ALBERT

Description: High-resolution (0.01/cm) absorption spectra of lean mixtures of CH4 in dry air were recorded with the McMath-Pierce Fourier transform spectrometer (FTS) of the National Solar Observatory on Kitt Peak at various temperatures between 24 and -61 C. The spectra have been analyzed to determine the values at room temperature of pressure-broadened widths and pressure-induced shifts of more than 740 transitions. The temperature dependence of air-broadened widths and pressure-induced shifts was deduced for approx. 370 transitions in the nu(sub 1) + nu(sub 4), nu(sub 3) + nu(sub 4), and nu(sub 2) + nu(sub 3) bands of (12)CH4 located between 4118 and 4615/cm. These results were obtained by analyzing a total of 29 spectra simultaneously using a multi-spectral non-linear least-squares fitting technique. This new technique allowed the determination of correlated spectral line parameters (e.g. intensity and broadening coefficient) better than the procedure of averaging values obtained by fitting the spectra individually. This method also provided a direct determination of the uncertainties in the retrieved parameters due to random errors. For each band analysed in this study the dependence of the various spectral line parameters upon the tetrahedral symmetry species and the rotational quantum numbers of the transitions is also presented.

Description: The integration of CLIPS into HyperCard combines the intuitive, interactive user interface of the Macintosh with the powerful symbolic computation of an expert system interpreter. HyperCard is an excellent environment for quickly developing the front end of an application with buttons, dialogs, and pictures, while the CLIPS interpreter provides a powerful inference engine for complex problem solving and analysis. In order to understand the benefit of integrating HyperCard and CLIPS, consider the following: HyperCard is an information storage and retrieval system which exploits the use of the graphics and user interface capabilities of the Apple Macintosh computer. The user can easily define buttons, dialog boxes, information templates, pictures, and graphic displays through the use of the HyperCard tools and scripting language. What is generally lacking in this environment is a powerful reasoning engine for complex problem solving, and this is where CLIPS plays a role. CLIPS 5.0 (C Language Integrated Production System, v5.0) was developed at the Johnson Space Center Software Technology Branch to allow artificial intelligence research, development, and delivery on conventional computers. CLIPS 5.0 supports forward chaining rule systems, object-oriented language, and procedural programming for the construction of expert systems. It features incremental reset, seven conflict resolution stategies, truth maintenance, and user-defined external functions. Since CLIPS is implemented in the C language it is highly portable; in addition, it is embeddable as a callable routine from a program written in another language such as Ada or Fortran. By integrating HyperCard and CLIPS the advantages and uses of both packages are made available for a wide range of applications: rapid prototyping of knowledge-based expert systems, interactive simulations of physical systems and intelligent control of hypertext processes, to name a few. HyperCLIPS 2.0 is written in C-Language (54%) and Pascal (46%) for Apple Macintosh computers running Macintosh System 6.0.2 or greater. HyperCLIPS requires HyperCard 1.2 or higher and at least 2Mb of RAM are recommended to run. An executable is provided. To compile the source code, the Macintosh Programmer's Workshop (MPW) version 3.0, CLIPS 5.0 (MSC-21927), and the MPW C-Language compiler are also required. NOTE: Installing this program under Macintosh System 7 requires HyperCard v2.1. This program is distributed on a 3.5 inch Macintosh format diskette. A copy of the program documentation is included on the diskette, but may be purchased separately. HyperCLIPS was developed in 1990 and version 2.0 was released in 1991. HyperCLIPS is a copyrighted work with all copyright vested in NASA. Apple, Macintosh, MPW, and HyperCard are registered trademarks of Apple Computer, Inc.

Description: The C Language Integrated Production System, CLIPS, is a shell for developing expert systems. It is designed to allow artificial intelligence research, development, and delivery on conventional computers. The primary design goals for CLIPS are portability, efficiency, and functionality. For these reasons, the program is written in C. CLIPS meets or outperforms most micro- and minicomputer based artificial intelligence tools. CLIPS is a forward chaining rule-based language. The program contains an inference engine and a language syntax that provide a framework for the construction of an expert system. It also includes tools for debugging an application. CLIPS is based on the Rete algorithm, which enables very efficient pattern matching. The collection of conditions and actions to be taken if the conditions are met is constructed into a rule network. As facts are asserted either prior to or during a session, CLIPS pattern-matches the number of fields. Wildcards and variables are supported for both single and multiple fields. CLIPS syntax allows the inclusion of externally defined functions (outside functions which are written in a language other than CLIPS). CLIPS itself can be embedded in a program such that the expert system is available as a simple subroutine call. Advanced features found in CLIPS version 4.3 include an integrated microEMACS editor, the ability to generate C source code from a CLIPS rule base to produce a dedicated executable, binary load and save capabilities for CLIPS rule bases, and the utility program CRSV (Cross-Reference, Style, and Verification) designed to facilitate the development and maintenance of large rule bases. Five machine versions are available. Each machine version includes the source and the executable for that machine. The UNIX version includes the source and binaries for IBM RS/6000, Sun3 series, and Sun4 series computers. The UNIX, DEC VAX, and DEC RISC Workstation versions are line oriented. The PC version and the Macintosh version each contain a windowing variant of CLIPS as well as the standard line oriented version. The mouse/window interface version for the PC works with a Microsoft compatible mouse or without a mouse. This window version uses the proprietary CURSES library for the PC, but a working executable of the window version is provided. The window oriented version for the Macintosh includes a version which uses a full Macintosh-style interface, including an integrated editor. This version allows the user to observe the changing fact base and rule activations in separate windows while a CLIPS program is executing. The IBM PC version is available bundled with CLIPSITS, The CLIPS Intelligent Tutoring System for a special combined price (COS-10025). The goal of CLIPSITS is to provide the student with a tool to practice the syntax and concepts covered in the CLIPS User's Guide. It attempts to provide expert diagnosis and advice during problem solving which is typically not available without an instructor. CLIPSITS is divided into 10 lessons which mirror the first 10 chapters of the CLIPS User's Guide. The program was developed for the IBM PC series with a hard disk. CLIPSITS is also available separately as MSC-21679. The CLIPS program is written in C for interactive execution and has been implemented on an IBM PC computer operating under DOS, a Macintosh and DEC VAX series computers operating under VMS or ULTRIX. The line oriented version should run on any computer system which supports a full (Kernighan and Ritchie) C compiler or the ANSI standard C language. CLIPS was developed in 1986 and Version 4.2 was released in July of 1988. Version 4.3 was released in June of 1989.

Description: There is no simple and perfect way to measure residual stresses in metal parts that have been welded or deformed to make complex structures such as pressure vessels and aircraft, yet these locked-in stresses can contribute to structural failure by fatigue and fracture. However, one proven and tested technique for determining the internal stress of a metal part is to drill a test hole while measuring the relieved strains around the hole, such as the hole-drilling strain gage method described in ASTM E 837. The program HOLEGAGE processes strain gage data and provides additional calculations of internal stress variations that are not obtained with standard E 837 analysis methods. The typical application of the technique uses a three gage rosette with a special hole-drilling fixture for drilling a hole through the center of the rosette to produce a hole with very small gage pattern eccentricity error. Another device is used to control the drilling and halt the drill at controlled depth steps. At each step, strains from all three strain gages are recorded. The influence coefficients used by HOLEGAGE to compute stresses from relieved hole strains were developed by published finite element method studies of thick plates for specific hole sizes and depths. The program uses a parabolic fit and an interpolating scheme to project the coefficients to other hole sizes and depths. Additionally, published experimental data are used to extend the coefficients to relatively thin plates. These influence coefficients are used to compute the stresses in the original part from the strain data. HOLEGAGE will compute interior planar stresses using strain data from each drilled hole depth layer. Planar stresses may be computed in three ways including: a least squares fit for a linear variation with depth, an integral method to give incremental stress data for each layer, or by a linear fit to the integral data (with some surface data points omitted) to predict surface stresses before strain gage sanding preparations introduced additional residual stresses. Options are included for estimating the effect of hole eccentricity on calculations, smoothing noise from the strain data, and inputting the program data either interactively or from a data file. HOLEGAGE was written in FORTRAN 77 for DEC VAX computers under VMS, and is transportable except for system-unique TIME and DATE system calls. The program requires 54K of main memory and was developed in 1990. The program is available on a 9-track 1600 BPI VAX BACKUP format magnetic tape (standard media) or a TK50 tape cartridge. The documentation is included on the tape. DEC VAX and VMS are trademarks of Digital Equipment Corporation.

Description: VASP is a variable dimension Fortran version of the Automatic Synthesis Program, ASP. The program is used to implement Kalman filtering and control theory. Basically, it consists of 31 subprograms for solving most modern control problems in linear, time-variant (or time-invariant) control systems. These subprograms include operations of matrix algebra, computation of the exponential of a matrix and its convolution integral, and the solution of the matrix Riccati equation. The user calls these subprograms by means of a FORTRAN main program, and so can easily obtain solutions to most general problems of extremization of a quadratic functional of the state of the linear dynamical system. Particularly, these problems include the synthesis of the Kalman filter gains and the optimal feedback gains for minimization of a quadratic performance index. VASP, as an outgrowth of the Automatic Synthesis Program, has the following improvements: more versatile programming language; more convenient input/output format; some new subprograms which consolidate certain groups of statements that are often repeated; and variable dimensioning. The pertinent difference between the two programs is that VASP has variable dimensioning and more efficient storage. The documentation for the VASP program contains a VASP dictionary and example problems. The dictionary contains a description of each subroutine and instructions on its use. The example problems include dynamic response, optimal control gain, solution of the sampled data matrix Riccati equation, matrix decomposition, and a pseudo-inverse of a matrix. This program is written in FORTRAN IV and has been implemented on the IBM 360. The VASP program was developed in 1971.

Description: The Simple Tool for Automated Reasoning program (STAR) is an interactive, interpreted programming language for the development and operation of artificial intelligence (AI) application systems. STAR provides an environment for integrating traditional AI symbolic processing with functions and data structures defined in compiled languages such as C, FORTRAN and PASCAL. This type of integration occurs in a number of AI applications including interpretation of numerical sensor data, construction of intelligent user interfaces to existing compiled software packages, and coupling AI techniques with numerical simulation techniques and control systems software. The STAR language was created as part of an AI project for the evaluation of imaging spectrometer data at NASA's Jet Propulsion Laboratory. Programming in STAR is similar to other symbolic processing languages such as LISP and CLIP. STAR includes seven primitive data types and associated operations for the manipulation of these structures. A semantic network is used to organize data in STAR, with capabilities for inheritance of values and generation of side effects. The AI knowledge base of STAR can be a simple repository of records or it can be a highly interdependent association of implicit and explicit components. The symbolic processing environment of STAR may be extended by linking the interpreter with functions defined in conventional compiled languages. These external routines interact with STAR through function calls in either direction, and through the exchange of references to data structures. The hybrid knowledge base may thus be accessed and processed in general by either side of the application. STAR is initially used to link externally compiled routines and data structures. It is then invoked to interpret the STAR rules and symbolic structures. In a typical interactive session, the user enters an expression to be evaluated, STAR parses the input, evaluates the expression, performs any file input/output required, and displays the results. The STAR interpreter is written in the C language for interactive execution. It has been implemented on a VAX 11/780 computer operating under VMS, and the UNIX version has been implemented on a Sun Microsystems 2/170 workstation. STAR has a memory requirement of approximately 200K of 8 bit bytes, excluding externally compiled functions and application-dependent symbolic definitions. This program was developed in 1985.

Description: The Simple Tool for Automated Reasoning program (STAR) is an interactive, interpreted programming language for the development and operation of artificial intelligence (AI) application systems. STAR provides an environment for integrating traditional AI symbolic processing with functions and data structures defined in compiled languages such as C, FORTRAN and PASCAL. This type of integration occurs in a number of AI applications including interpretation of numerical sensor data, construction of intelligent user interfaces to existing compiled software packages, and coupling AI techniques with numerical simulation techniques and control systems software. The STAR language was created as part of an AI project for the evaluation of imaging spectrometer data at NASA's Jet Propulsion Laboratory. Programming in STAR is similar to other symbolic processing languages such as LISP and CLIP. STAR includes seven primitive data types and associated operations for the manipulation of these structures. A semantic network is used to organize data in STAR, with capabilities for inheritance of values and generation of side effects. The AI knowledge base of STAR can be a simple repository of records or it can be a highly interdependent association of implicit and explicit components. The symbolic processing environment of STAR may be extended by linking the interpreter with functions defined in conventional compiled languages. These external routines interact with STAR through function calls in either direction, and through the exchange of references to data structures. The hybrid knowledge base may thus be accessed and processed in general by either side of the application. STAR is initially used to link externally compiled routines and data structures. It is then invoked to interpret the STAR rules and symbolic structures. In a typical interactive session, the user enters an expression to be evaluated, STAR parses the input, evaluates the expression, performs any file input/output required, and displays the results. The STAR interpreter is written in the C language for interactive execution. It has been implemented on a VAX 11/780 computer operating under VMS, and the UNIX version has been implemented on a Sun Microsystems 2/170 workstation. STAR has a memory requirement of approximately 200K of 8 bit bytes, excluding externally compiled functions and application-dependent symbolic definitions. This program was developed in 1985.

Description: CLIPS, the C Language Integrated Production System, is a complete environment for developing expert systems -- programs which are specifically intended to model human expertise or knowledge. It is designed to allow artificial intelligence research, development, and delivery on conventional computers. CLIPS 6.0 provides a cohesive tool for handling a wide variety of knowledge with support for three different programming paradigms: rule-based, object-oriented, and procedural. Rule-based programming allows knowledge to be represented as heuristics, or "rules-of-thumb" which specify a set of actions to be performed for a given situation. Object-oriented programming allows complex systems to be modeled as modular components (which can be easily reused to model other systems or create new components). The procedural programming capabilities provided by CLIPS 6.0 allow CLIPS to represent knowledge in ways similar to those allowed in languages such as C, Pascal, Ada, and LISP. Using CLIPS 6.0, one can develop expert system software using only rule-based programming, only object-oriented programming, only procedural programming, or combinations of the three. CLIPS provides extensive features to support the rule-based programming paradigm including seven conflict resolution strategies, dynamic rule priorities, and truth maintenance. CLIPS 6.0 supports more complex nesting of conditional elements in the if portion of a rule ("and", "or", and "not" conditional elements can be placed within a "not" conditional element). In addition, there is no longer a limitation on the number of multifield slots that a deftemplate can contain. The CLIPS Object-Oriented Language (COOL) provides object-oriented programming capabilities. Features supported by COOL include classes with multiple inheritance, abstraction, encapsulation, polymorphism, dynamic binding, and message passing with message-handlers. CLIPS 6.0 supports tight integration of the rule-based programming features of CLIPS with COOL (that is, a rule can pattern match on objects created using COOL). CLIPS 6.0 provides the capability to define functions, overloaded functions, and global variables interactively. In addition, CLIPS can be embedded within procedural code, called as a subroutine, and integrated with languages such as C, FORTRAN and Ada. CLIPS can be easily extended by a user through the use of several well-defined protocols. CLIPS provides several delivery options for programs including the ability to generate stand alone executables or to load programs from text or binary files. CLIPS 6.0 provides support for the modular development and execution of knowledge bases with the defmodule construct. CLIPS modules allow a set of constructs to be grouped together such that explicit control can be maintained over restricting the access of the constructs by other modules. This type of control is similar to global and local scoping used in languages such as C or Ada. By restricting access to deftemplate and defclass constructs, modules can function as blackboards, permitting only certain facts and instances to be seen by other modules. Modules are also used by rules to provide execution control. The CRSV (Cross-Reference, Style, and Verification) utility included with previous version of CLIPS is no longer supported. The capabilities provided by this tool are now available directly within CLIPS 6.0 to aid in the development, debugging, and verification of large rule bases. COSMIC offers four distribution versions of CLIPS 6.0: UNIX (MSC-22433), VMS (MSC-22434), MACINTOSH (MSC-22429), and IBM PC (MSC-22430). Executable files, source code, utilities, documentation, and examples are included on the program media. All distribution versions include identical source code for the command line version of CLIPS 6.0. This source code should compile on any platform with an ANSI C compiler. Each distribution version of CLIPS 6.0, except that for the Macintosh platform, includes an executable for the command line version. For the UNIX version of CLIPS 6.0, the command line interface has been successfully implemented on a Sun4 running SunOS, a DECstation running DEC RISC ULTRIX, an SGI Indigo Elan running IRIX, a DEC Alpha AXP running OSF/1, and an IBM RS/6000 running AIX. Command line interface executables are included for Sun4 computers running SunOS 4.1.1 or later and for the DEC RISC ULTRIX platform. The makefiles may have to be modified slightly to be used on other UNIX platforms. The UNIX, Macintosh, and IBM PC versions of CLIPS 6.0 each have a platform specific interface. Source code, a makefile, and an executable for the Windows 3.1 interface version of CLIPS 6.0 are provided only on the IBM PC distribution diskettes. Source code, a makefile, and an executable for the Macintosh interface version of CLIPS 6.0 are provided only on the Macintosh distribution diskettes. Likewise, for the UNIX version of CLIPS 6.0, only source code and a makefile for an X-Windows interface are provided. The X-Windows interface requires MIT's X Window System, Version 11, Release 4 (X11R4), the Athena Widget Set, and the Xmu library. The source code for the Athena Widget Set is provided on the distribution medium. The X-Windows interface has been successfully implemented on a Sun4 running SunOS 4.1.2 with the MIT distribution of X11R4 (not OpenWindows), an SGI Indigo Elan running IRIX 4.0.5, and a DEC Alpha AXP running OSF/1 1.2. The VAX version of CLIPS 6.0 comes only with the generic command line interface. ASCII makefiles for the command line version of CLIPS are provided on all the distribution media for UNIX, VMS, and DOS. Four executables are provided with the IBM PC version: a windowed interface executable for Windows 3.1 built using Borland C++ v3.1, an editor for use with the windowed interface, a command line version of CLIPS for Windows 3.1, and a 386 command line executable for DOS built using Zortech C++ v3.1. All four executables are capable of utilizing extended memory and require an 80386 CPU or better. Users needing an 8086/8088 or 80286 executable must recompile the CLIPS source code themselves. Users who wish to recompile the DOS executable using Borland C++ or MicroSoft C must use a DOS extender program to produce an executable capable of using extended memory. The version of CLIPS 6.0 for IBM PC compatibles requires DOS v3.3 or later and/or Windows 3.1 or later. It is distributed on a set of three 1.4Mb 3.5 inch diskettes. A hard disk is required. The Macintosh version is distributed in compressed form on two 3.5 inch 1.4Mb Macintosh format diskettes, and requires System 6.0.5, or higher, and 1Mb RAM. The version for DEC VAX/VMS is available in VAX BACKUP format on a 1600 BPI 9-track magnetic tape (standard distribution medium) or a TK50 tape cartridge. The UNIX version is distributed in UNIX tar format on a .25 inch streaming magnetic tape cartridge (Sun QIC-24). For the UNIX version, alternate distribution media and formats are available upon request. The CLIPS 6.0 documentation includes a User's Guide and a three volume Reference Manual consisting of Basic and Advanced Programming Guides and an Interfaces Guide. An electronic version of the documentation is provided on the distribution medium for each version: in MicroSoft Word format for the Macintosh and PC versions of CLIPS, and in both PostScript format and MicroSoft Word for Macintosh format for the UNIX and DEC VAX versions of CLIPS. CLIPS was developed in 1986 and Version 6.0 was released in 1993.

