ALBERT

Description: It is not yet clear of what type, and how much, intelligence is needed for a planetary rover to function semi-autonomously on a planetary surface. Current designs assume an advanced AI system that maintains a detailed map of its journeys and the surroundings, and that carefully calculates and tests every move in advance. To achieve these abilities, and because of the limitations of space-qualified electronics, the supporting rover is quite sizable, massing a large fraction of a ton, and requiring technology advances in everything from power to ground operations. An alternative approach is to use a behavior driven control scheme. Recent research has shown that many complex tasks may be achieved by programming a robot with a set of behaviors and activation or deactivating a subset of those behaviors as required by the specific situation in which the robot finds itself. Behavior control requires much less computation than is required by tradition AI planning techniques. The reduced computation requirements allows the entire rover to be scaled down as appropriate (only down-link communications and payload do not scale under these circumstances). The missions that can be handled by the real-time control and operation of a set of small, semi-autonomous, interacting, behavior-controlled planetary rovers are discussed.

Description: A planetary rover will be traversing largely unknown and often unknowable terrain. In addition to geometric obstacles such as cliffs, rocks, and holes, it may also have to deal with non-geometric hazards such as soft soil and surface breakthroughs which often cannot be detected until rover is in imminent danger. Therefore, the rover must monitor its progress throughout a traverse, making sure to stay on course and to detect and act on any previously unseen hazards. Its onboard planning system must decide what sensors to monitor, what landmarks to take position readings from, and what actions to take if something should go wrong. The planning systems being developed for the Pathfinder Planetary Rover to perform these execution monitoring tasks are discussed. This system includes a network of planners to perform path planning, expectation generation, path analysis, sensor and reaction selection, and resource allocation.

Description: This paper documents a powerful set of software tools used for developing situated skills. These situated skills form the reactive level of a three-tiered intelligent agent architecture. The architecture is designed to allow these skills to be manipulated by a task level engine which is monitoring the current situation and selecting skills necessary for the current task. The idea is to coordinate the dynamic activations and deactivations of these situated skills in order to configure the reactive layer for the task at hand. The heart of the skills environment is a data flow mechanism which pipelines the currently active skills for execution. A front end graphical interface serves as a debugging facility during skill development and testing. We are able to integrate skills developed in different languages into the skills environment. The power of the skills environment lies in the amount of time it saves for the programmer to develop code for the reactive layer of a robot.

Description: The Intelligent Systems and Robots Center at Sandia National Laboratories is developing technologies for the automation of processes associated with environmental remediation and information-driven manufacturing. These technologies, which focus on automated planning and programming and sensor-based and model-based control, are used to build intelligent systems which are able to generate plans of action, program the necessary devices, and use sensors to react to changes in the environment. By automating tasks through the use of programmable devices tied to computer models which are augmented by sensing, requirements for faster, safer, and cheaper systems are being satisfied. However, because of the need for rapid cost-effect prototyping and multi-laboratory teaming, it is also necessary to define a consistent approach to the construction of controllers for such systems. As a result, the Generic Intelligent System Controller (GISC) concept has been developed. This concept promotes the philosophy of producing generic tool kits which can be used and reused to build intelligent control systems.

Description: Many people who are mobility impaired are incapable, for a variety of reasons, of using an ordinary wheelchair. These people must rely on either a power wheelchair, which they control, or another person to push and guide them while they are in an ordinary or power wheelchair. Power wheelchairs can be difficult to operate. If a person has additional disabilities, either in perception or fine motor control of their hands, a power chair can be difficult or impossible for them to use safely. Having one person push and guide a person who is mobility impaired is very expensive, and if the disabled person is otherwise independent, very inefficient and frustrating. This paper describes a low-cost robotic addition to a power wheelchair that assists the rider of the chair in avoiding obstacles, going to pre-designated places, and maneuvering through doorways and other narrow or crowded areas. This system can be interfaced to a variety of input devices, and can give the operator as much or as little moment by moment control of the chair as they wish.

Description: The Space Engineering Research Center (SERC) at MIT, started in Jul. 1988, has completed two years of research. The Center is approaching the operational phase of its first testbed, is midway through the construction of a second testbed, and is in the design phase of a third. We presently have seven participating faculty, four participating staff members, ten graduate students, and numerous undergraduates. This report reviews the testbed programs, individual graduate research, other SERC activities not funded by the Center, interaction with non-MIT organizations, and SERC milestones. Published papers made possible by SERC funding are included at the end of the report.

