Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
2 ASSESSMENT OF INDIVIDUAL ARMY PROGRAMS In recent years, the Army has used limited resources to develop selected robotics and artificial intelligence (AI) systems that would demonstrate performance in: ⢠improving soldier survivability, ⢠providing force multipliers, ⢠reducing operating and support costs, and ⢠enhancing training and education of personnel. Army presentations to the committee indicated that the most important of the funded programs include: Teleoperated Mobile Anti-Armor Project (TMAP), Field Material Handling Robot (FMR), Soldier/Robot Interface Project (SRIP), Robotic Combat Vehicles, manned and unmanned (RCV), Advanced Ground Vehicle Technology (AGVT), and Hawk Al-based Maintenance Tutor (HAWK-MACH III). In addition, the Army is supporting a small number of basic research projects in AI, done primarily by key university researchers, and a large number of expert system developmental projects, done primarily in Army laboratories and organizations. While it was impossible to assess each of these programs in detail, the committee's consensus was that the projects had been well conceived and, if successfully completed, would significantly advance the introduction of AI and robotics into Army operations. A major concern of the committee, however, was that the planned present and future funding would not enable most of these programs (especially the robotic developments) to be translated into effective field-worthy systems in less than 10 to 15
years. Sufficient financial support could reduce the lead time to 5 years. After hearing in October 1986 that large portions of the robotics programs were unfunded for FY 1987, the committee was pleased to learn that funds have been redirected to these programs. The Army's three primary robotics programs--the Teleoperated Mobile Anti-Armor Project, the Field Material Handling Robot, and Robotic Combat Vehicles--are now funded at levels at or near the amount requested. Those amounts are still modest, how- ever, and more will be needed to develop the engineering prototypes and specifications that necessarily precede manufacturing. The remainder of this chapter addresses each of the major programs, plus a worthwhile program of legged robots that the Army is not yet supporting. A summary of technical issues and a chart of the status of Army programs follows. THE TELEOPERATED MOBILE ANTI-ARMOR PROGRAM Members of the committee observed the concept evalua- tion program test of the Robotic Ranger conducted by the U.S. Army Infantry Board at Ft. Benning on June 11-12, 1986. The Ranger, developed by Grumman for the Missile Command, is the original demonstrator (two are now funded) of the TMAP concept. It is a teleoperated mobile vehicle equipped with weapon and reconnaissance systems. The program is important not only because of the Ranger's potential as an effective anti-tank weapon, but also because the concept could be adapted for many other teleoperated mobile robot applications. Successful extensions of this technology could achieve large gains in soldier survivability at relatively low cost and offer the potential for significant force multiplication. The value of developing and implementing large numbers of low-cost, expendable anti-armor (especially anti-tank) robots cannot be overestimated. The availability of such weapons to counterbalance the persistent numerical tank advantage of the Soviet bloc could have a profound effect on U.S. nuclear strategy. Moreover, the potential vulnerability of tanks to relatively inexpensive anti-tank robots could fundamentally affect Army tactics and weapons requirements for the field. A major deficiency in the present TMAP program is the use of a fiber optic umbilical cord between the operator--
that is, the soldier--and the teleoperated vehicle. The fiber optic cord provides an excellent means for communi- cations. It is wideband, does not require line of sight, does not radiate, and is not subject to interception. However, the cord has the serious problems of limited range and possible entanglement or cutting by the user or the enemy. This communication problem has received much attention in the past with little success. Nonetheless, because of the importance not only to the Army but to all the services, the committee believes that a sustained research and development effort should be undertaken immediately to develop a better solution. Technologies made available in the past few years could be used in effecting secure communications without an umbilical cord. Specifically, the Army should investigate burst (i.e., non-continuous) communications, spread spectrum systems (which code and recapture signals in novel ways across wider bands), and other technologies for short-range secure communications. The broad scope and applicability of such research makes it an ideal candidate for support by the services and, in particular, DARPA. The Tech Base Enhancement for Autonomous Vehicles program, which was too new for the committee to receive a detailed briefing, plans to explore alternatives to to the fiber optic link. In keeping with the strategy of combining short-term demonstrators with planned upgrades, it is appropriate that the Army plans to add a much higher degree of autonomy to the TMAP in the future. One promising opportunity would be to incorporate more advanced sensors for target acquisition, which would partially relieve the system of the need for continuous communication service with a human operator. This program appears to be adequately funded to meet the stated near-term objectives. Full-scale engineering development, originally planned for early calendar year 1988, has--appropriately--been delayed until after a proof of principle demonstration in October 1988 and a 2 to 3 year preproduction engineering project beginning in 1989. Without intensive research to replace the umbilical cord, however, the expanded version of the TMAP is unlikely to be ready for deployment on schedule. ROBOTIC MATERIAL HANDLING EQUIPMENT The Army's projects in robotics material handling are the Field Material Handling Robot (FMR) and the Soldier/
Robot Interface Project (SKIP). Although the total amount of funding by the Army for FMR can be termed modest, at best, the establishment of individual projects on a joint basis with DARPA, plus exploitation of earlier NBS and industry IR&D efforts, presents an opportunity for successful transition to the 6.4 category of funding-- engineering development with intent to produce for field use--by 1990. Soldier/Robot Interface Project A promising development in the Army's robotic material handling program is a very strong lightweight flexible robot arm with a deflection-compensating control system. The Army has gained leverage for its investment by involving Oak Ridge National Laboratories, with its experience in teleoperation in hostile environments, as a major participant. Odetics, Incorporated, is also a participant through the Small Business Innovation Research program. Successful development would make possible many material handling robot operations in the field where low weight and high strength are of paramount importance; these attributes are currently missing from state-of-the- art industrial robots. The Army projects are exploring the many applications in which both small and very large versions of this principle would be most useful. ROBOTIC COMBAT VEHICLES The manned and unmanned Robotic Combat Vehicle (RCV) projects supported by DARPA, the Marines, and the Army are commendable. They are, however, all at a very early stage of development and will not be implemented until much later. For fully autonomous operation of the unmanned vehicle, the technical problems are severe. Automatic target acquisition, driving, and loading remain formidable problems, even for the manned tank with reduced manning. Scheduled funding indicates some opportunities for implementation, especially with the increase in FY 1987. The Army is funding three RCV projects: Advanced Ground Vehicle Technology (AGVT), Tech Base Enhancement for Autonomous Machines (TEAM), and Robotic Command Center (RCC). The committee was briefed primarily on the AGVT; the other two are relatively new.