Description: CLIPS, the C Language Integrated Production System, is a complete environment for developing expert systems -- programs which are specifically intended to model human expertise or knowledge. It is designed to allow artificial intelligence research, development, and delivery on conventional computers. CLIPS 6.0 provides a cohesive tool for handling a wide variety of knowledge with support for three different programming paradigms: rule-based, object-oriented, and procedural. Rule-based programming allows knowledge to be represented as heuristics, or "rules-of-thumb" which specify a set of actions to be performed for a given situation. Object-oriented programming allows complex systems to be modeled as modular components (which can be easily reused to model other systems or create new components). The procedural programming capabilities provided by CLIPS 6.0 allow CLIPS to represent knowledge in ways similar to those allowed in languages such as C, Pascal, Ada, and LISP. Using CLIPS 6.0, one can develop expert system software using only rule-based programming, only object-oriented programming, only procedural programming, or combinations of the three. CLIPS provides extensive features to support the rule-based programming paradigm including seven conflict resolution strategies, dynamic rule priorities, and truth maintenance. CLIPS 6.0 supports more complex nesting of conditional elements in the if portion of a rule ("and", "or", and "not" conditional elements can be placed within a "not" conditional element). In addition, there is no longer a limitation on the number of multifield slots that a deftemplate can contain. The CLIPS Object-Oriented Language (COOL) provides object-oriented programming capabilities. Features supported by COOL include classes with multiple inheritance, abstraction, encapsulation, polymorphism, dynamic binding, and message passing with message-handlers. CLIPS 6.0 supports tight integration of the rule-based programming features of CLIPS with COOL (that is, a rule can pattern match on objects created using COOL). CLIPS 6.0 provides the capability to define functions, overloaded functions, and global variables interactively. In addition, CLIPS can be embedded within procedural code, called as a subroutine, and integrated with languages such as C, FORTRAN and Ada. CLIPS can be easily extended by a user through the use of several well-defined protocols. CLIPS provides several delivery options for programs including the ability to generate stand alone executables or to load programs from text or binary files. CLIPS 6.0 provides support for the modular development and execution of knowledge bases with the defmodule construct. CLIPS modules allow a set of constructs to be grouped together such that explicit control can be maintained over restricting the access of the constructs by other modules. This type of control is similar to global and local scoping used in languages such as C or Ada. By restricting access to deftemplate and defclass constructs, modules can function as blackboards, permitting only certain facts and instances to be seen by other modules. Modules are also used by rules to provide execution control. The CRSV (Cross-Reference, Style, and Verification) utility included with previous version of CLIPS is no longer supported. The capabilities provided by this tool are now available directly within CLIPS 6.0 to aid in the development, debugging, and verification of large rule bases. COSMIC offers four distribution versions of CLIPS 6.0: UNIX (MSC-22433), VMS (MSC-22434), MACINTOSH (MSC-22429), and IBM PC (MSC-22430). Executable files, source code, utilities, documentation, and examples are included on the program media. All distribution versions include identical source code for the command line version of CLIPS 6.0. This source code should compile on any platform with an ANSI C compiler. Each distribution version of CLIPS 6.0, except that for the Macintosh platform, includes an executable for the command line version. For the UNIX version of CLIPS 6.0, the command line interface has been successfully implemented on a Sun4 running SunOS, a DECstation running DEC RISC ULTRIX, an SGI Indigo Elan running IRIX, a DEC Alpha AXP running OSF/1, and an IBM RS/6000 running AIX. Command line interface executables are included for Sun4 computers running SunOS 4.1.1 or later and for the DEC RISC ULTRIX platform. The makefiles may have to be modified slightly to be used on other UNIX platforms. The UNIX, Macintosh, and IBM PC versions of CLIPS 6.0 each have a platform specific interface. Source code, a makefile, and an executable for the Windows 3.1 interface version of CLIPS 6.0 are provided only on the IBM PC distribution diskettes. Source code, a makefile, and an executable for the Macintosh interface version of CLIPS 6.0 are provided only on the Macintosh distribution diskettes. Likewise, for the UNIX version of CLIPS 6.0, only source code and a makefile for an X-Windows interface are provided. The X-Windows interface requires MIT's X Window System, Version 11, Release 4 (X11R4), the Athena Widget Set, and the Xmu library. The source code for the Athena Widget Set is provided on the distribution medium. The X-Windows interface has been successfully implemented on a Sun4 running SunOS 4.1.2 with the MIT distribution of X11R4 (not OpenWindows), an SGI Indigo Elan running IRIX 4.0.5, and a DEC Alpha AXP running OSF/1 1.2. The VAX version of CLIPS 6.0 comes only with the generic command line interface. ASCII makefiles for the command line version of CLIPS are provided on all the distribution media for UNIX, VMS, and DOS. Four executables are provided with the IBM PC version: a windowed interface executable for Windows 3.1 built using Borland C++ v3.1, an editor for use with the windowed interface, a command line version of CLIPS for Windows 3.1, and a 386 command line executable for DOS built using Zortech C++ v3.1. All four executables are capable of utilizing extended memory and require an 80386 CPU or better. Users needing an 8086/8088 or 80286 executable must recompile the CLIPS source code themselves. Users who wish to recompile the DOS executable using Borland C++ or MicroSoft C must use a DOS extender program to produce an executable capable of using extended memory. The version of CLIPS 6.0 for IBM PC compatibles requires DOS v3.3 or later and/or Windows 3.1 or later. It is distributed on a set of three 1.4Mb 3.5 inch diskettes. A hard disk is required. The Macintosh version is distributed in compressed form on two 3.5 inch 1.4Mb Macintosh format diskettes, and requires System 6.0.5, or higher, and 1Mb RAM. The version for DEC VAX/VMS is available in VAX BACKUP format on a 1600 BPI 9-track magnetic tape (standard distribution medium) or a TK50 tape cartridge. The UNIX version is distributed in UNIX tar format on a .25 inch streaming magnetic tape cartridge (Sun QIC-24). For the UNIX version, alternate distribution media and formats are available upon request. The CLIPS 6.0 documentation includes a User's Guide and a three volume Reference Manual consisting of Basic and Advanced Programming Guides and an Interfaces Guide. An electronic version of the documentation is provided on the distribution medium for each version: in MicroSoft Word format for the Macintosh and PC versions of CLIPS, and in both PostScript format and MicroSoft Word for Macintosh format for the UNIX and DEC VAX versions of CLIPS. CLIPS was developed in 1986 and Version 6.0 was released in 1993.

Description: CLIPS, the C Language Integrated Production System, is a complete environment for developing expert systems -- programs which are specifically intended to model human expertise or knowledge. It is designed to allow artificial intelligence research, development, and delivery on conventional computers. CLIPS 6.0 provides a cohesive tool for handling a wide variety of knowledge with support for three different programming paradigms: rule-based, object-oriented, and procedural. Rule-based programming allows knowledge to be represented as heuristics, or "rules-of-thumb" which specify a set of actions to be performed for a given situation. Object-oriented programming allows complex systems to be modeled as modular components (which can be easily reused to model other systems or create new components). The procedural programming capabilities provided by CLIPS 6.0 allow CLIPS to represent knowledge in ways similar to those allowed in languages such as C, Pascal, Ada, and LISP. Using CLIPS 6.0, one can develop expert system software using only rule-based programming, only object-oriented programming, only procedural programming, or combinations of the three. CLIPS provides extensive features to support the rule-based programming paradigm including seven conflict resolution strategies, dynamic rule priorities, and truth maintenance. CLIPS 6.0 supports more complex nesting of conditional elements in the if portion of a rule ("and", "or", and "not" conditional elements can be placed within a "not" conditional element). In addition, there is no longer a limitation on the number of multifield slots that a deftemplate can contain. The CLIPS Object-Oriented Language (COOL) provides object-oriented programming capabilities. Features supported by COOL include classes with multiple inheritance, abstraction, encapsulation, polymorphism, dynamic binding, and message passing with message-handlers. CLIPS 6.0 supports tight integration of the rule-based programming features of CLIPS with COOL (that is, a rule can pattern match on objects created using COOL). CLIPS 6.0 provides the capability to define functions, overloaded functions, and global variables interactively. In addition, CLIPS can be embedded within procedural code, called as a subroutine, and integrated with languages such as C, FORTRAN and Ada. CLIPS can be easily extended by a user through the use of several well-defined protocols. CLIPS provides several delivery options for programs including the ability to generate stand alone executables or to load programs from text or binary files. CLIPS 6.0 provides support for the modular development and execution of knowledge bases with the defmodule construct. CLIPS modules allow a set of constructs to be grouped together such that explicit control can be maintained over restricting the access of the constructs by other modules. This type of control is similar to global and local scoping used in languages such as C or Ada. By restricting access to deftemplate and defclass constructs, modules can function as blackboards, permitting only certain facts and instances to be seen by other modules. Modules are also used by rules to provide execution control. The CRSV (Cross-Reference, Style, and Verification) utility included with previous version of CLIPS is no longer supported. The capabilities provided by this tool are now available directly within CLIPS 6.0 to aid in the development, debugging, and verification of large rule bases. COSMIC offers four distribution versions of CLIPS 6.0: UNIX (MSC-22433), VMS (MSC-22434), MACINTOSH (MSC-22429), and IBM PC (MSC-22430). Executable files, source code, utilities, documentation, and examples are included on the program media. All distribution versions include identical source code for the command line version of CLIPS 6.0. This source code should compile on any platform with an ANSI C compiler. Each distribution version of CLIPS 6.0, except that for the Macintosh platform, includes an executable for the command line version. For the UNIX version of CLIPS 6.0, the command line interface has been successfully implemented on a Sun4 running SunOS, a DECstation running DEC RISC ULTRIX, an SGI Indigo Elan running IRIX, a DEC Alpha AXP running OSF/1, and an IBM RS/6000 running AIX. Command line interface executables are included for Sun4 computers running SunOS 4.1.1 or later and for the DEC RISC ULTRIX platform. The makefiles may have to be modified slightly to be used on other UNIX platforms. The UNIX, Macintosh, and IBM PC versions of CLIPS 6.0 each have a platform specific interface. Source code, a makefile, and an executable for the Windows 3.1 interface version of CLIPS 6.0 are provided only on the IBM PC distribution diskettes. Source code, a makefile, and an executable for the Macintosh interface version of CLIPS 6.0 are provided only on the Macintosh distribution diskettes. Likewise, for the UNIX version of CLIPS 6.0, only source code and a makefile for an X-Windows interface are provided. The X-Windows interface requires MIT's X Window System, Version 11, Release 4 (X11R4), the Athena Widget Set, and the Xmu library. The source code for the Athena Widget Set is provided on the distribution medium. The X-Windows interface has been successfully implemented on a Sun4 running SunOS 4.1.2 with the MIT distribution of X11R4 (not OpenWindows), an SGI Indigo Elan running IRIX 4.0.5, and a DEC Alpha AXP running OSF/1 1.2. The VAX version of CLIPS 6.0 comes only with the generic command line interface. ASCII makefiles for the command line version of CLIPS are provided on all the distribution media for UNIX, VMS, and DOS. Four executables are provided with the IBM PC version: a windowed interface executable for Windows 3.1 built using Borland C++ v3.1, an editor for use with the windowed interface, a command line version of CLIPS for Windows 3.1, and a 386 command line executable for DOS built using Zortech C++ v3.1. All four executables are capable of utilizing extended memory and require an 80386 CPU or better. Users needing an 8086/8088 or 80286 executable must recompile the CLIPS source code themselves. Users who wish to recompile the DOS executable using Borland C++ or MicroSoft C must use a DOS extender program to produce an executable capable of using extended memory. The version of CLIPS 6.0 for IBM PC compatibles requires DOS v3.3 or later and/or Windows 3.1 or later. It is distributed on a set of three 1.4Mb 3.5 inch diskettes. A hard disk is required. The Macintosh version is distributed in compressed form on two 3.5 inch 1.4Mb Macintosh format diskettes, and requires System 6.0.5, or higher, and 1Mb RAM. The version for DEC VAX/VMS is available in VAX BACKUP format on a 1600 BPI 9-track magnetic tape (standard distribution medium) or a TK50 tape cartridge. The UNIX version is distributed in UNIX tar format on a .25 inch streaming magnetic tape cartridge (Sun QIC-24). For the UNIX version, alternate distribution media and formats are available upon request. The CLIPS 6.0 documentation includes a User's Guide and a three volume Reference Manual consisting of Basic and Advanced Programming Guides and an Interfaces Guide. An electronic version of the documentation is provided on the distribution medium for each version: in MicroSoft Word format for the Macintosh and PC versions of CLIPS, and in both PostScript format and MicroSoft Word for Macintosh format for the UNIX and DEC VAX versions of CLIPS. CLIPS was developed in 1986 and Version 6.0 was released in 1993.