Description: Lidar remote sensing instruments can make a significant contribution to satisfying many of the required measurements of atmospheric and surface parameters for future spaceborne platforms, including topographic altimeters, atmospheric profiles of, wind, humidity, temperature, trace molecules, aerosols, and clouds. It is highly desirable to have wide measurement swaths for rapid coverage rather than just the narrow ribbon of data that is obtained with a nadir only observation. For most applications global coverage is required, and for wind measurements scanning or pointing is required in order to retrieve the full 3-D wind vector from multiple line-of-sight Doppler measurements. Conventional lidar receivers make up a substantial portion of the instrument's size and weight. Wide angle scanning typically requires a large scanning mirror in front of the receiver telescope, or pointing the entire telescope and aft optics assembly, Either of these methods entails the use of large bearings, motors, gearing and their associated electronics. Spaceborne instruments also need reaction wheels to counter the torque applied to the spacecraft by these motions. NASA has developed simplified conical scanning telescopes using Holographic Optical Elements (HOEs) to reduce the size, mass, angular momentum, and cost of scanning lidar systems. NASA has developed two operating lidar systems based on 40 cm diameter HOEs. The first such system, named Prototype Holographic Atmospheric Scanner for Environmental Remote Sensing (PHASERS) was a joint development between NASA Goddard Space Flight Center (GSFC) and the University of Maryland College Park. PHASERS is based on a reflection HOE for use at the doubled Nd:YAG laser wavelength of 532 nm and has recently undergone a number of design changes in a collaborative effort between GSFC and Saint Anselm College in New Hampshire. The next step was to develop IR transmission HOEs for use with the Nd:YAG fundamental in the Holographic Airborne Rotating Lidar Instrument Experiment (HARLIE). The HOE spins like a compact disk in a large ring ball bearing. In an aircraft the HOE faces down, looking out through a window at an angle of 45 degrees off-nadir. The HOE diffracts 85% of the incident 532 nm light into a 160 micron spot at a focal length of 1 meter. HARLIE is a field deployable lidar measuring aerosol, cloud, and boundary layer backscatter for atmospheric research. It has flown several times and is also used from a ground-based trailer in an upward-looking mode. The HOE generates a 45 degree conical scan pattern by rotating at speeds up to 30 rpm. Like PHASERS, the HOE in HARLIE serves both as the laser collimating lens as well as the receiver telescope primary optic. The telescope is coupled to the receiver package via fiber optic. The transmitter is a diode pumped Nd:YAG laser operating at 1064 nm, delivering 1 mJ pulses at a 5 KHz rep-rate. The receiver has a 200 microradian field-of-view and a 0.5 nm optical bandpass. The photon counting data system utilizes a single Geiger-mode silicon avalanche photodiode detector, This new technology has also presented us with new data visualization challenges as well as new measurement techniques. The backscatter data obtained from a stationary (i.e. ground-based) scanning HOE lidar is on the surface of a cone, which when viewed over many consecutive scans can reveal atmospheric motions on this surface over time as the atmosphere advects over the site. In a moving platform such as an airplane or satellite, the data from consecutive scans cover different areas under the flight path, revealing atmospheric structure in 3-dimensions. An example of a visualization of HARLIE ground-based data is presented, showing aerosol backscatter on a 90 degree conical surface generated from one 360 degree scan of the lidar during the HOLO-1 field campaign on the afternoon of 10 March 1999. Higher backscatter levels are rendered as lighter signal against a dark background. Breaking Kelvin-Helmholtz waves are evident on the north side of the scan at an altitude of 10-11 km. Time series of successive scans made at regular intervals render unique views of atmospheric motions, from which vertical profiles of atmospheric wind vectors can be obtained using a unique data analysis approach. Wind vectors obtained from the lidar were compared with co-located radiosonde wind profiles during an intensive operating period in September-October 2000 at the Atmospheric Radiation Measurement Program's Southern Great Plains Central Facility.

Description: Two aerosol backscatter lidar measurement campaigns were conducted using two holographic scanning lidars and one zenith staring lidar for the purposes of reliability testing under field conditions three new lidar systems and to develop new scanning measurement techniques and applications. The first campaign took place near the campus of Utah State University in Logan Utah in March of 1999 and is called HOLO-1. HOLO-2 was conducted in June of 1999 on the campus of Saint Anselm College, near the city of Manchester, New Hampshire. Each campaign covered a period of approximately one week of nearly continuous observation of cloud and aerosol backscatter in the visible and near infrared by lidar, and wide field visible sky images by video camera in the daytime. The scanning capability coupled with a high rep-rate, high average power laser enables both high spatial and high temporal resolution observations that Particularly intriguing is the possibility of deriving atmospheric wind profiles from temporal analysis of aerosol backscatter spatial structure obtained by conical scan without the use of Doppler techniques.

Description: Background noise reduction of War signals is one of the most important factors in achieving better signal to noise ratio and precise atmospheric data from Mar measurements. Fahey Perot etalons have been used in several lidar systems as narrow band pass filters in the reduction of scattered sunlight. An slalom with spectral bandwidth, (Delta)v=0.23/cm, free spectral range, FSR=6.7/cm, and diameter, d=24mm was installed in a fiber coupled box which included a 500 pm bandwidth interference Filter. The slalom box couples the telescope and detector with 200 pm core fibers and 21 mm focal length collimators. The angular magnification is M=48. The etalon box was inserted into the Holographic Airborne Rotating Lidar Instrument Experiment (HARLIE) system and tested during the HARGLO-2 intercomparison campaign conducted in November 2001 at Wallops Island, Virginia. This paper presents the preliminary test results of the slalom and a complete analysis will be presented at the conference.

Description: Due to sensor errors, uncertainty, incomplete knowledge, and a dynamic world, robot plans will not always be executed exactly as planned. This paper describes an implemented robot planning system that enhances the traditional sense-think-act cycle in ways that allow the robot system monitor its behavior and react in emergencies in real-time. A proposal on how robot systems can completely break away from the traditional three-step cycle is also made.