Advanced Ground Vehicle Technology Committee members visited Martin Marietta in Denver on August 26, 1986. The trip involved both a visit to the Martin Marietta Autonomous Land Vehicle (ALV) laboratory and a DARPA-sponsored demonstration of a teleoperated Advanced Ground Vehicle built by FMC Corporation. The committee was unable to see another version of the Advanced Ground Vehicle Technology (AGVT), built by General Dynamics, that was demonstrated subsequently. The committee was pleased to note the leverage that the AGVT gained by integrating the results of FMC Corpora- tion's IR&D program with DARPA research on the ALV. The demonstration of the teleoperated vehicle was encouraging in that it will help set realistic requirements for robotic systems capable of performing combat and combat support missions. The contrast between the terrain on which the ALV was operating and that on which it will eventually have to operate suggests that the ALV is a long-term solution that will not have useful short-term applications. The ALV was able to assist in developing the obstacle avoidance algorithms while operating on a flat surface with hay bales as obstacles; however, that is very different from the extremely rugged terrain that the ALV would eventually have to traverse to demonstrate its ability to operate autonomously. Despite the Army's strong need for autonomous vehicles, there are major problems ahead which are unlikely to be solved soon. As with most areas of research, the time to commercialize or achieve field use is longer than desired. Although the committee understands that DARPA intends to devote its resources and efforts solely to autonomous mobility, teleoperation still appears to be the answer for the immediate future. That is, people will still be needed to operate the system for some time to come. That being the case, how do we make the operator (and the system) safer and more efficient? Specific needs are: ⢠more concern about counter-countermeasures: non-cable, non-line-of-sight, jam-resistant, highly reliable and secure data links with low probability of intercept; ⢠better stereoscopic or 3D vision to improve the teleoperator's ability to identify ditches, boulders, and other obstacles that could be obscured by vegetation or smoke; and
10 ⢠better and easier map reading systems as aids to operators. HAWK MAINTENANCE TUTOR The development by the Army Research Institute (ARI) of the Hawk Maintenance Tutor is highly commendable. This system uses AI expert system technology for maintenance training. It can satisfy the need for training highly skilled technicians, who are in short supply now and for the foreseeable future. Quality of training and consider- ably reduced training time are impressive outcomes of this program. Further, this system can serve as a model for many other tutoring systems required for even more sophisticated high-technology weapon systems. In addition to the strong applications-based uses of AI, such as for the Hawk Mach-III intelligent maintenance tutor, ARI is making good use of AI and expert systems in association with intelligent computer-aided instruction (ICAI) and is sponsoring a coherent Army program of research at some of the finest universities to help develop the needed training aids. LEGGED MACHINES The Army and DARPA have concentrated their support on the development of mobile vehicles that are either wheeled or tracked. DARPA is supporting several legged vehicle programs at Carnegie-Melion and Ohio State Universities. These programs, however, are still in the research stage, far from use in practical situations. It appears that compact legged (or composite wheeled and legged) vehicles could have significant advantages in rough terrains. About 5 years ago, Odetics Incorporated unveiled a teleoperated "functionoid"--a six-legged vehicle controlled by a person with computer assistance. This clever machine was creatively designed to be strong and versatile. However, it moved slowly and had no sensory feedback to help control it autonomously. Since then, the company has added a stabilizing sensory system to maintain equilibrium and leg sensors to enable the machine to maneuver over rough ground. If a robot with multiple degrees of freedom (i.e., roll, pitch, and yaw, as well as movement along the x, y,
11 and z axes) were added to the multi-legged platform, it could become a teleoperated aid to a soldier. Such a system could carry loads, perform repetitive tasks, be a stable mount for a sensor, rocket launcher, or other weapon, handle hazardous or toxic materials, and perform other tasks. The soldier-operator would control the machine with an appropriate human-engineered interface, either close by or remotely. The machine could have some degree of autonomy in repetitive tasks which the operator would train it to do. Such a system could be light (about 200 pounds) and strong (handle much more than its weight); it could be equipped with a small gasoline-driven electric recharger for a truck-sized battery to enable the system to operate over a good part of a day. We recommend that the Army carefully assess the potential of such a legged, teleoperated "soldier's assistant." It could multiply the effectiveness of the armed forces--or reduce the number of soldiers needed--in the near future. Further, this work is an example of the advantages to the Army of monitoring technical develop- ments in industry and adapting them to Army needs. SUMMARY OF TECHNICAL AREAS The review of selected Army programs and the site visits brought out seven technical areas that the Army should consider including in its robotics and AI programs: 1. Communication links for teleoperation, 2. Stereoscopic vision, 3. Map reading, 4. Sensors for target acquisition, 5. Deflection-compensated robot arms, 6. Stabilizing systems, and 7. Small, self-contained power sources. The Army should carefully review the seven areas to answer these questions: ⢠Do programs in the Army or other service cover each area properly? ⢠Is more research and development needed? ⢠What other programs could use these technologies?