Description: CLIPS, the C Language Integrated Production System, is a complete environment for developing expert systems -- programs which are specifically intended to model human expertise or knowledge. It is designed to allow artificial intelligence research, development, and delivery on conventional computers. CLIPS 6.0 provides a cohesive tool for handling a wide variety of knowledge with support for three different programming paradigms: rule-based, object-oriented, and procedural. Rule-based programming allows knowledge to be represented as heuristics, or "rules-of-thumb" which specify a set of actions to be performed for a given situation. Object-oriented programming allows complex systems to be modeled as modular components (which can be easily reused to model other systems or create new components). The procedural programming capabilities provided by CLIPS 6.0 allow CLIPS to represent knowledge in ways similar to those allowed in languages such as C, Pascal, Ada, and LISP. Using CLIPS 6.0, one can develop expert system software using only rule-based programming, only object-oriented programming, only procedural programming, or combinations of the three. CLIPS provides extensive features to support the rule-based programming paradigm including seven conflict resolution strategies, dynamic rule priorities, and truth maintenance. CLIPS 6.0 supports more complex nesting of conditional elements in the if portion of a rule ("and", "or", and "not" conditional elements can be placed within a "not" conditional element). In addition, there is no longer a limitation on the number of multifield slots that a deftemplate can contain. The CLIPS Object-Oriented Language (COOL) provides object-oriented programming capabilities. Features supported by COOL include classes with multiple inheritance, abstraction, encapsulation, polymorphism, dynamic binding, and message passing with message-handlers. CLIPS 6.0 supports tight integration of the rule-based programming features of CLIPS with COOL (that is, a rule can pattern match on objects created using COOL). CLIPS 6.0 provides the capability to define functions, overloaded functions, and global variables interactively. In addition, CLIPS can be embedded within procedural code, called as a subroutine, and integrated with languages such as C, FORTRAN and Ada. CLIPS can be easily extended by a user through the use of several well-defined protocols. CLIPS provides several delivery options for programs including the ability to generate stand alone executables or to load programs from text or binary files. CLIPS 6.0 provides support for the modular development and execution of knowledge bases with the defmodule construct. CLIPS modules allow a set of constructs to be grouped together such that explicit control can be maintained over restricting the access of the constructs by other modules. This type of control is similar to global and local scoping used in languages such as C or Ada. By restricting access to deftemplate and defclass constructs, modules can function as blackboards, permitting only certain facts and instances to be seen by other modules. Modules are also used by rules to provide execution control. The CRSV (Cross-Reference, Style, and Verification) utility included with previous version of CLIPS is no longer supported. The capabilities provided by this tool are now available directly within CLIPS 6.0 to aid in the development, debugging, and verification of large rule bases. COSMIC offers four distribution versions of CLIPS 6.0: UNIX (MSC-22433), VMS (MSC-22434), MACINTOSH (MSC-22429), and IBM PC (MSC-22430). Executable files, source code, utilities, documentation, and examples are included on the program media. All distribution versions include identical source code for the command line version of CLIPS 6.0. This source code should compile on any platform with an ANSI C compiler. Each distribution version of CLIPS 6.0, except that for the Macintosh platform, includes an executable for the command line version. For the UNIX version of CLIPS 6.0, the command line interface has been successfully implemented on a Sun4 running SunOS, a DECstation running DEC RISC ULTRIX, an SGI Indigo Elan running IRIX, a DEC Alpha AXP running OSF/1, and an IBM RS/6000 running AIX. Command line interface executables are included for Sun4 computers running SunOS 4.1.1 or later and for the DEC RISC ULTRIX platform. The makefiles may have to be modified slightly to be used on other UNIX platforms. The UNIX, Macintosh, and IBM PC versions of CLIPS 6.0 each have a platform specific interface. Source code, a makefile, and an executable for the Windows 3.1 interface version of CLIPS 6.0 are provided only on the IBM PC distribution diskettes. Source code, a makefile, and an executable for the Macintosh interface version of CLIPS 6.0 are provided only on the Macintosh distribution diskettes. Likewise, for the UNIX version of CLIPS 6.0, only source code and a makefile for an X-Windows interface are provided. The X-Windows interface requires MIT's X Window System, Version 11, Release 4 (X11R4), the Athena Widget Set, and the Xmu library. The source code for the Athena Widget Set is provided on the distribution medium. The X-Windows interface has been successfully implemented on a Sun4 running SunOS 4.1.2 with the MIT distribution of X11R4 (not OpenWindows), an SGI Indigo Elan running IRIX 4.0.5, and a DEC Alpha AXP running OSF/1 1.2. The VAX version of CLIPS 6.0 comes only with the generic command line interface. ASCII makefiles for the command line version of CLIPS are provided on all the distribution media for UNIX, VMS, and DOS. Four executables are provided with the IBM PC version: a windowed interface executable for Windows 3.1 built using Borland C++ v3.1, an editor for use with the windowed interface, a command line version of CLIPS for Windows 3.1, and a 386 command line executable for DOS built using Zortech C++ v3.1. All four executables are capable of utilizing extended memory and require an 80386 CPU or better. Users needing an 8086/8088 or 80286 executable must recompile the CLIPS source code themselves. Users who wish to recompile the DOS executable using Borland C++ or MicroSoft C must use a DOS extender program to produce an executable capable of using extended memory. The version of CLIPS 6.0 for IBM PC compatibles requires DOS v3.3 or later and/or Windows 3.1 or later. It is distributed on a set of three 1.4Mb 3.5 inch diskettes. A hard disk is required. The Macintosh version is distributed in compressed form on two 3.5 inch 1.4Mb Macintosh format diskettes, and requires System 6.0.5, or higher, and 1Mb RAM. The version for DEC VAX/VMS is available in VAX BACKUP format on a 1600 BPI 9-track magnetic tape (standard distribution medium) or a TK50 tape cartridge. The UNIX version is distributed in UNIX tar format on a .25 inch streaming magnetic tape cartridge (Sun QIC-24). For the UNIX version, alternate distribution media and formats are available upon request. The CLIPS 6.0 documentation includes a User's Guide and a three volume Reference Manual consisting of Basic and Advanced Programming Guides and an Interfaces Guide. An electronic version of the documentation is provided on the distribution medium for each version: in MicroSoft Word format for the Macintosh and PC versions of CLIPS, and in both PostScript format and MicroSoft Word for Macintosh format for the UNIX and DEC VAX versions of CLIPS. CLIPS was developed in 1986 and Version 6.0 was released in 1993.

Description: NETS, A Tool for the Development and Evaluation of Neural Networks, provides a simulation of Neural Network algorithms plus an environment for developing such algorithms. Neural Networks are a class of systems modeled after the human brain. Artificial Neural Networks are formed from hundreds or thousands of simulated neurons, connected to each other in a manner similar to brain neurons. Problems which involve pattern matching readily fit the class of problems which NETS is designed to solve. NETS uses the back propagation learning method for all of the networks which it creates. The nodes of a network are usually grouped together into clumps called layers. Generally, a network will have an input layer through which the various environment stimuli are presented to the network, and an output layer for determining the network's response. The number of nodes in these two layers is usually tied to some features of the problem being solved. Other layers, which form intermediate stops between the input and output layers, are called hidden layers. NETS allows the user to customize the patterns of connections between layers of a network. NETS also provides features for saving the weight values of a network during the learning process, which allows for more precise control over the learning process. NETS is an interpreter. Its method of execution is the familiar "read-evaluate-print" loop found in interpreted languages such as BASIC and LISP. The user is presented with a prompt which is the simulator's way of asking for input. After a command is issued, NETS will attempt to evaluate the command, which may produce more prompts requesting specific information or an error if the command is not understood. The typical process involved when using NETS consists of translating the problem into a format which uses input/output pairs, designing a network configuration for the problem, and finally training the network with input/output pairs until an acceptable error is reached. NETS allows the user to generate C code to implement the network loaded into the system. This permits the placement of networks as components, or subroutines, in other systems. In short, once a network performs satisfactorily, the Generate C Code option provides the means for creating a program separate from NETS to run the network. Other features: files may be stored in binary or ASCII format; multiple input propagation is permitted; bias values may be included; capability to scale data without writing scaling code; quick interactive testing of network from the main menu; and several options that allow the user to manipulate learning efficiency. NETS is written in ANSI standard C language to be machine independent. The Macintosh version (MSC-22108) includes code for both a graphical user interface version and a command line interface version. The machine independent version (MSC-21588) only includes code for the command line interface version of NETS 3.0. The Macintosh version requires a Macintosh II series computer and has been successfully implemented under System 7. Four executables are included on these diskettes, two for floating point operations and two for integer arithmetic. It requires Think C 5.0 to compile. A minimum of 1Mb of RAM is required for execution. Sample input files and executables for both the command line version and the Macintosh user interface version are provided on the distribution medium. The Macintosh version is available on a set of three 3.5 inch 800K Macintosh format diskettes. The machine independent version has been successfully implemented on an IBM PC series compatible running MS-DOS, a DEC VAX running VMS, a SunIPC running SunOS, and a CRAY Y-MP running UNICOS. Two executables for the IBM PC version are included on the MS-DOS distribution media, one compiled for floating point operations and one for integer arithmetic. The machine independent version is available on a set of three 5.25 inch 360K MS-DOS format diskettes (standard distribution medium) or a .25 inch streaming magnetic tape cartridge in UNIX tar format. NETS was developed in 1989 and updated in 1992. IBM PC is a registered trademark of International Business Machines. MS-DOS is a registered trademark of Microsoft Corporation. DEC, VAX, and VMS are trademarks of Digital Equipment Corporation. SunIPC and SunOS are trademarks of Sun Microsystems, Inc. CRAY Y-MP and UNICOS are trademarks of Cray Research, Inc.

Description: The CLIPS Intelligent Tutoring System (CLIPSITS) is designed to be used to learn CLIPS, the C-language Integrated Production System expert system shell developed by the Software Technology Branch at Johnson Space Center. The goal of CLIPSITS is to provide the student with a tool to practice the syntax and concepts covered in the CLIPS User's Guide. It attempts to provide expert diagnosis and advice during problem solving which is typically not available without an instructor. CLIPSITS is divided into 10 lessons which mirror the first 10 chapters of the CLIPS User's Guide. This version of CLIPSITS is compatible with the Version 4.2 and 4.3 CLIPS User's Guide. However, the program does not cover any new features of CLIPS v4.3 that were added since the release of v4.2. The chapter numbers in the CLIPS User's Guide correspond directly with the lesson numbers in CLIPSITS. Each lesson in the program contains anywhere from 1 to 10 problems. Most of these have multiple parts. The student is given a subset of these problems from each lesson to work. The actual number of problems presented depends on how well the student masters the previous problem(s). The progression through these lessons is maintained in a personalized file under the student's name. As with most computer languages, there is usually more than one way to solve a problem. CLIPSITS attempts to be as flexible as possible and to allow as many correct solutions as possible. CLIPSITS gives the student the option of setting his/her own colors for the screen interface and the option of redefining special keystroke combinations used within the program. CLIPSITS requires an IBM PC compatible with 640K RAM and optional 2 or 3 button mouse. A 286- or 386-based machine is preferable. Performance will be somewhat slower on an XT class machine. The program must be installed on a hard disk with 825 KB space available. The program was developed in 1989. The standard distribution media is three 5.25" IBM PC DOS format diskettes. The program is also sold bundled with CLIPS for a special combined price as COS-10025. NOTE: Only the executable code is distributed. Supporting documentation is included on the diskettes. IBM, IBM PC and XT are registered trademarks of International Business Machines Corporation.

Description: F77NNS (A FORTRAN-77 Neural Network Simulator) simulates the popular back error propagation neural network. F77NNS is an ANSI-77 FORTRAN program designed to take advantage of vectorization when run on machines having this capability, but it will run on any computer with an ANSI-77 FORTRAN Compiler. Artificial neural networks are formed from hundreds or thousands of simulated neurons, connected to each other in a manner similar to biological nerve cells. Problems which involve pattern matching or system modeling readily fit the class of problems which F77NNS is designed to solve. The program's formulation trains a neural network using Rumelhart's back-propagation algorithm. Typically the nodes of a network are grouped together into clumps called layers. A network will generally have an input layer through which the various environmental stimuli are presented to the network, and an output layer for determining the network's response. The number of nodes in these two layers is usually tied to features of the problem being solved. Other layers, which form intermediate stops between the input and output layers, are called hidden layers. The back-propagation training algorithm can require massive computational resources to implement a large network such as a network capable of learning text-to-phoneme pronunciation rules as in the famous Sehnowski experiment. The Sehnowski neural network learns to pronounce 1000 common English words. The standard input data defines the specific inputs that control the type of run to be made, and input files define the NN in terms of the layers and nodes, as well as the input/output (I/O) pairs. The program has a restart capability so that a neural network can be solved in stages suitable to the user's resources and desires. F77NNS allows the user to customize the patterns of connections between layers of a network. The size of the neural network to be solved is limited only by the amount of random access memory (RAM) available to the user. The program has a memory requirement of about 900K. The standard distribution medium for this package is a .25 inch streaming magnetic tape cartridge in UNIX tar format. It is also available on a 3.5 inch diskette in UNIX tar format. F77NNS was developed in 1989.

Description: NETS, A Tool for the Development and Evaluation of Neural Networks, provides a simulation of Neural Network algorithms plus an environment for developing such algorithms. Neural Networks are a class of systems modeled after the human brain. Artificial Neural Networks are formed from hundreds or thousands of simulated neurons, connected to each other in a manner similar to brain neurons. Problems which involve pattern matching readily fit the class of problems which NETS is designed to solve. NETS uses the back propagation learning method for all of the networks which it creates. The nodes of a network are usually grouped together into clumps called layers. Generally, a network will have an input layer through which the various environment stimuli are presented to the network, and an output layer for determining the network's response. The number of nodes in these two layers is usually tied to some features of the problem being solved. Other layers, which form intermediate stops between the input and output layers, are called hidden layers. NETS allows the user to customize the patterns of connections between layers of a network. NETS also provides features for saving the weight values of a network during the learning process, which allows for more precise control over the learning process. NETS is an interpreter. Its method of execution is the familiar "read-evaluate-print" loop found in interpreted languages such as BASIC and LISP. The user is presented with a prompt which is the simulator's way of asking for input. After a command is issued, NETS will attempt to evaluate the command, which may produce more prompts requesting specific information or an error if the command is not understood. The typical process involved when using NETS consists of translating the problem into a format which uses input/output pairs, designing a network configuration for the problem, and finally training the network with input/output pairs until an acceptable error is reached. NETS allows the user to generate C code to implement the network loaded into the system. This permits the placement of networks as components, or subroutines, in other systems. In short, once a network performs satisfactorily, the Generate C Code option provides the means for creating a program separate from NETS to run the network. Other features: files may be stored in binary or ASCII format; multiple input propagation is permitted; bias values may be included; capability to scale data without writing scaling code; quick interactive testing of network from the main menu; and several options that allow the user to manipulate learning efficiency. NETS is written in ANSI standard C language to be machine independent. The Macintosh version (MSC-22108) includes code for both a graphical user interface version and a command line interface version. The machine independent version (MSC-21588) only includes code for the command line interface version of NETS 3.0. The Macintosh version requires a Macintosh II series computer and has been successfully implemented under System 7. Four executables are included on these diskettes, two for floating point operations and two for integer arithmetic. It requires Think C 5.0 to compile. A minimum of 1Mb of RAM is required for execution. Sample input files and executables for both the command line version and the Macintosh user interface version are provided on the distribution medium. The Macintosh version is available on a set of three 3.5 inch 800K Macintosh format diskettes. The machine independent version has been successfully implemented on an IBM PC series compatible running MS-DOS, a DEC VAX running VMS, a SunIPC running SunOS, and a CRAY Y-MP running UNICOS. Two executables for the IBM PC version are included on the MS-DOS distribution media, one compiled for floating point operations and one for integer arithmetic. The machine independent version is available on a set of three 5.25 inch 360K MS-DOS format diskettes (standard distribution medium) or a .25 inch streaming magnetic tape cartridge in UNIX tar format. NETS was developed in 1989 and updated in 1992. IBM PC is a registered trademark of International Business Machines. MS-DOS is a registered trademark of Microsoft Corporation. DEC, VAX, and VMS are trademarks of Digital Equipment Corporation. SunIPC and SunOS are trademarks of Sun Microsystems, Inc. CRAY Y-MP and UNICOS are trademarks of Cray Research, Inc.

Description: The primary purpose of NNETS (Neural Network Environment on a Transputer System) is to provide users a high degree of flexibility in creating and manipulating a wide variety of neural network topologies at processing speeds not found in conventional computing environments. To accomplish this purpose, NNETS supports back propagation and back propagation related algorithms. The back propagation algorithm used is an implementation of Rumelhart's Generalized Delta Rule. NNETS was developed on the INMOS Transputer. NNETS predefines a Back Propagation Network, a Jordan Network, and a Reinforcement Network to assist users in learning and defining their own networks. The program also allows users to configure other neural network paradigms from the NNETS basic architecture. The Jordan network is basically a feed forward network that has the outputs connected to a pseudo input layer. The state of the network is dependent on the inputs from the environment plus the state of the network. The Reinforcement network learns via a scalar feedback signal called reinforcement. The network propagates forward randomly. The environment looks at the outputs of the network to produce a reinforcement signal that is fed back to the network. NNETS was written for the INMOS C compiler D711B version 1.3 or later (MS-DOS version). A small portion of the software was written in the OCCAM language to perform the communications routing between processors. NNETS is configured to operate on a 4 X 10 array of Transputers in sequence with a Transputer based graphics processor controlled by a master IBM PC 286 (or better) Transputer. A RGB monitor is required which must be capable of 512 X 512 resolution. It must be able to receive red, green, and blue signals via BNC connectors. NNETS is meant for experienced Transputer users only. The program is distributed on 5.25 inch 1.2Mb MS-DOS format diskettes. NNETS was developed in 1991. Transputer and OCCAM are registered trademarks of Inmos Corporation. MS-DOS is a registered trademark of Microsoft Corporation. IBM PC is a registered trademark of International Business Machines.

Description: The C Language Integrated Production System, CLIPS, is a shell for developing expert systems. It is designed to allow artificial intelligence research, development, and delivery on conventional computers. The primary design goals for CLIPS are portability, efficiency, and functionality. For these reasons, the program is written in C. CLIPS meets or outperforms most micro- and minicomputer based artificial intelligence tools. CLIPS is a forward chaining rule-based language. The program contains an inference engine and a language syntax that provide a framework for the construction of an expert system. It also includes tools for debugging an application. CLIPS is based on the Rete algorithm, which enables very efficient pattern matching. The collection of conditions and actions to be taken if the conditions are met is constructed into a rule network. As facts are asserted either prior to or during a session, CLIPS pattern-matches the number of fields. Wildcards and variables are supported for both single and multiple fields. CLIPS syntax allows the inclusion of externally defined functions (outside functions which are written in a language other than CLIPS). CLIPS itself can be embedded in a program such that the expert system is available as a simple subroutine call. Advanced features found in CLIPS version 4.3 include an integrated microEMACS editor, the ability to generate C source code from a CLIPS rule base to produce a dedicated executable, binary load and save capabilities for CLIPS rule bases, and the utility program CRSV (Cross-Reference, Style, and Verification) designed to facilitate the development and maintenance of large rule bases. Five machine versions are available. Each machine version includes the source and the executable for that machine. The UNIX version includes the source and binaries for IBM RS/6000, Sun3 series, and Sun4 series computers. The UNIX, DEC VAX, and DEC RISC Workstation versions are line oriented. The PC version and the Macintosh version each contain a windowing variant of CLIPS as well as the standard line oriented version. The mouse/window interface version for the PC works with a Microsoft compatible mouse or without a mouse. This window version uses the proprietary CURSES library for the PC, but a working executable of the window version is provided. The window oriented version for the Macintosh includes a version which uses a full Macintosh-style interface, including an integrated editor. This version allows the user to observe the changing fact base and rule activations in separate windows while a CLIPS program is executing. The IBM PC version is available bundled with CLIPSITS, The CLIPS Intelligent Tutoring System for a special combined price (COS-10025). The goal of CLIPSITS is to provide the student with a tool to practice the syntax and concepts covered in the CLIPS User's Guide. It attempts to provide expert diagnosis and advice during problem solving which is typically not available without an instructor. CLIPSITS is divided into 10 lessons which mirror the first 10 chapters of the CLIPS User's Guide. The program was developed for the IBM PC series with a hard disk. CLIPSITS is also available separately as MSC-21679. The CLIPS program is written in C for interactive execution and has been implemented on an IBM PC computer operating under DOS, a Macintosh and DEC VAX series computers operating under VMS or ULTRIX. The line oriented version should run on any computer system which supports a full (Kernighan and Ritchie) C compiler or the ANSI standard C language. CLIPS was developed in 1986 and Version 4.2 was released in July of 1988. Version 4.3 was released in June of 1989.

Description: The C Language Integrated Production System, CLIPS, is a shell for developing expert systems. It is designed to allow artificial intelligence research, development, and delivery on conventional computers. The primary design goals for CLIPS are portability, efficiency, and functionality. For these reasons, the program is written in C. CLIPS meets or outperforms most micro- and minicomputer based artificial intelligence tools. CLIPS is a forward chaining rule-based language. The program contains an inference engine and a language syntax that provide a framework for the construction of an expert system. It also includes tools for debugging an application. CLIPS is based on the Rete algorithm, which enables very efficient pattern matching. The collection of conditions and actions to be taken if the conditions are met is constructed into a rule network. As facts are asserted either prior to or during a session, CLIPS pattern-matches the number of fields. Wildcards and variables are supported for both single and multiple fields. CLIPS syntax allows the inclusion of externally defined functions (outside functions which are written in a language other than CLIPS). CLIPS itself can be embedded in a program such that the expert system is available as a simple subroutine call. Advanced features found in CLIPS version 4.3 include an integrated microEMACS editor, the ability to generate C source code from a CLIPS rule base to produce a dedicated executable, binary load and save capabilities for CLIPS rule bases, and the utility program CRSV (Cross-Reference, Style, and Verification) designed to facilitate the development and maintenance of large rule bases. Five machine versions are available. Each machine version includes the source and the executable for that machine. The UNIX version includes the source and binaries for IBM RS/6000, Sun3 series, and Sun4 series computers. The UNIX, DEC VAX, and DEC RISC Workstation versions are line oriented. The PC version and the Macintosh version each contain a windowing variant of CLIPS as well as the standard line oriented version. The mouse/window interface version for the PC works with a Microsoft compatible mouse or without a mouse. This window version uses the proprietary CURSES library for the PC, but a working executable of the window version is provided. The window oriented version for the Macintosh includes a version which uses a full Macintosh-style interface, including an integrated editor. This version allows the user to observe the changing fact base and rule activations in separate windows while a CLIPS program is executing. The IBM PC version is available bundled with CLIPSITS, The CLIPS Intelligent Tutoring System for a special combined price (COS-10025). The goal of CLIPSITS is to provide the student with a tool to practice the syntax and concepts covered in the CLIPS User's Guide. It attempts to provide expert diagnosis and advice during problem solving which is typically not available without an instructor. CLIPSITS is divided into 10 lessons which mirror the first 10 chapters of the CLIPS User's Guide. The program was developed for the IBM PC series with a hard disk. CLIPSITS is also available separately as MSC-21679. The CLIPS program is written in C for interactive execution and has been implemented on an IBM PC computer operating under DOS, a Macintosh and DEC VAX series computers operating under VMS or ULTRIX. The line oriented version should run on any computer system which supports a full (Kernighan and Ritchie) C compiler or the ANSI standard C language. CLIPS was developed in 1986 and Version 4.2 was released in July of 1988. Version 4.3 was released in June of 1989.

Description: The PYROLASER package is an operating system for the Pyrometer Instrument Company's Pyrolaser. There are 6 individual programs in the PYROLASER package: two main programs, two lower level subprograms, and two programs which, although independent, function predominantly as macros. The package provides a quick and easy way to setup, control, and program a standard Pyrolaser. Temperature and emissivity measurements may be either collected as if the Pyrolaser were in the manual operations mode, or displayed on real time strip charts and stored in standard spreadsheet format for post-test analysis. A shell is supplied to allow macros, which are test-specific, to be easily added to the system. The Pyrolaser Simple Operation program provides full on-screen remote operation capabilities, thus allowing the user to operate the Pyrolaser from the computer just as it would be operated manually. The Pyrolaser Simple Operation program also allows the use of "quick starts". Quick starts provide an easy way to permit routines to be used as setup macros for specific applications or tests. The specific procedures required for a test may be ordered in a sequence structure and then the sequence structure can be started with a simple button in the cluster structure provided. One quick start macro is provided for continuous Pyrolaser operation. A subprogram, Display Continuous Pyr Data, is used to display and store the resulting data output. Using this macro, the system is set up for continuous operation and the subprogram is called to display the data in real time on strip charts. The data is simultaneously stored in a spreadsheet format. The resulting spreadsheet file can be opened in any one of a number of commercially available spreadsheet programs. The Read Continuous Pyrometer program is provided as a continuously run subprogram for incorporation of the Pyrolaser software into a process control or feedback control scheme in a multi-component system. The program requires the Pyrolaser to be set up using the Pyrometer String Transfer macro. It requires no inputs and provides temperature and emissivity as outputs. The Read Continuous Pyrometer program can be run continuously and the data can be sampled as often or as seldom as updates of temperature and emissivity are required. PYROLASER is written using the Labview software for use on Macintosh series computers running System 6.0.3 or later, Sun Sparc series computers running OpenWindows 3.0 or MIT's X Window System (X11R4 or X11R5), and IBM PC or compatibles running Microsoft Windows 3.1 or later. Labview requires a minimum of 5Mb of RAM on a Macintosh, 24Mb of RAM on a Sun, and 8Mb of RAM on an IBM PC or compatible. The Labview software is a product of National Instruments (Austin,TX; 800-433-3488), and is not included with this program. The standard distribution medium for PYROLASER is a 3.5 inch 800K Macintosh format diskette. It is also available on a 3.5 inch 720K MS-DOS format diskette, a 3.5 inch diskette in UNIX tar format, and a .25 inch streaming magnetic tape cartridge in UNIX tar format. An electronic copy of the documentation in Macintosh WordPerfect version 2.0.4 format is included on the distribution medium. Printed documentation is included in the price of the program. PYROLASER was developed in 1992.

Description: The Nickel Cadmium Battery Expert System-2 (NICBES2) is a prototype diagnostic expert system for Nickel Cadmium Battery Health Management. NICBES2 is intended to support evaluation of the performance of Hubble Space Telescope spacecraft batteries, and to alert personnel to possible malfunctions. To achieve this, NICBES2 provides a reasoning system supported by appropriate battery domain knowledge. NICBES2 oversees the status of the batteries by evaluating data gathered in orbit packets, and when the status so merits, raises an alarm and provides fault diagnosis as well as advice on the actions to be taken to remedy the particular alarm. In addition to diagnosis and advice, it provides status history of the batteries' health, and a graphical display capability to help in assimilation of the information by the operator. NICBES2 is composed of three cooperating processes driven by a program written in SunOS C. A serial port process gathers incoming data from an RS-232 connection and places it into a raw data pipe. The data handler processes read this information from the raw data pipe and perform statistical data reduction to generate a set of reduced data files per orbit. The expert system process starts the Quintus Prolog interpreter and the expert system and then uses the reduced data files for the generation of status and advice information. The expert system presents the user with an interface window composed of six subwindows: Battery Status, Advice Selection, Support, Battery Selection, Graphics, and Actions. The Battery status subwindow can provide a display of the current status of a battery. Similarly, advice on battery reconditioning, charging, and workload can be obtained from the Advice Selection subwindow. A display of trends for the last orbit and over a sequence of the last twelve orbits is available in the Graph subwindow. A WHY button is available to give the user an explanation of the rules that the expert system used in determining the current information. The Support subwindow contains an editor for altering the knowledge base. NICBES2 is written in C-language and Quintus Prolog for Sun series computers running SunOS. It requires 8Mb of RAM for execution. The Quintus ProWindows graphics system is required for graphical display, and a Postscript printer is required to print graphics. A DEC LSI-11 is required to send telemetry via a RS-232 connection. The program is available on a .25 inch streaming magnetic tape cartridge in UNIX tar format. NICBES2 was developed in 1989. Sun and SunOS are trademarks of Sun Microsystems, Inc. PostScript is a registered trademark of Adobe Systems Incorporated. UNIX is a registered trademark of AT&T Bell Laboratories. DEC LSI-11 is a trademark of Digital Equipment Corporation.

Description: Scale-space filtering is used to screen information obtained from signals that produce a complex curve (such as that in geographic and thermal analysis) to gain a truer representation of the area under analysis. PSF extends this technique to extract non-periodic hills and valleys from a signal. Because the signal's information is sometimes too complex to determine with certainty if some features are real or artificial, PSF calculates probabilities, with the extracted features corresponding to real events, in order to aid in determining the signal's accuracy. Since the probabilities associated with the features are derived from domain-specific statistics, it is (most likely) necessary to modify the program code to correspond to the user's particular domain. PSF also provides a standard scale-space filtering algorithm for use when the desired features can be identified with certainty or when it is not practical to get the domain-specific statistics. The PSF algorithm is based on Witkin's scale-space filtering theory. The program detects signal variations by finding the points of inflection in the input signal. The number and position of these points are dependent upon the scale of the derivative operators used to detect them. Therefore, instead of assuming any single scale to be correct, PSF identifies points of inflection in a large number of different scales. It then describes the curve according to the groups of points of inflection, across all scales, caused by the same physical process. PSF provides an output table giving the following information: the abscissa of the first inflection of the peak, the type of peak, the distance between the first and second inflection points, the abscissa of the peak, and the probability of the feature corresponding to a real event in the curve. The program will also list points representing a graphical image of the signal and detected peaks. This data can be used with a standard plotting program (not included) to display the signal and its features graphically. PSF is written in C language (49%) and Common LISP (51%) for use on a Sun SPARC workstation running the UNIX operating system. PSF requires 4Mb of RAM. The standard distribution medium for this program is a .25 streaming magnetic tape cartridge in UNIX tar format. It is also available on a 3.5 inch diskette in UNIX tar format. PSF was developed in 1991. Sun and SPARC are trademarks of Sun Microsystems, Inc. UNIX is a registered trademark of AT&T Bell Laboratories.

Description: The complex environment of the typical research laboratory requires flexible process control. This program provides natural language process control from an IBM PC or compatible machine. Sometimes process control schedules require changes frequently, even several times per day. These changes may include adding, deleting, and rearranging steps in a process. This program sets up a process control system that can either run without an operator, or be run by workers with limited programming skills. The software system includes three programs. Two of the programs, written in FORTRAN77, record data and control research processes. The third program, written in Pascal, generates the FORTRAN subroutines used by the other two programs to identify the user commands with the user-written device drivers. The software system also includes an input data set which allows the user to define the user commands which are to be executed by the computer. To set the system up the operator writes device driver routines for all of the controlled devices. Once set up, this system requires only an input file containing natural language command lines which tell the system what to do and when to do it. The operator can make up custom commands for operating and taking data from external research equipment at any time of the day or night without the operator in attendance. This process control system requires a personal computer operating under MS-DOS with suitable hardware interfaces to all controlled devices. The program requires a FORTRAN77 compiler and user-written device drivers. This program was developed in 1989 and has a memory requirement of about 62 Kbytes.

Description: CO-ST-IN is a program developed for NASA to help facilitate the study of Control Structure Interaction, the dynamic coupling between control systems and flexible structures. Current space structures are larger and more flexible than previous designs. At the same time, increased demands are being placed on the performance of control systems. For many space structures it is essential to analyze the interaction of control systems with structural flexibility. CO-ST-IN was designed to complement and enhance rather than to replace the structural dynamics and control system analysis tools already available at NASA. The functions performed by CO-ST-IN can be roughly divided into three areas: 1) data transfer between structural dynamics and control systems software (MSC/NASTRAN, I-DEAS, EASY5 and MATRIXx are currently supported to varying degrees); 2) modal selection at both the component and system level as a means of model reduction; and 3) simulation of the coupled system (given simple controllers). CO-ST-IN reduces the size of the structural model by selecting system modes on the basis of input/output coupling (three algorithms along with a number of other options are offered). This allows the analyst to use far fewer modes in the coupled analysis, since the program will select those which are most closely coupled to the structural inputs and outputs. Another special capability is the calculation of structural outputs such as element forces and stresses using either the mode acceleration or mode displacement approach directly within the coupled simulation. This eliminates the need to return to MSC/NASTRAN for recovery of this data, accelerating the turnaround time of analyses. The transfer of input forces for transient analysis in MSC/NASTRAN is also supported. CO-ST-IN was implemented on a DEC VAX with the VMS operating system. This FORTRAN77 program has a memory requirement of 9.4 MB. CO-ST-IN was developed in 1989.

Description: The Vibration Pattern Imager (VPI) system was designed to control and acquire data from laser vibrometer sensors. The PC computer based system uses a digital signal processing (DSP) board and an analog I/O board to control the sensor and to process the data. The VPI system was originally developed for use with the Ometron VPI Sensor (Ometron Limited, Kelvin House, Worsley Bridge Road, London, SE26 5BX, England), but can be readily adapted to any commercially available sensor which provides an analog output signal and requires analog inputs for control of mirror positioning. VPI's graphical user interface allows the operation of the program to be controlled interactively through keyboard and mouse-selected menu options. The main menu controls all functions for setup, data acquisition, display, file operations, and exiting the program. Two types of data may be acquired with the VPI system: single point or "full field". In the single point mode, time series data is sampled by the A/D converter on the I/O board at a user-defined rate for the selected number of samples. The position of the measuring point, adjusted by mirrors in the sensor, is controlled via a mouse input. In the "full field" mode, the measurement point is moved over a user-selected rectangular area with up to 256 positions in both x and y directions. The time series data is sampled by the A/D converter on the I/O board and converted to a root-mean-square (rms) value by the DSP board. The rms "full field" velocity distribution is then uploaded for display and storage. VPI is written in C language and Texas Instruments' TMS320C30 assembly language for IBM PC series and compatible computers running MS-DOS. The program requires 640K of RAM for execution, and a hard disk with 10Mb or more of disk space is recommended. The program also requires a mouse, a VGA graphics display, a Four Channel analog I/O board (Spectrum Signal Processing, Inc.; Westborough, MA), a break-out box and a Spirit-30 board (Sonitech International, Inc.; Wellesley, MA) which includes a TMS320C30 DSP processor, 256Kb zero wait state SRAM, and a daughter board with 8Mb one wait state DRAM. Please contact COSMIC for additional information on required hardware and software. In order to compile the provided VPI source code, a Microsoft C version 6.0 compiler, a Texas Instruments' TMS320C30 assembly language compiler, and the Spirit 30 run time libraries are required. A math co-processor is highly recommended. A sample MS-DOS executable is provided on the distribution medium. The standard distribution medium for this program is one 5.25 inch 360K MS-DOS format diskette. The contents of the diskettes are compressed using the PKWARE archiving tools. The utility to unarchive the files, PKUNZIP.EXE, is included. VPI was developed in 1991-1992.