12 1. Communication Links for Teleoperation Two of the Army's three major robotic programs--the Robotic Combat Vehicle and the Teleoperated Mobile Anti- Armor Program--need a broadband, jam-resistant, secure, and reliable data link. In particular, the TMAP's use of a fiber optic cable seriously limits the environments in which it can be used. The shielded cable by itself provides an excellent non-radiating, broad bandwidth (with three-dimensional capability) data link. However, optical cables and wires are not appropriate for field use. They have limited range and can easily be destroyed by either the enemy or the user. Increased research is needed to find durable and secure communication links. 2. Stereoscopic Vision Teleoperation by scope observation and feedback showed the need for better three-dimensional or depth percep- tion. This was confirmed by discussions with operators who have had many hours of training. Because teleopera- tion will be used for a long time, the Army should be working to improve stereoscopic vision. 3. Map Reading At the sites, the soldiers' difficulties in reading were often mentioned as a fundamental problem. Proper map reading for land navigation is basic to teleoperation. Today's technologies should offer many ways to expedite and improve map reading, possibly with operator-assist systems and electronic storage and retrieval using optical disks. The ability to read maps "interactively" could be the technological solution to this continuing problem. A. Sensors for Target Acquisition Robotic sentries will be in positions where sensitive target sensors for object detection, identification, and location could add greatly to the robots' survival and ability to react quickly. The work needed in this area is to apply sensors--primarily imaging sensors--and to improve algorithms for dealing with noisy signals. Good
13 sensors could give the robot a degree of autonomy without requiring constant communications to the teleoperator. They could also permit the robotic sentry to operate in nuclear, biological, or chemical (NBC) contamination or to warn of the presence of NBC contaminants. 5. Deflection-Compensated Robot Arms Deflection compensation systems--including lasers, multi-linkages, and perhaps others--are being developed for robot arms. These devices could make commercial as well as Army-developed robot systems successful in areas where deflections of the robot arms would have caused some failures. People working in the application of robotics should understand these compensating principles. 6. Stabilizing System Teleoperated vehicles will be exposed to terrain that challenges balance. It is important, therefore, to understand stabilizing systems. A stabilizing system has been developed for a multi-legged robot. This suggests that the principle should be studied for possible applications to teleoperated vehicles. 7. Small. Self-Contained Power Source Teleoperated vehicles, especially small systems, have relatively short power life. Small, self-contained power sources could add greatly to the useful field time. A combination of intelligent sensors with teleoperation could give the best timing for recharging.
14 TABLE 2 Status of Selected Army Projects in Robotics and Artificial Intelligence. June 1987 PROGRAM PROJECT PROGRESS SINCE 1983 ORGANIZATIONAL AND TECHNICAL ISSUES Teleoperated Mobile Anti-Armor Program Robotic Ranger Short-term demonstrator tested for battlefield; proof of principle planned October 1988 Optical cables not acceptable in field; need improved ste- reoscopic vision and advanced sensors for target acquisition Robotic Material Handling Field Material Handling Robot Early development; transition to 6.4 by 1990 possible if work with DARPA & use of NBS work succeed Total program funding should be increased Soldier/Robot Interface Project Potential payoff of deflection-compensated flexible robotic arm Robotic Combat Vehicles Advanced Ground Vehicle Technology Demonstrator is func- tional; map reading system & terrain analy- sis under development; full autonomy still long-term possibility; teleoperation only option for near term Not demonstrably prepared for very rugged terrain; need better 3D vision; tenuous communica- tion links are vulnerable to interception Hawk Maintenance Tutor Hawk Mach-III Intelligent Maintenance Tutor Army- sponsored univer- sity research is well planned and promising; strong applications of AI & expert systems