Description: This control theory design package, called Optimal Regulator Algorithms for the Control of Linear Systems (ORACLS), was developed to aid in the design of controllers and optimal filters for systems which can be modeled by linear, time-invariant differential and difference equations. Optimal linear quadratic regulator theory, currently referred to as the Linear-Quadratic-Gaussian (LQG) problem, has become the most widely accepted method of determining optimal control policy. Within this theory, the infinite duration time-invariant problems, which lead to constant gain feedback control laws and constant Kalman-Bucy filter gains for reconstruction of the system state, exhibit high tractability and potential ease of implementation. A variety of new and efficient methods in the field of numerical linear algebra have been combined into the ORACLS program, which provides for the solution to time-invariant continuous or discrete LQG problems. The ORACLS package is particularly attractive to the control system designer because it provides a rigorous tool for dealing with multi-input and multi-output dynamic systems in both continuous and discrete form. The ORACLS programming system is a collection of subroutines which can be used to formulate, manipulate, and solve various LQG design problems. The ORACLS program is constructed in a manner which permits the user to maintain considerable flexibility at each operational state. This flexibility is accomplished by providing primary operations, analysis of linear time-invariant systems, and control synthesis based on LQG methodology. The input-output routines handle the reading and writing of numerical matrices, printing heading information, and accumulating output information. The basic vector-matrix operations include addition, subtraction, multiplication, equation, norm construction, tracing, transposition, scaling, juxtaposition, and construction of null and identity matrices. The analysis routines provide for the following computations: the eigenvalues and eigenvectors of real matrices; the relative stability of a given matrix; matrix factorization; the solution of linear constant coefficient vector-matrix algebraic equations; the controllability properties of a linear time-invariant system; the steady-state covariance matrix of an open-loop stable system forced by white noise; and the transient response of continuous linear time-invariant systems. The control law design routines of ORACLS implement some of the more common techniques of time-invariant LQG methodology. For the finite-duration optimal linear regulator problem with noise-free measurements, continuous dynamics, and integral performance index, a routine is provided which implements the negative exponential method for finding both the transient and steady-state solutions to the matrix Riccati equation. For the discrete version of this problem, the method of backwards differencing is applied to find the solutions to the discrete Riccati equation. A routine is also included to solve the steady-state Riccati equation by the Newton algorithms described by Klein, for continuous problems, and by Hewer, for discrete problems. Another routine calculates the prefilter gain to eliminate control state cross-product terms in the quadratic performance index and the weighting matrices for the sampled data optimal linear regulator problem. For cases with measurement noise, duality theory and optimal regulator algorithms are used to calculate solutions to the continuous and discrete Kalman-Bucy filter problems. Finally, routines are included to implement the continuous and discrete forms of the explicit (model-in-the-system) and implicit (model-in-the-performance-index) model following theory. These routines generate linear control laws which cause the output of a dynamic time-invariant system to track the output of a prescribed model. In order to apply ORACLS, the user must write an executive (driver) program which inputs the problem coefficients, formulates and selects the routines to be used to solve the problem, and specifies the desired output. There are three versions of ORACLS source code available for implementation: CDC, IBM, and DEC. The CDC version has been implemented on a CDC 6000 series computer with a central memory of approximately 13K (octal) of 60 bit words. The CDC version is written in FORTRAN IV, was developed in 1978, and last updated in 1986. The IBM version has been implemented on an IBM 370 series computer with a central memory requirement of approximately 300K of 8 bit bytes. The IBM version is written in FORTRAN IV and was generated in 1981. The DEC version has been implemented on a VAX series computer operating under VMS. The VAX version is written in FORTRAN 77 and was generated in 1986.

Description: The Interactive Controls Analysis (INCA) program was developed to provide a user friendly environment for the design and analysis of linear control systems, primarily feedback control systems. INCA is designed for use with both small and large order systems. Using the interactive graphics capability, the INCA user can quickly plot a root locus, frequency response, or time response of either a continuous time system or a sampled data system. The system configuration and parameters can be easily changed, allowing the INCA user to design compensation networks and perform sensitivity analysis in a very convenient manner. A journal file capability is included. This stores an entire sequence of commands, generated during an INCA session into a file which can be accessed later. Also included in INCA are a context-sensitive help library, a screen editor, and plot windows. INCA is robust to VAX-specific overflow problems. The transfer function is the basic unit of INCA. Transfer functions are automatically saved and are available to the INCA user at any time. A powerful, user friendly transfer function manipulation and editing capability is built into the INCA program. The user can do all transfer function manipulations and plotting without leaving INCA, although provisions are made to input transfer functions from data files. By using a small set of commands, the user may compute and edit transfer functions, and then examine these functions by using the ROOT_LOCUS, FREQUENCY_RESPONSE, and TIME_RESPONSE capabilities. Basic input data, including gains, are handled as single-input single-output transfer functions. These functions can be developed using the function editor or by using FORTRAN- like arithmetic expressions. In addition to the arithmetic functions, special functions are available to 1) compute step, ramp, and sinusoid functions, 2) compute closed loop transfer functions, 3) convert from S plane to Z plane with optional advanced Z transform, and 4) convert from Z plane to W plane and back. These capabilities allow the INCA user to perform block diagram algebraic manipulations quickly for functions in the S, Z, and W domains. Additionally, a versatile digital control capability has been included in INCA. Special plane transformations allow the user to easily convert functions from one domain to another. Other digital control capabilities include: 1) totally independent open loop frequency response analyses on a continuous plant, discrete control system with a delay, 2) advanced Z-transform capability for systems with delays, and 3) multirate sampling analyses. The current version of INCA includes Dynamic Functions (which change when a parameter changes), standard filter generation, PD and PID controller generation, incorporation of the QZ-algorithm (function addition, inverse Laplace), and describing functions that allow the user to calculate the gain and phase characteristics of a nonlinear device. The INCA graphic modes provide the user with a convenient means to document and study frequency response, time response, and root locus analyses. General graphics features include: 1) zooming and dezooming, 2) plot documentation, 3) a table of analytic computation results, 4) multiple curves on the same plot, and 5) displaying frequency and gain information for a specific point on a curve. Additional capabilities in the frequency response mode include: 1) a full complement of graphical methods Bode magnitude, Bode phase, Bode combined magnitude and phase, Bode strip plots, root contour plots, Nyquist, Nichols, and Popov plots; 2) user selected plot scaling; and 3) gain and phase margin calculation and display. In the time response mode, additional capabilities include: 1) support for inverse Laplace and inverse Z transforms, 2) support for various input functions, 3) closed loop response evaluation, 4) loop gain sensitivity analyses, 5) intersample time response for discrete systems using the advanced Z transform, and 6) closed loop time response using mixed plane (S, Z, W) operations with delay. A Graphics mode command was added to the current version of INCA, version 3.13, to produce Metafiles (graphic files) of the currently displayed plot. The metafile can be displayed and edited using the QPLOT Graphics Editor and Replotter for Metafiles (GERM) program included with the INCA package. The INCA program is written in Pascal and FORTRAN for interactive or batch execution and has been implemented on a DEC VAX series computer under VMS. Both source code and executable code are supplied for INCA. Full INCA graphics capabilities are supported for various Tektronix 40xx and 41xx terminals; DEC VT graphics terminals; many PC and Macintosh terminal emulators; TEK014 hardcopy devices such as the LN03 Laserprinter; and bit map graphics external hardcopy devices. Also included for the TEK4510 rasterizer users are a multiple copy feature, a wide line feature, and additional graphics fonts. The INCA program was developed in 1985, Version 2.04 was released in 1986, Version 3.00 was released in 1988, and Version 3.13 was released in 1989. An INCA version 2.0X conversion program is included.

Description: Expensive analysis programs are often combined with optimization procedures to solve engineering problems. An optimal solution requires numerous iterations between the analysis program and an optimizer. This often becomes prohibitive due to cost and amount of computer time needed to converge to an optimal solution. NETS/PROSSS was developed to provide a system for combining NETS (MSC-21588), a neural network program developed at NASA's Johnson Space Center, and the optimization program CONMIN (Constrained Function Minimization, ARC-10836) developed at Ames Research Center. After training, NETS approximates the results from the analysis program, possibly allowing the user to reach a near-optimal solution in much less time than before. These results can then be used as a starting point in a normal optimization process, possibly allowing the user to converge to an optimal solution in significantly fewer iterations. NETS/PROSSS is written in C-language and FORTRAN 77 for Sun series computers running SunOS. The required CONMIN and NETS v3.0 files are included in this package. The documentation for CONMIN and NETS are included with the documentation of NETS/PROSSS. The program requires 342K of RAM for execution. The standard distribution medium for this program is a .25 inch streaming magnetic tape cartridge in UNIX tar format. It is also available on a 3.5 inch diskette in UNIX tar format. NETS/PROSSS was developed in 1991.

Description: AESOP was developed to solve a number of problems associated with the design of controls and state estimators for linear time-invariant systems. The systems considered are modeled in state-variable form by a set of linear differential and algebraic equations with constant coefficients. Two key problems solved by AESOP are the linear quadratic regulator (LQR) design problem and the steady-state Kalman filter design problem. AESOP is designed to be used in an interactive manner. The user can solve design problems and analyze the solutions in a single interactive session. Both numerical and graphical information are available to the user during the session. The AESOP program is structured around a list of predefined functions. Each function performs a single computation associated with control, estimation, or system response determination. AESOP contains over sixty functions and permits the easy inclusion of user defined functions. The user accesses these functions either by inputting a list of desired functions in the order they are to be performed, or by specifying a single function to be performed. The latter case is used when the choice of function and function order depends on the results of previous functions. The available AESOP functions are divided into several general areas including: 1) program control, 2) matrix input and revision, 3) matrix formation, 4) open-loop system analysis, 5) frequency response, 6) transient response, 7) transient function zeros, 8) LQR and Kalman filter design, 9) eigenvalues and eigenvectors, 10) covariances, and 11) user-defined functions. The most important functions are those that design linear quadratic regulators and Kalman filters. The user interacts with AESOP when using these functions by inputting design weighting parameters and by viewing displays of designed system response. Support functions obtain system transient and frequency responses, transfer functions, and covariance matrices. AESOP can also provide the user with open-loop system information including stability, controllability, and observability. The AESOP program is written in FORTRAN IV for interactive execution and has been implemented on an IBM 3033 computer using TSS 370. As currently configured, AESOP has a central memory requirement of approximately 2 Megs of 8 bit bytes. Memory requirements can be reduced by redimensioning arrays in the AESOP program. Graphical output requires adaptation of the AESOP plot routines to whatever device is available. The AESOP program was developed in 1984.

Description: The Automation Technology Branch of NASA's Langley Research Center is employing increasingly complex degrees of operator/robot cooperation (telerobotics). A good relationship between the operator and computer is essential for smooth performance by a telerobotic system. ESG (Expert Script Generator) is a software package that automatically generates high-level task objective commands from the NASA Intelligent Systems Research Lab's (ISRL's) complex menu-driven language. ESG reduces errors and makes the telerobotics lab accessible to researchers who are not familiar with the comprehensive language developed by ISRL for interacting with the various systems of the ISRL testbed. ESG incorporates expert system technology to capture the typical rules of operation that a skilled operator would use. The result is an operator interface which optimizes the system's capability to perform a task remotely in a hazardous environment, in a timely manner, and without undue stress to the operator, while minimizing the chance for operator errors that may damage equipment. The intricate menu-driven command interface which provides for various control modes of both manipulators and their associated sensors in the TeleRobotic System Simulation (TRSS) has a syntax which is both irregular and verbose. ESG eliminates the following two problems with this command "language": 1) knowing the correct command sequence to accomplish a task, and 2) inputting a known command sequence without typos and other errors. ESG serves as an additional layer of interface, working in conjunction with the menu command processor, not supplanting it. By specifying task-level commands, such as GRASP, CONNECT, etc., ESG will generate the appropriate menu elements to accomplish the task. These elements will be collected in a script file which can then be executed by the ISRL menu command processor. In addition, the operator can extend the list of task-level commands to include customized tasks composed of sub-task commands. This mechanism gives the operator the ability to build a task-hierarchy tree of increasingly powerful commands. ESG also provides automatic regeneration of scripts based on system knowledge of telerobotic environment updates. The commands generated by ESG may be displayed at the terminal screen and/or stored. ESG is implemented as a rule-based expert system written in CLIPS (C Language Integrated Production System). The system consists of a knowledge-base of task heuristics, a static (unchanged during execution) database which describes the physical features of objects, and a dynamic (may change as a result of task achievement) database which maintains changes in the environment. Capabilites are provided for adding new environmental objects and for modifying existing objects and configuration data. Options are available for interactively viewing both the static and dynamic attribute values of database items. Execution of the ESG may be suspended to allow access to system-level functions. ESG was implemented on a VAX 11/780 with the VMS 4.7 operating system using a VT100 compatible terminal. Its source code is 47% CLIPS and 53% C-language, with a memory requirement of approximately 205 KB. The program was developed in 1988.

Description: The expert system called EXADS was developed to aid users of the Automated Design Synthesis (ADS) general purpose optimization program. Because of the general purpose nature of ADS, it is difficult for a nonexpert to select the best choice of strategy, optimizer, and one-dimensional search options from the one hundred or so combinations that are available. EXADS aids engineers in determining the best combination based on their knowledge of the problem and the expert knowledge previously stored by experts who developed ADS. EXADS is a customized application of the AESOP artificial intelligence program (the general version of AESOP is available separately from COSMIC. The ADS program is also available from COSMIC.) The expert system consists of two main components. The knowledge base contains about 200 rules and is divided into three categories: constrained, unconstrained, and constrained treated as unconstrained. The EXADS inference engine is rule-based and makes decisions about a particular situation using hypotheses (potential solutions), rules, and answers to questions drawn from the rule base. EXADS is backward-chaining, that is, it works from hypothesis to facts. The rule base was compiled from sources such as literature searches, ADS documentation, and engineer surveys. EXADS will accept answers such as yes, no, maybe, likely, and don't know, or a certainty factor ranging from 0 to 10. When any hypothesis reaches a confidence level of 90% or more, it is deemed as the best choice and displayed to the user. If no hypothesis is confirmed, the user can examine explanations of why the hypotheses failed to reach the 90% level. The IBM PC version of EXADS is written in IQ-LISP for execution under DOS 2.0 or higher with a central memory requirement of approximately 512K of 8 bit bytes. This program was developed in 1986.

Description: CFORM was developed by the Kennedy Space Center Robotics Lab to assist in linear control system design and analysis using closed form and transient response mechanisms. The program computes the closed form solution and transient response of a linear (constant coefficient) differential equation. CFORM allows a choice of three input functions: the Unit Step (a unit change in displacement); the Ramp function (step velocity); and the Parabolic function (step acceleration). It is only accurate in cases where the differential equation has distinct roots, and does not handle the case for roots at the origin (s=0). Initial conditions must be zero. Differential equations may be input to CFORM in two forms - polynomial and product of factors. In some linear control analyses, it may be more appropriate to use a related program, Linear Control System Design and Analysis (KSC-11376), which uses root locus and frequency response methods. CFORM was written in VAX FORTRAN for a VAX 11/780 under VAX VMS 4.7. It has a central memory requirement of 30K. CFORM was developed in 1987.

Description: This control theory design package, called Optimal Regulator Algorithms for the Control of Linear Systems (ORACLS), was developed to aid in the design of controllers and optimal filters for systems which can be modeled by linear, time-invariant differential and difference equations. Optimal linear quadratic regulator theory, currently referred to as the Linear-Quadratic-Gaussian (LQG) problem, has become the most widely accepted method of determining optimal control policy. Within this theory, the infinite duration time-invariant problems, which lead to constant gain feedback control laws and constant Kalman-Bucy filter gains for reconstruction of the system state, exhibit high tractability and potential ease of implementation. A variety of new and efficient methods in the field of numerical linear algebra have been combined into the ORACLS program, which provides for the solution to time-invariant continuous or discrete LQG problems. The ORACLS package is particularly attractive to the control system designer because it provides a rigorous tool for dealing with multi-input and multi-output dynamic systems in both continuous and discrete form. The ORACLS programming system is a collection of subroutines which can be used to formulate, manipulate, and solve various LQG design problems. The ORACLS program is constructed in a manner which permits the user to maintain considerable flexibility at each operational state. This flexibility is accomplished by providing primary operations, analysis of linear time-invariant systems, and control synthesis based on LQG methodology. The input-output routines handle the reading and writing of numerical matrices, printing heading information, and accumulating output information. The basic vector-matrix operations include addition, subtraction, multiplication, equation, norm construction, tracing, transposition, scaling, juxtaposition, and construction of null and identity matrices. The analysis routines provide for the following computations: the eigenvalues and eigenvectors of real matrices; the relative stability of a given matrix; matrix factorization; the solution of linear constant coefficient vector-matrix algebraic equations; the controllability properties of a linear time-invariant system; the steady-state covariance matrix of an open-loop stable system forced by white noise; and the transient response of continuous linear time-invariant systems. The control law design routines of ORACLS implement some of the more common techniques of time-invariant LQG methodology. For the finite-duration optimal linear regulator problem with noise-free measurements, continuous dynamics, and integral performance index, a routine is provided which implements the negative exponential method for finding both the transient and steady-state solutions to the matrix Riccati equation. For the discrete version of this problem, the method of backwards differencing is applied to find the solutions to the discrete Riccati equation. A routine is also included to solve the steady-state Riccati equation by the Newton algorithms described by Klein, for continuous problems, and by Hewer, for discrete problems. Another routine calculates the prefilter gain to eliminate control state cross-product terms in the quadratic performance index and the weighting matrices for the sampled data optimal linear regulator problem. For cases with measurement noise, duality theory and optimal regulator algorithms are used to calculate solutions to the continuous and discrete Kalman-Bucy filter problems. Finally, routines are included to implement the continuous and discrete forms of the explicit (model-in-the-system) and implicit (model-in-the-performance-index) model following theory. These routines generate linear control laws which cause the output of a dynamic time-invariant system to track the output of a prescribed model. In order to apply ORACLS, the user must write an executive (driver) program which inputs the problem coefficients, formulates and selects the routines to be used to solve the problem, and specifies the desired output. There are three versions of ORACLS source code available for implementation: CDC, IBM, and DEC. The CDC version has been implemented on a CDC 6000 series computer with a central memory of approximately 13K (octal) of 60 bit words. The CDC version is written in FORTRAN IV, was developed in 1978, and last updated in 1989. The IBM version has been implemented on an IBM 370 series computer with a central memory requirement of approximately 300K of 8 bit bytes. The IBM version is written in FORTRAN IV and was generated in 1981. The DEC version has been implemented on a VAX series computer operating under VMS. The VAX version is written in FORTRAN 77 and was generated in 1986.

Description: The objective of this work is to validate mathematically derived clock synchronization theories and their associated algorithms through experiment. Two theories are considered, the Interactive Convergence Clock Synchronization Algorithm and the Mid-Point Algorithm. Special clock circuitry was designed and built so that several operating conditions and failure modes (including malicious failures) could be tested. Both theories are shown to predict conservative upper bounds (i.e., measured values of clock skew were always less than the theory prediction). Insight gained during experimentation led to alternative derivations of the theories. These new theories accurately predict the clock system's behavior. It is found that a 100% penalty is paid to tolerate worst case failures. It is also shown that under optimal conditions (with minimum error and no failures) the clock skew can be as much as 3 clock ticks. Clock skew grows to 6 clock ticks when failures are present. Finally, it is concluded that one cannot rely solely on test procedures or theoretical analysis to predict worst case conditions. conditions.

Description: The paper presents the synthesis of neural network based feedback laws for dynamic systems using the computed optimal and time histories of the state and control variables. The efficacy of the proposed approach has been successfully demonstrated on a minimum time orbit injection problem. If the method is found to be effective to real life problems with many state and control variables, it can used for a variety of guidance and control problems.

Description: A preliminary investigation has been conducted on instrumented fasteners for use as sensors to measure the shear loads transmitted by individual fasteners installed in double-splice joints. Calibration and load verification tests were conducted for instrumented fasteners installed at three fastener torque levels. Results from calibration tests show that the shear strains obtained from the instrumented fasteners vary linearly with the applied load and that the instrumented fasteners can be effectively used to measure shear loads transmitted by individual fasteners installed in double-splice joints. Tests were also conducted with three instumented fasteners installed in a typical double-splice joint. The test results showed that the load distribution between individual fasteners is dependent on the location of the fastener in the joint and the fastener torque level.

Description: This paper is on the control of nonlinear-nonstationary vibration of a frame-stringer structure resulting from high levels of excitation from a nearby supersonic jet exhaust. The structure exhibits periodic, chaotic, or random behaviors when forced by high-intensity sound from a supersonic jet exhaust with 'shock' loading superimposed on a broadband response. The time history of the pressure, showing the rotation and flapping of the shock structure in the jet column due to large-scale instabilities, indicates that the response is not only nonlinear but also nonstationary. The acoustic pressure radiated by the structure also contains shocks and the formation of harmonics with distance. Control of the structural response is achieved by actively forcing the structure with an actuator at the shock oscillation frequency whose amplitude is locked into a self-control cycle. Results show that the peak power level is reduced by a factor of 63, or 18 dB. As a result, new broadband components emerge with at least four harmonics. At accelerating and decelerating supersonic speeds, the exhaust from the jet induces higher transient loading on the nearby flexible structure due to the occurrence of multiple shock from the jet.

Description: This paper presents a novel method to design decentralized controllers for large complex flexible structures by using the idea of joint decoupling. Decoupling of joint degrees of freedom from the interior degrees of freedom is achieved by setting the joint actuator commands to cancel the internal forces exerting on the joint degrees of freedom. By doing so, the interactions between substructures are eliminated. The global structure control design problem is then decomposed into several substructure control design problems. Control commands for interior actuators are set to be localized state feedback using decentralized observers for state estimation. The proposed decentralized controllers can operate successfully at the individual substructure level as well as at the global structure level. Not only control design but also control implementation is decentralized. A two-component mass-spring-damper system is used as an example to demonstrate the proposed method.

Description: A common approach to supervised classification and prediction in artificial intelligence and statistical pattern recognition is the use of decision trees. A tree is "grown" from data using a recursive partitioning algorithm to create a tree which has good prediction of classes on new data. Standard algorithms are CART (by Breiman Friedman, Olshen and Stone) and ID3 and its successor C4 (by Quinlan). As well as reimplementing parts of these algorithms and offering experimental control suites, IND also introduces Bayesian and MML methods and more sophisticated search in growing trees. These produce more accurate class probability estimates that are important in applications like diagnosis. IND is applicable to most data sets consisting of independent instances, each described by a fixed length vector of attribute values. An attribute value may be a number, one of a set of attribute specific symbols, or it may be omitted. One of the attributes is delegated the "target" and IND grows trees to predict the target. Prediction can then be done on new data or the decision tree printed out for inspection. IND provides a range of features and styles with convenience for the casual user as well as fine-tuning for the advanced user or those interested in research. IND can be operated in a CART-like mode (but without regression trees, surrogate splits or multivariate splits), and in a mode like the early version of C4. Advanced features allow more extensive search, interactive control and display of tree growing, and Bayesian and MML algorithms for tree pruning and smoothing. These often produce more accurate class probability estimates at the leaves. IND also comes with a comprehensive experimental control suite. IND consists of four basic kinds of routines: data manipulation routines, tree generation routines, tree testing routines, and tree display routines. The data manipulation routines are used to partition a single large data set into smaller training and test sets. The generation routines are used to build classifiers. The test routines are used to evaluate classifiers and to classify data using a classifier. And the display routines are used to display classifiers in various formats. IND is written in C-language for Sun4 series computers. It consists of several programs with controlling shell scripts. Extensive UNIX man entries are included. IND is designed to be used on any UNIX system, although it has only been thoroughly tested on SUN platforms. The standard distribution medium for IND is a .25 inch streaming magnetic tape cartridge in UNIX tar format. An electronic copy of the documentation in PostScript format is included on the distribution medium. IND was developed in 1992.

Description: The program AUTOCLASS III, Automatic Class Discovery from Data, uses Bayesian probability theory to provide a simple and extensible approach to problems such as classification and general mixture separation. Its theoretical basis is free from ad hoc quantities, and in particular free of any measures which alter the data to suit the needs of the program. As a result, the elementary classification model used lends itself easily to extensions. The standard approach to classification in much of artificial intelligence and statistical pattern recognition research involves partitioning of the data into separate subsets, known as classes. AUTOCLASS III uses the Bayesian approach in which classes are described by probability distributions over the attributes of the objects, specified by a model function and its parameters. The calculation of the probability of each object's membership in each class provides a more intuitive classification than absolute partitioning techniques. AUTOCLASS III is applicable to most data sets consisting of independent instances, each described by a fixed length vector of attribute values. An attribute value may be a number, one of a set of attribute specific symbols, or omitted. The user specifies a class probability distribution function by associating attribute sets with supplied likelihood function terms. AUTOCLASS then searches in the space of class numbers and parameters for the maximally probable combination. It returns the set of class probability function parameters, and the class membership probabilities for each data instance. AUTOCLASS III is written in Common Lisp, and is designed to be platform independent. This program has been successfully run on Symbolics and Explorer Lisp machines. It has been successfully used with the following implementations of Common LISP on the Sun: Franz Allegro CL, Lucid Common Lisp, and Austin Kyoto Common Lisp and similar UNIX platforms; under the Lucid Common Lisp implementations on VAX/VMS v5.4, VAX/Ultrix v4.1, and MIPS/Ultrix v4, rev. 179; and on the Macintosh personal computer. The minimum Macintosh required is the IIci. This program will not run under CMU Common Lisp or VAX/VMS DEC Common Lisp. A minimum of 8Mb of RAM is required for Macintosh platforms and 16Mb for workstations. The standard distribution medium for this program is a .25 inch streaming magnetic tape cartridge in UNIX tar format. It is also available on a 3.5 inch diskette in UNIX tar format and a 3.5 inch diskette in Macintosh format. An electronic copy of the documentation is included on the distribution medium. AUTOCLASS was developed between March 1988 and March 1992. It was initially released in May 1991. Sun is a trademark of Sun Microsystems, Inc. UNIX is a registered trademark of AT&T Bell Laboratories. DEC, VAX, VMS, and ULTRIX are trademarks of Digital Equipment Corporation. Macintosh is a trademark of Apple Computer, Inc. Allegro CL is a registered trademark of Franz, Inc.

Description: Pixel Pusher is a Macintosh application used for viewing and performing minor enhancements on imagery. It will read image files in JPL's two primary image formats- VICAR and PDS - as well as the Macintosh PICT format. VICAR (NPO-18076) handles an array of image processing capabilities which may be used for a variety of applications including biomedical image processing, cartography, earth resources, and geological exploration. Pixel Pusher can also import VICAR format color lookup tables for viewing images in pseudocolor (256 colors). This program currently supports only eight bit images but will work on monitors with any number of colors. Arbitrarily large image files may be viewed in a normal Macintosh window. Color and contrast enhancement can be performed with a graphical "stretch" editor (as in contrast stretch). In addition, VICAR images may be saved as Macintosh PICT files for exporting into other Macintosh programs, and individual pixels can be queried to determine their locations and actual data values. Pixel Pusher is written in Symantec's Think C and was developed for use on a Macintosh SE30, LC, or II series computer running System Software 6.0.3 or later and 32 bit QuickDraw. Pixel Pusher will only run on a Macintosh which supports color (whether a color monitor is being used or not). The standard distribution medium for this program is a set of three 3.5 inch Macintosh format diskettes. The program price includes documentation. Pixel Pusher was developed in 1991 and is a copyrighted work with all copyright vested in NASA. Think C is a trademark of Symantec Corporation. Macintosh is a registered trademark of Apple Computer, Inc.

Description: The S3 Data Acquisition and Control System, S3DACS, was developed for the Environmental Test Laboratory and Space Simulator at NASA's Jet Propulsion Laboratory. The program is used for monitoring, controlling, and recording information acquired during tests and presenting this information in various formats for easy access by a large number of users. All testing is initiated by a setup procedure that defines what will be tested, limits to be checked, formulas to use, etc. Test results (e.g. temperature, resistance) are then automatically stored in a database for real time display and for future reference. Measurements obtained may be used in various computations defined for the test and selectively presented in tabular, graphical, or electronic representation. Reports may show current or historical events. The S3DACS network software is written in FoxPro/LAN 1.02 and 80386 Assembler for IBM PC and compatibles running MS-DOS 3.31 or higher. Machine requirements include: an 80386 33MHz machine with 10Mb RAM set up as a file server; an 80386 33MHz machine with 4Mb RAM connected to a FLUKE 2240B or 2280 data acquisition device; and an 80386 20MHz machine with 5Mb RAM used as a workstation. Also needed is a National Instruments General Purpose Interface Bus-compatible (GP-IB) Board to enable S3DACS to communicate with IEEE-488 control instruments. Software requirements include: Novell Netware 386 for network management; FoxPro/LAN 1.02 for database management; QEMM 386 version 5.0 for memory management; and DGE version 4, Saywhat, Viewlib, and DBSHOW for graphics and screen displays. The previous list of hardware is the minimum configuration which will allow installation of S3DACS. The addition of workstations and data acquisition devices can occur transparently. S3DACS is distributed on one 5.25 inch 1.2Mb MS-DOS format diskette. The extensive documentation includes a Quick Reference Guide, a Software User's Manual, a Computer Systems Operator's Manual, and a Software Programmer's Manual. The source code is provided in PKZIP format, and the PKUNZIP utility is included. Compiling the assembler source requires Microsoft's Assembler version 5.10. Due to the complexity of this software package, COSMIC strongly recommends the purchase and review of the documentation prior to the purchase of the program. S3DACS was developed in 1990. S3DACS is a copyrighted work with all copyright vested in NASA. All tradenames used are the property of their respective corporate owners.

Description: The goal of this work is development of fully electrostatically suspended and rebalancing angular rate sensing micro-gyroscope fabricated according to standard VLSI techniques. Fabrication of test structures is proceeding. Off chip electronics for the electrostatic sensing and driving circuits has been tested. The prototype device will be assembled in a hybrid construction including the FET input stages of the sensors.

Description: In this paper we study the robustness with respect to stability of the closed-loop system with collocated rate sensor using LQG (mean square rate) optimized compensators. Our main result is that the transmission zeros of the compensator are precisely the structure modes when the actuator/sensor locations are 'pinned' and/or 'clamped': i.e., motion in the direction sensed is not allowed. We have stability even under parameter mismatch, except in the unlikely situation where such a mode frequency of the assumed system coincides with an undamped mode frequency of the real system and the corresponding mode shape is an eigenvector of the compensator transfer function matrix at that frequency. For a truncated modal model - such as that of the NASA LaRC Phase Zero Evolutionary model - the transmission zeros of the corresponding compensator transfer function can be interpreted as the structure modes when motion in the directions sensed is prohibited.

Description: In optimal placement of actuators for stochastic systems, it is commonly assumed that the actuator noise variances are not related to the feedback matrix and the actuator locations. In this paper, we will discuss the limitation of that assumption and develop a more practical noise variance model. Various properties associated with optimal actuator placement under the assumption of this noise variance model are discovered through the analytical study of a second order system.

Description: Optimal regulation of hyperbolic systems in the presence of unknown disturbances is considered. Necessary conditions for determining the optimal control that tracks a desired trajectory in the presence of the worst possible perturbations are developed. The results also characterize the worst possible disturbance that the system will be able to tolerate before any degradation of the system performance. Numerical results on the control of a vibrating beam are presented.

Description: This paper presents the development of a general-purpose fuzzy logic (FL) control methodology for isolating the external vibratory disturbances of space-based devices. According to the desired performance specifications, a full investigation regarding the development of an FL controller was done using different scenarios, such as variances of passive reaction-compensating components and external disturbance load. It was shown that the proposed FL controller is robust in that the FL-controlled system closely follows the prespecified ideal reference model. The comparative study also reveals that the FL-controlled system achieves significant improvement in reducing vibrations over passive systems.

Description: Remote viewing is critical for teleoperations, but the inherent limitations of standard video reduce the operator's effectiveness. These limitations have been compensated for in many ways, from using the operator's adaptability, to augmenting his capability with feedback from a variety of sensors and simulations. Omniview can overcome some of these limitations and improve the operator's efficiency without adding additional sensors or computational burden. It can minimize the potential collisions with facility equipment, provide peripheral vision, and display multiple images simultaneously from a single input device. The Omniview technology provides electronic pan, tilt, magnify, and rotational orientation within a hemispherical field-of-view without any moving parts. Image sizes, viewing directions, scale, offset, etc., may be adjusted to fit operator needs. This paper discusses the derivation of the image transformation, the design of the electronics, and two applications to telepresence that are under development. These are Video Emulated Tweening (VET), and Manipulator Guidance and Positioning (ManGAP). The VET effort uses Omniview to compensate for time-delayed video in teleoperation of remote vehicles. In ManGAP two Omniview systems are used to provide two sets of orientation vectors to points in the field-of-view (FOV). These vectors then provide absolute position information to both control the position of the telerobot, and to avoid collisions with the work sight equipment.

Description: A fuzzy-neural control system simulation was developed for the control of a camera platform used to observe aircraft on final approach to an aircraft carrier. The fuzzy-neural approach to control combines the structure of a fuzzy knowledge base with a supervised neural network's ability to adapt and improve. The performance characteristics of this hybrid system were compared to those of a fuzzy system and a neural network system developed independently to determine if the fusion of these two technologies offers any advantage over the use of one or the other. The results of this study indicate that the fuzzy-neural approach to control offers some advantages over either fuzzy or neural control alone.

Description: A fuzzy classifier system that discovers rules for controlling a mathematical model of a pH titration system was developed by researchers at the U.S. Bureau of Mines (USBM). Fuzzy classifier systems successfully combine the strengths of learning classifier systems and fuzzy logic controllers. Learning classifier systems resemble familiar production rule-based systems, but they represent their IF-THEN rules by strings of characters rather than in the traditional linguistic terms. Fuzzy logic is a tool that allows for the incorporation of abstract concepts into rule based-systems, thereby allowing the rules to resemble the familiar 'rules-of-thumb' commonly used by humans when solving difficult process control and reasoning problems. Like learning classifier systems, fuzzy classifier systems employ a genetic algorithm to explore and sample new rules for manipulating the problem environment. Like fuzzy logic controllers, fuzzy classifier systems encapsulate knowledge in the form of production rules. The results presented in this paper demonstrate the ability of fuzzy classifier systems to generate a fuzzy logic-based process control system.

Description: Neural networks are an outgrowth of interdisciplinary studies concerning the brain. These studies are guiding the field of Artificial Intelligence towards the, so-called, 6th Generation Computer. Enormous amounts of resources have been poured into R/D. Wavelet Transforms (WT) have replaced Fourier Transforms (FT) in Wideband Transient (WT) cases since the discovery of WT in 1985. The list of successful applications includes the following: earthquake prediction; radar identification; speech recognition; stock market forecasting; FBI finger print image compression; and telecommunication ISDN-data compression.

Description: Artificial neural systems (ANS), also known as neural networks, are an attempt to develop computer systems that emulate the neural reasoning behavior of biological neural systems (e.g. the human brain). As such, they are loosely based on biological neural networks. The ANS consists of a series of nodes (neurons) and weighted connections (axons) that, when presented with a specific input pattern, can associate specific output patterns. It is essentially a highly complex, nonlinear, mathematical relationship or transform. These constructs have two significant properties that have proven useful to the authors in signal processing and process modeling: noise tolerance and complex pattern recognition. Specifically, the authors have developed a new network learning algorithm that has resulted in the successful application of ANS's to high speed signal processing and to developing models of highly complex processes. Two of the applications, the Weld Bead Geometry Control System and the Welding Penetration Monitoring System, are discussed in the body of this paper.

Description: Neural network capabilities include automatic and organized handling of complex information, quick adaptation to continuously changing environments, nonlinear modeling, and parallel implementation. This viewgraph presentation presents Bellcore work on applications, learning chip computational function, learning system block diagram, neural network equalization, broadband access control, calling-card fraud detection, software reliability prediction, and conclusions.

Description: Over the past three years, our group has concentrated on the application of neural network methods to the training of controllers for real-world systems. This presentation describes our approach, surveys what we have found to be important, mentions some contributions to the field, and shows some representative results. Topics discussed include: (1) executing model studies as rehearsal for experimental studies; (2) the importance of correct derivatives; (3) effective training with second-order (DEKF) methods; (4) the efficacy of time-lagged recurrent networks; (5) liberation from the tyranny of the control cycle using asynchronous truncated backpropagation through time; and (6) multistream training for robustness. Results from model studies of automotive idle speed control serve as examples for several of these topics.

Description: The research mission is the development of computer assisted diagnostic (CAD) methods for improved diagnosis of medical images including digital x-ray sensors and tomographic imaging modalities. The CAD algorithms include advanced methods for adaptive nonlinear filters for image noise suppression, hybrid wavelet methods for feature segmentation and enhancement, and high convergence neural networks for feature detection and VLSI implementation of neural networks for real time analysis. Other missions include (1) implementation of CAD methods on hospital based picture archiving computer systems (PACS) and information networks for central and remote diagnosis and (2) collaboration with defense and medical industry, NASA, and federal laboratories in the area of dual use technology conversion from defense or aerospace to medicine.

Description: This viewgraph presentation presents four working analog VLSI vision chips: (1) time-derivative retina, (2) zero-crossing chip, (3) resistive fuse, and (4) figure-ground chip; work in progress on computing motion and neuromorphic systems; and conceptual and practical lessons learned.

Description: The problem under consideration in this viewgraph presentation is to understand, predict, and control the fluid mechanics of dynamic maneuvers, unsteady boundary layers, and vortex dominated flows. One solution is the application of neural networks demonstrating closed-loop control. Neural networks offer unique opportunities: simplify modeling of three dimensional, vortex dominated, unsteady separated flow fields; are effective means for controlling unsteady aerodynamics; and address integration of sensors, controllers, and time lags into adaptive control systems.

Description: Irvine Sensors Corporation (ISC), working closely with JPL under BMDO/ONR sponsorship, is developing a radically new neural computing technology. Primarily aimed at discrimination and target recognition for BMDO missile interceptor applications, it appears to have near term commercial applicability to such problems as handwriting and face recognition, just to name two. In its earliest form it will be able to perform inner product computation using 262 thousand 64x64 templates (weighted synapse arrays) where the 64(exp 5) weights can all be changed every millisecond. Internal switching provides an inherent capability to zoom, translate, or rotate the templates. The 3D silicon architecture is manufactured on a commercial, high volume DRAM production line at very low cost, enabling its commercialization. Two technology thrusts are beginning: in the first, the 64 layer capability of 3DANN-I will be extended to 1024 layers and beyond. In the second layer size will be shrunk to 2-3 millimeters to reduce layer costs. Our workshop goal is to expose this technology to the neural network community as an emerging tool for their use and to obtain their desire for its future development.

Description: Electronic and optoelectronic hardware implementations of highly parallel computing architectures address several ill-defined and/or computation-intensive problems not easily solved by conventional computing techniques. The concurrent processing architectures developed are derived from a variety of advanced computing paradigms including neural network models, fuzzy logic, and cellular automata. Hardware implementation technologies range from state-of-the-art digital/analog custom-VLSI to advanced optoelectronic devices such as computer-generated holograms and e-beam fabricated Dammann gratings. JPL's concurrent processing devices group has developed a broad technology base in hardware implementable parallel algorithms, low-power and high-speed VLSI designs and building block VLSI chips, leading to application-specific high-performance embeddable processors. Application areas include high throughput map-data classification using feedforward neural networks, terrain based tactical movement planner using cellular automata, resource optimization (weapon-target assignment) using a multidimensional feedback network with lateral inhibition, and classification of rocks using an inner-product scheme on thematic mapper data. In addition to addressing specific functional needs of DOD and NASA, the JPL-developed concurrent processing device technology is also being customized for a variety of commercial applications (in collaboration with industrial partners), and is being transferred to U.S. industries. This viewgraph p resentation focuses on two application-specific processors which solve the computation intensive tasks of resource allocation (weapon-target assignment) and terrain based tactical movement planning using two extremely different topologies. Resource allocation is implemented as an asynchronous analog competitive assignment architecture inspired by the Hopfield network. Hardware realization leads to a two to four order of magnitude speed-up over conventional techniques and enables multiple assignments, (many to many), not achievable with standard statistical approaches. Tactical movement planning (finding the best path from A to B) is accomplished with a digital two-dimensional concurrent processor array. By exploiting the natural parallel decomposition of the problem in silicon, a four order of magnitude speed-up over optimized software approaches has been demonstrated.

Description: We have previously reported on the use of neural networks for detection and identification of faults in complex microprocessor controlled powertrain systems. The data analyzed in those studies consisted of the full spectrum of signals passing between the engine and the real-time microprocessor controller. The specific task of the classification system was to classify system operation as nominal or abnormal and to identify the fault present. The primary concern in earlier work was the identification of faults, in sensors or actuators in the powertrain system as it was exercised over its full operating range. The use of data from a variety of sources, each contributing some potentially useful information to the classification task, is commonly referred to as sensor fusion and typifies the type of problems successfully addressed using neural networks. In this work we explore the application of neural networks to a different diagnostic problem, the diagnosis of faults in newly manufactured engines and the utility of neural networks for process control.

Description: A recording density of 10 Gbit/sq. in. is being pursued by a number of companies and universities in the National Storage Industry Consortium. It is widely accepted that this goal will be achieved in the laboratory within a few years. In this paper approaches to achieving 100 Gbit/sq. in. storage densities are considered. A major obstacle to continued scaling of magnetic recording to higher densities is that as the bit size is reduced, the grain size in the magnetic media must be reduced in order that media noise does not become so large that the signal to noise ratio (SNR) degrades sufficiently to make detection impossible. At 100 Gbit/sq. in., the bit size is only 0.006 square micrometers, which, in order to achieve 30 dB SNR, requires a grain size of about 2.5 nm. Such small grains are subject to thermal instability, and the recorded information will degrade over time unless the magnetic anisotropy of the materials used is increased significantly, or the media thickness is made much larger than expected on the basis of scaling today's longitudinal media thickness.

Description: The Intelligent Satellite Control Software (ISACS) for the geoMagnetic tail observation satellite named GEOTAIL (launched in July 1992) has been successfully developed. ISACS has made it possible by applying Artificial Intelligence (AI) technology including an expert system to autonomously generate a tracking schedule, which originally used to be conducted manually. Using ISACS, a satellite operator can generate a maximum four day period of stored command stream autonomously and can easily confirm its safety. The ISACS system has another function -- to diagnose satellite troubles and to suggest necessary remedies. The workload of satellite operators has drastically been reduced since ISACS has been introduced into the operations of GEOTAIL.

Description: Time delay and small capacity of communication are the primary constraint in super long distance telerobotic systems such as astronautical robotic tasks. Intelligent telerobotics is thought to break this constraint. We aim to realize this super long distance telerobotic system with object handling knowledge base and intelligent monitoring. We will discuss physical and technical factors for this purpose.

Description: This paper reports experiments involving the handling of flexible parts (e.g. wires) when using a teleoperated system with time delay. The task is principally a peg-in-hole task involving the wrapping of a wire around two posts on the task-board. It is difficult to estimate the effects of the flexible parts; therefore, on-line teleoperation is indispensable for this class of unpredictable task. We first propose a teleoperation system based on the predictive image display, then describe an experimental teleoperation testbed with a four second transmission time delay. Finally, we report on wire handling operations that were performed to evaluate the performance of this system. Those experiments will contribute to future advanced experiments for the MITI ETS-7 mission.

Description: The focus of this paper is on the detection of natural, as opposed to human-made, objects. The distinction is important because, in the context of image analysis, natural objects tend to possess much greater variability in appearance than human-made objects. Hence, we shall focus primarily on the use of algorithms that 'learn by example' as the basis for image exploration. The 'learn by example' approach is potentially more generally applicable compared to model-based vision methods since domain scientists find it relatively easier to provide examples of what they are searching for versus describing a model.

Description: A new containerless image furnace with a microwave discharge plasma lamp and electrostatic positioning device was developed for the use of the microgravity experiment on the Japanese experimental module (JEM). The electrostatic positioning system was tested under the reduced gravity environment in the MU-300 aircraft. Solid specimens (maximum weight is 1.3 gr and 10 mm in diameter) and water drops (maximum weight is 0.11 gr and 6 mm in diameter) were successfully controlled under the 0.02G environment. Rotation control of the dielectric specimen was also possible by means of supplying a rotating electric field while the specimen is levitating. The measured rotation speed of the glass shell specimen (0.08 gr, 10 mm) was up to 110 rpm, when the rotating field frequency was 6Hz.

Description: KATE (Knowledge-based Autonomous Test Engineer) is a model-based software system developed in the Artificial Intelligence Laboratory at the Kennedy Space Center for monitoring, fault detection, and control of launch vehicles and ground support systems. This report describes two software efforts which enhance the functionality and usability of KATE. The first addition, a flow solver, adds to KATE a tool for modeling the flow of liquid in a pipe system. The second addition adds support for editing KATE knowledge base files to the Emacs editor. The body of this report discusses design and implementation issues having to do with these two tools. It will be useful to anyone maintaining or extending either the flow solver or the editor enhancements.

Description: One of the most important contributions to neural networks comes from Kohonen, Helsinki/Espoo, Finland, who had the idea of self-organizating maps in 1981. He verified his idea by an algorithm of which many applications make use of. The impetus for this idea came from biology, a field where the Europeans have always been very active at several research laboratories. The challenge was to model the self-organization found in the brain. Today one goal is the development of more sophisticated neurons which model the biological neurons more exactly. They should come to a better performance of neural nets with only a few complex neurons instead of many simple ones. A lot of application concepts arise from this idea: Kohonen himself applied it to speech recognition, but the project did not overcome much more than the recognition of the numerals one to ten at that time. A more promising application for self-organizing maps is process control and process monitoring. Several proposals were made which concern parameter classification of semiconductor technologies, design of integrated circuits, and control of chemical processes. Self-organizing maps were applied to robotics. The neural concept was introduced into electric power systems. At Dortmund we are working on a system which has to monitor the quality and the reliability of gears and electrical motors in equipment installed in coal mines. The results are promising and the probability to apply the system in the field is very high. A special feature of the system is that linguistic rules which are embedded in a fuzzy controller analyze the data of the self-organizing map in regard to life expectation of the gears. It seems that the fuzzy technique will introduce the technology of neural networks in a tandem mode. These technologies together with the genetic algorithms start to form the attractive field of computational intelligence.

Description: Neural networks plus hidden Markov models (HMM) can provide excellent detection and false alarm rate performance in fault detection applications, as shown in this viewgraph presentation. Modified models allow for novelty detection. Key contributions of neural network models are: (1) excellent nonparametric discrimination capability; (2) a good estimator of posterior state probabilities, even in high dimensions, and thus can be embedded within overall probabilistic model (HMM); and (3) simple to implement compared to other nonparametric models. Neural network/HMM monitoring model is currently being integrated with the new Deep Space Network (DSN) antenna controller software and will be on-line monitoring a new DSN 34-m antenna (DSS-24) by July, 1994.

Description: There are available today many data storage devices that serve the diverse application requirements of the consumer, professional entertainment, and computer data processing industries. Storage technologies include semiconductors, several varieties of optical disk, optical tape, magnetic disk, and many varieties of magnetic tape. In some cases, devices are developed with specific characteristics to meet specification requirements. In other cases, an existing storage device is modified and adapted to a different application. For magnetic tape storage devices, examples of the former case are 3480/3490 and QIC device types developed for the high end and low end segments of the data processing industry respectively, VHS, Beta, and 8 mm formats developed for consumer video applications, and D-1, D-2, D-3 formats developed for professional video applications. Examples of modified and adapted devices include 4 mm, 8 mm, 12.7 mm and 19 mm computer data storage devices derived from consumer and professional audio and video applications. With the conversion of the consumer and professional entertainment industries from analog to digital storage and signal processing, there have been increasing references to the 'convergence' of the computer data processing and entertainment industry technologies. There has yet to be seen, however, any evidence of convergence of data storage device types. There are several reasons for this. The diversity of application requirements results in varying degrees of importance for each of the tape storage characteristics.

Description: Current operational capabilities of tape recording for Very Long Baseline Interferometry (VLBI) at the Haystack Observatory allow 0.7 terabytes (12 hours at 128 Mb/s) of data to be stored in a 128 cu. inch volume. On-going efforts are aimed at full time 1 Gb/s operation with two 36-channel headstacks. Applications for linear digital tape recording, with suitable development of thin-film head arrays, suggest a volume density exceeding 1 TB/cu. inch to be achievable in the future.

Description: A plan has been formulated and selected for a NASA Phase 2 SBIR award for using the VLBA tape recorder for recording general data. The VLBA tape recorder is a high-speed, high-density linear tape recorder developed for Very Long Baseline Interferometry (VLBI) which is presently capable of recording at rates up to 2 Gbit/sec and holding up to 1 Terabyte of data on one tape, using a special interface and not employing error correction. A general-purpose interface and error correction will be added so that the recorder can be used in other high-speed, high-capacity applications.

Description: The state-of-the-art in computing technology is rapidly attaining the performance necessary to implement many early vision algorithms at real-time rates. This new capability is helping to accelerate progress in vision research by improving our ability to evaluate the performance of algorithms in dynamic environments. In particular, we are becoming much more aware of the relative stability of various visual measurements in the presence of camera motion and system noise. This new processing speed is also allowing us to raise our sights toward accomplishing much higher-level processing tasks, such as figure-ground separation and active object tracking, in real-time. This paper describes a methodology for using early visual measurements to accomplish higher-level tasks; it then presents an overview of the high-speed accelerators developed at Teleos to support early visual measurements. The final section describes the successful deployment of a real-time vision system to provide visual perception for the Extravehicular Activity Helper/Retriever robotic system in tests aboard NASA's KC135 reduced gravity aircraft.

Description: In this paper, we propose a new fuzzy structural matching scheme for space stereo vision which is based on the fuzzy properties of regions of images and effectively reduces the computational burden in the following low level matching process. Three dimensional distance images of a space truss structural model are estimated using this scheme from stereo images sensed by Charge Coupled Device (CCD) TV cameras.

Description: Future space robots require position and orientation tracking with visual feedback control to track and capture floating objects and satellites. We developed a four-circle mark that is useful for this purpose. With this mark, four geometric center positions as feature points can be extracted from the mark by simple image processing. We also developed a position and orientation measurement method that uses the four feature points in our mark. The mark gave good enough image measurement accuracy to let space robots approach and contact objects. A visual feedback control system using this mark enabled a robot arm to track a target object accurately. The control system was able to tolerate a time delay of 2 seconds.

Description: When trying to fly an aircraft as smoothly as possible it is a good idea to use the derivatives of the pilot command instead of using the actual control. This idea was implemented with splines and control theory, in a system that tries to model an aircraft. Computer calculations in Matlab show that it is impossible to receive enough smooth control signals by this way. This is due to the fact that the splines not only try to approximate the test function, but also its derivatives. A perfect traction is received but we have to pay in very peaky control signals and accelerations.

Description: In this report the problem we are going to study is the interpolation of a set of points in the plane with the use of control theory. We will discover how different systems generate different kinds of splines, cubic and exponential, and investigate the effect that the different systems have on the tracking problems. Actually we will see that the important parameters will be the two eigenvalues of the control matrix.

Description: There has been considerable interest recently in constrained mechanics and variational problems. This is in part due to applied interests (such as 'non-holonomic mechanics in robotics') and in other part due to the fact that several schools of 'pure' mathematics have found that this classical subject is of importance for what they are trying to do. I have made various attempts at developing these subjects since my Lincoln lab days of the late 1950's. In this Chapter, I will sketch a Unified point of view, using Cartan's approach with differential forms. This has the advantage from the C-O-R viewpoint being developed in this Volume that the extension from 'smooth' to 'generalized' data is very systematic and algebraic. (I will only deal with the 'smooth' point of view in this Chapter; I will develop the 'generalized function' material at a later point.) The material presented briefly here about Variational Calculus and Constrained Mechanics can be found in more detail in my books, 'Differential Geometry and the Calculus of Variations', 'Lie Algebras and Quantum Mechanics', and 'Geometry, Physics and Systems'.

Description: In addition to major crown projects such as the Mobile Servicing System for Space Station, the Canadian Space Agency is also engaged in internal, industrial and academic research and development activities in robotics and other space-related areas of science and technology. These activities support current and future space projects, and lead to technology development which can be spun off to terrestrial applications, thus satisfying the Agency's objective of providing economic benefits to the public at large through its space-related work.

Description: About 40% of the budget of a scientific spacecraft mission is usually consumed by Mission Operations & Data Analysis (MO&DA) with MO driving these costs. In the current practice, MO is separated from spacecraft design and comes in focus relatively late in the mission life cycle. As a result, spacecraft may be designed that are very difficult to operate. NASA centers have extensive MO expertise but often lessons learned in one mission are not exploited for other parallel or future missions. A significant reduction of MO costs is essential to ensure a continuing and growing access to space for the scientific community. We are addressing some of these issues with a highly automated payload operations and command system for an existing mission, the Extreme Ultraviolet Explorer (EUVE). EUVE is currently operated jointly by the Goddard Space Flight Center (GSFC), responsible for spacecraft operations, and the Center for Extreme Ultraviolet Astrophysics (CEA) of the University of California, Berkeley, which controls the telescopes and scientific instruments aboard the satellite. The new automated system is being developed by a team including personnel from the NASA Ames Research Center (ARC), the Jet Propulsion Laboratory (JPL) and the Center for EUV Astrophysics (CEA). An important goal of the project is to provide AI-based technology that can be easily operated by nonspecialists in AI. Another important goal is the reusability of the techniques for other missions. Models of the EUVE spacecraft need to be built both for planning/scheduling and for monitoring. In both cases, our modeling tools allow the assembly of a spacecraft model from separate sub-models of the various spacecraft subsystems. These sub-models are reusable; therefore, building mission operations systems for another small satellite mission will require choosing pre-existing modules, reparametrizing them with respect to the actual satellite telemetry information, and reassembling them in a new model. We briefly describe the EUVE mission and indicate why it is particularly suitable for the task. Then we briefly outline our current work in mission planning/scheduling and spacecraft and instrument health monitoring.

Description: DI is short for Distributed Intelligence for Ground/Space Systems and the DI Study is one in a series of ESA projects concerned with the development of new concepts and architectures for future autonomous spacecraft systems. The kick-off of DI was in January 1994 and the planned duration is three years. The background of DI is the desire to design future ground/space systems with a higher degree of autonomy than seen in today's missions. The aim of introducing autonomy in spacecraft systems is to: (1) lift the role of the spacecraft operators from routine work and basic troubleshooting to supervision; (2) ease access to and increase availability of spacecraft resources; (3) carry out basic mission planning for users; (4) enable missions which have not yet been feasible due to eg. propagation delays, insufficient ground station coverage etc.; and (5) possibly reduce mission cost. The study serves to identify the feasibility of using state-of-the-art technologies in the area of planning, scheduling, fault detection using model-based diagnosis and knowledge processing to obtain a higher level of autonomy in ground/space systems.

Description: Although the analysis of anomalous behavior of satellites is difficult because it is a very complex process, it is important to be able to make an accurate assessment in a timely manner when the anomaly is observed. Spacecraft operators may have to take corrective action or to 'safe' the spacecraft; space-environment forecasters may have to assess the environmental situation and issue warnings and alerts regarding hazardous conditions, and scientists and engineers may want to gain knowledge for future designs to mitigate the problems. Anomalies can be hardware problems, software errors, environmentally induced, or even the cause of workmanship. Spacecraft anomalies attributable to electrostatic discharges have been known to cause command errors. A goal is to develop an automated system based on this concept to reduce the number of personnel required to operate large programs or missions such as Hubble Space Telescope (HST) and Mission to Planet Earth (MTPE). Although expert systems to detect anomalous behavior of satellites during operations are established, diagnosis of the anomaly is a complex procedure and is a new development.

Description: At McDonnell Douglas Aerospace (MDA), an artificial neural network based control system has been developed and implemented to control laser heating for the fiber placement composite manufacturing process. This neurocontroller learns an approximate inverse model of the process on-line to provide performance that improves with experience and exceeds that of conventional feedback control techniques. When untrained, the control system behaves as a proportional plus integral (PI) controller. However after learning from experience, the neural network feedforward control module provides control signals that greatly improve temperature tracking performance. Faster convergence to new temperature set points and reduced temperature deviation due to changing feed rate have been demonstrated on the machine. A Cerebellar Model Articulation Controller (CMAC) network is used for inverse modeling because of its rapid learning performance. This control system is implemented in an IBM compatible 386 PC with an A/D board interface to the machine.

Description: Image compression for both still and moving images is an extremely important area of investigation, with numerous applications to videoconferencing, interactive education, home entertainment, and potential applications to earth observations, medical imaging, digital libraries, and many other areas. We describe work on a neural network methodology to compress/decompress still and moving images. We use the 'point-process' type neural network model which is closer to biophysical reality than standard models, and yet is mathematically much more tractable. We currently achieve compression ratios of the order of 120:1 for moving grey-level images, based on a combination of motion detection and compression. The observed signal-to-noise ratio varies from values above 25 to more than 35. The method is computationally fast so that compression and decompression can be carried out in real-time. It uses the adaptive capabilities of a set of neural networks so as to select varying compression ratios in real-time as a function of quality achieved. It also uses a motion detector which will avoid retransmitting portions of the image which have varied little from the previous frame. Further improvements can be achieved by using on-line learning during compression, and by appropriate compensation of nonlinearities in the compression/decompression scheme. We expect to go well beyond the 250:1 compression level for color images with good quality levels.

Description: Included in this viewgraph presentation on intelligent neuroprocessors for launch vehicle health management systems (HMS) are the following: where the flight failures have been in launch vehicles; cumulative delay time; breakdown of operations hours; failure of Mars Probe; vehicle health management (VHM) cost optimizing curve; target HMS-STS auxiliary power unit location; APU monitoring and diagnosis; and integration of neural networks and fuzzy logic.

Description: The purpose of neurophysiological monitoring of the 'acute care' patient is to allow the accurate recognition of changing or deteriorating neurological function as close to the moment of occurrence as possible, thus permitting immediate intervention. Results confirm that: (1) neural networks are able to accurately identify electroencephalogram (EEG) patterns and evoked potential (EP) wave components, and measuring EP waveform latencies and amplitudes; (2) neural networks are able to accurately detect EP and EEG recordings that have been contaminated by noise; (3) the best performance was obtained consistently with the back propagation network for EP and the HONN for EEG's; (4) neural network performed consistently better than other methods evaluated; and (5) neural network EEG and EP analyses are readily performed on multichannel data.

Description: Since 1989, Martin Marietta has invested in the development of an innovative concept for robust real-time pattern recognition for any two-dimensioanal sensor. This concept has been tested in simulation, and in laboratory and field hardware, for a number of DOD and commercial uses from automatic target recognition to manufacturing inspection. We have now joined Rose Health Care Systems in developing its use for medical diagnostics. The concept is based on determining regions of interest by using optical Fourier bandpassing as a scene segmentation technique, enhancing those regions using wavelet filters, passing the enhanced regions to a neural network for analysis and initial pattern identification, and following this initial identification with confirmation by optical correlation. The optical scene segmentation and pattern confirmation are performed by the same optical module. The neural network is a recursive error minimization network with a small number of connections and nodes that rapidly converges to a global minimum.

Description: Engineering neural network systems are best known for their abilities to adapt to the changing characteristics of the surrounding environment by adjusting system parameter values during the learning process. Rapid advances in analog current-mode design techniques have made possible the implementation of major neural network functions in custom VLSI chips. An electrically programmable analog synapse cell with large dynamic range can be realized in a compact silicon area. New designs of the synapse cells, neurons, and analog processor are presented. A synapse cell based on Gilbert multiplier structure can perform the linear multiplication for back-propagation networks. A double differential-pair synapse cell can perform the Gaussian function for radial-basis network. The synapse cells can be biased in the strong inversion region for high-speed operation or biased in the subthreshold region for low-power operation. The voltage gain of the sigmoid-function neurons is externally adjustable which greatly facilitates the search of optimal solutions in certain networks. Various building blocks can be intelligently connected to form useful industrial applications. Efficient data communication is a key system-level design issue for large-scale networks. We also present analog neural processors based on perceptron architecture and Hopfield network for communication applications. Biologically inspired neural networks have played an important role towards the creation of powerful intelligent machines. Accuracy, limitations, and prospects of analog current-mode design of the biologically inspired vision processing chips and cellular neural network chips are key design issues.

Description: Document analysis is one of the main applications of machine vision today and offers great opportunities for neural net circuits. Despite more and more data processing with computers, the number of paper documents is still increasing rapidly. A fast translation of data from paper into electronic format is needed almost everywhere, and when done manually, this is a time consuming process. Markets range from small scanners for personal use to high-volume document analysis systems, such as address readers for the postal service or check processing systems for banks. A major concern with present systems is the accuracy of the automatic interpretation. Today's algorithms fail miserably when noise is present, when print quality is poor, or when the layout is complex. A common approach to circumvent these problems is to restrict the variations of the documents handled by a system. In our laboratory, we had the best luck with circuits implementing basic functions, such as convolutions, that can be used in many different algorithms. To illustrate the flexibility of this approach, three applications of the NET32K circuit are described in this short viewgraph presentation: locating address blocks, cleaning document images by removing noise, and locating areas of interest in personal checks to improve image compression. Several of the ideas realized in this circuit that were inspired by neural nets, such as analog computation with a low resolution, resulted in a chip that is well suited for real-world document analysis applications and that compares favorably with alternative, 'conventional' circuits.

Description: A flow geometry and flow rate for mixed flowing gas testing is proposed. Use of an impinging jet of humid polluted air can provide a uniform and reproducible exposure of coupons of metal-based magnetic media. Numerical analysis of the fluid flow and mass transfer in such as system has shown that samples confined within a distance equal to the nozzle radius on the surface of impingement are uniformly accessible to pollutants in the impinging gas phase. The critical factor is the nozzle height above the surface of impingement. In particular, the uniformity of exposure is less than plus/minus 2% for a volumetric flow rate of 1600 cm(exp 3)/minute total flow with the following specifications: For a one inch nozzle, the height of the nozzle opening above the stage should be 0.177 inches; for a 2 inch nozzle - 0.390 inches. Not only is the distribution uniform, but one can calculate the maximum delivery rate of pollutants to the samples for comparison with the observed deterioration.

Description: Digital recording may take advantage of many types of media, but usually a preferred type of drive or transport emerges for each. In magnetic tape recording, two approaches have emerged in which essentially the same medium is tracked in two radically different ways. This paper compares the characteristics of Rotary- and Stationary-Head transports in an attempt to establish which approach might be considered for a given application. The conclusion is that in many cases there is no obvious choice based on recording physics and that often the choice will be made on the experimental knowledge of the designer.

Description: Current concepts of robot-supported operations for space laboratories (payload servicing, inspection, repair, and ORU exchange) are mainly based on the concept of 'interactive autonomy' which implies autonomous behavior of the robot according to predefined timelines, predefined sequences of elementary robot operations and within predefined world models supplying geometrical and other information for parameter instantiation on the one hand, and the ability to override and change the predefined course of activities by human intervention on the other hand. Although in principle a very powerful and useful concept, in practice the confinement of the robot to the abstract world models and predefined activities appears to reduce the robot's stability within real world uncertainties and its applicability to non-predefined parts of the world, calling for frequent corrective interaction by the operator, which in itself may be tedious and time-consuming. Methods are presented to improve this situation by incorporating 'robotic skills' into the concept of interactive autonomy.

Description: Many dynamic systems operate in select operating regions, each exhibiting characteristic modes of behavior. It is traditional to employ standard adjustable gain proportional-integral-derivative (PID) loops in such systems where no apriori model information is available. However, for controlling inlet pressure for rocket engine testing, problems in fine tuning, disturbance accommodation, and control gains for new profile operating regions (for research and development) are typically encountered. Because of the capability of capturing I/O peculiarities, using NETS, a back propagation trained neural network is specified. For select operating regions, the neural network controller is simulated to be as robust as the PID controller. For a comparative analysis, the higher order moment neural array (HOMNA) method is used to specify a second neural controller by extracting critical exemplars from the I/O data set. Furthermore, using the critical exemplars from the HOMNA method, a third neural controller is developed using NETS back propagation algorithm. All controllers are benchmarked against each other.

Description: Artificial neural network models can make amazing computations. These models are explained along with their application in problems associated with fighting crime. Specific problems addressed are identification of people using face recognition, speaker identification, and fingerprint and handwriting analysis (biometric authentication).

Description: Several years ago when INTEL and China Lake designed the ETANN chip, analog VLSI appeared to be the only way to do high density neural computing. In the last five years, however, digital parallel processing chips capable of performing neural computation functions have evolved to the point of rough equality with analog chips in system level computational density. The Naval Air Warfare Center, China Lake, has developed a real time, hardware and software system designed to implement and evaluate biologically inspired retinal and cortical models. The hardware is based on the Adaptive Solutions Inc. massively parallel CNAPS system COHO boards. Each COHO board is a standard size 6U VME card featuring 256 fixed point, RISC processors running at 20 MHz in a SIMD configuration. Each COHO board has a companion board built to support a real time VSB interface to an imaging seeker, a NTSC camera, and to other COHO boards. The system is designed to have multiple SIMD machines each performing different corticomorphic functions. The system level software has been developed which allows a high level description of corticomorphic structures to be translated into the native microcode of the CNAPS chips. Corticomorphic structures are those neural structures with a form similar to that of the retina, the lateral geniculate nucleus, or the visual cortex. This real time hardware system is designed to be shrunk into a volume compatible with air launched tactical missiles. Initial versions of the software and hardware have been completed and are in the early stages of integration with a missile seeker.

Description: This paper describes the Multimission VICAR (Video Image Communication and Retrieval) Planner (MVP) (Chien 1994) system, which uses artificial intelligence planning techniques (Iwasaki & Friedland, 1985, Pemberthy & Weld, 1992, Stefik, 1981) to automatically construct executable complex image processing procedures (using models of the smaller constituent image processing subprograms) in response to image processing requests made to the JPL Multimission Image Processing Laboratory (MIPL). The MVP system allows the user to specify the image processing requirements in terms of the various types of correction required. Given this information, MVP derives unspecified required processing steps and determines appropriate image processing programs and parameters to achieve the specified image processing goals. This information is output as an executable image processing program which can then be executed to fill the processing request.

Description: In earth observation or planetary exploration it is necessary to have more and, more autonomous systems, able to adapt to unpredictable situations. This imposes the use, in artificial systems, of new concepts in cognition, based on the fact that perception should not be separated from recognition and decision making levels. This means that low level signal processing (perception level) should interact with symbolic and high level processing (decision level). This paper is going to describe the new concept of active vision, implemented in Distributed Artificial Intelligence by Dassault Aviation following a 'structuralist' principle. An application to spatial image interpretation is given, oriented toward flexible robotics.

Description: The Jet Propulsion Laboratory (JPL) and the Electrotechnical Laboratory (ETL) have recently initiated a cooperative R&D effort in telerobotics. This new effort, sponsored by the U.S. National Aeronautics and Space Administration (NASA) and Japan's Ministry of International Trade and Industry (MITI), has two major themes. First, our work broadens the outreach of space telerobotics R&D to international technical collaboration and facilities usage in the United States and Japan. Second, our work fosters development and demonstration of new operator interface technologies to improve the flexibility and reliability of ground-to-orbit telerobotic operations. This new technology is important, given the continuing imperatives to off-load platform maintenance from the extravehicular activity/intravehicular activity (EVA/IVA) crew to on-board robot assists under direct ground mission control.

Description: Beginning next century, several schemes for sending a planetary rover to the moon or Mars are being planned. As part of the development program, autonomous navigation technology is being studied to allow the rover the ability to move autonomously over a long range of unknown planetary surface. In the previous study, we ran the autonomous navigation experiment on an outdoor test terrain by using a rover test-bed that was controlled by a conventional sense-plan-act method. In some cases during the experiment, a problem occurred with the rover moving into untraversable areas. To improve this situation, a new control technique has been developed that gives the rover the ability of reacting to the outputs of the proximity sensors, a reaction behavior if you will. We have developed a new rover test-bed system on which an autonomous navigation experiment was performed using the newly developed control technique. In this outdoor experiment, the new control technique effectively produced the control command for the rover to avoid obstacles and be guided to the goal point safely.

Description: The paper proposes a new architecture for autonomously generating and managing movement plans of planetary rovers. The system utilizes the uniform representation of the instantaneous subgoals in the form of virtual sensor states and the autonomous generation of the subsumption type plan network, which are expected to lead to the capability to pursue the overall goal while efficiently managing various unpredicted anomalies in a partially unknown, ill-structured environment such as a planetary surface.

Description: Any attempt to introduce automation into the monitoring of complex physical systems must start from a robust anomaly detection capability. This task is far from straightforward, for a single definition of what constitutes an anomaly is difficult to come by. In addition, to make the monitoring process efficient, and to avoid the potential for information overload on human operators, attention focusing must also be addressed. When an anomaly occurs, more often than not several sensors are affected, and the partially redundant information they provide can be confusing, particularly in a crisis situation where a response is needed quickly. The focus of this paper is a new technique for attention focusing. The technique involves reasoning about the distance between two frequency distributions, and is used to detect both anomalous system parameters and 'broken' causal dependencies. These two forms of information together isolate the locus of anomalous behavior in the system being monitored.