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.
3 THE ARMY ENVIRONMENT FOR ROBOTICS AND ARTIFICIAL INTELLIGENCE The environment for research and development of robotics and artificial intelligence in the Army appears to have become much more robust since the committee's original study. Increased interest and activity in the application of robotics and artificial intelligence (AI) is evident at many levels of the Army. This chapter addresses the management and organiza- tional issues that apply across the specific robotics and artificial intelligence programs. These include inter- and intra-agency cooperation, the need for leadership and a champion, funding of applications, and industrial appli- cations. The most urgent of the Army needs--education and training â is the subject of chapter A. INTER- AND INTRA-AGENCY COORDINATION The committee commends the Army for its heightened awareness of AI and robotics programs. This awareness is reflected in the 1987 establishment of technology base groups for robotics and AI, the 1986 survey of Army AI projects compiled by the Army Research Office, the 1986 summer workshop, the accomplishments of the AI and robotics programs described in the preceding chapter, and enhanced training using AI techniques. The survey, however, tends to overstate the extent of the Army programs. The next workshop should broaden its scope with respect to invitees. Some of the best-known people in the field were not present at the first one, nor were their organizations represented. Nonetheless, the Army has shown a commendable initiative in seeking other sources of funding and assistance, especially in several cooperative efforts with DARPA and in exploitation of work either 15
1C completed or under way at NBS, in universities, and in industry IR&D. The TRADOC definition of AI and robotic requirements shows clear evidence of improvement. The designation of TRADOC "lead agents" will enhance the coordination of battlefield robotic development. Similar improvement is noted in the communication between the user and the materiel developers in the Army Materiel Command. Cooperation has improved markedly between the Army research community and DARPA, and between the Army and the other services. As noted in chapter 2, the Army has joint projects not only with DARPA, but also with the Navy, Department of Energy, and the National Bureau of Standards, and a joint project is being discussed with the Marines. Although sponsored by DARPA, as a ground vehicle the Autonomous Land Vehicle program is of significant interest to the Army. Here government (DARPA, Army), industry (General Electric, SRI-International, Honeywell, Hughes, ADS, Martin Marietta), and academia (Carnegie-Mellon and Colum- bia Universities, Massachusetts Institute of Technology, and the Universities of Southern California, Rochester, Maryland, and Massachusetts) are cooperating in a move toward common robotic research and development goals. Concurrently, it promotes the educational environment and increases the involvement of people in AI and robotics as a discipline. The committee noted an increasing division of the Army's R&D activity into robotics on the one hand and artificial intelligence on the other, with a third area that constitutes the intersection of the two fields. This is a natural and inevitable delineation of activity that can make all related activity more manageable, given top-level oversight and coordination. Communications and networking need to be improved. Such a capability would provide the Army with a critical mass of expertise, geographically distributed. It could offer an intellectual environment that enabled individuals worldwide to work with colleagues in the same field, even when not located at the same laboratory. Networks such as MILNET and ARPANET, and electronic mailboxes, offer the opportunity for Army civilian and military personnel to get hands-on experience in this technology. The establishment of the Robotics and Artificial Intelligence Database is a step in this direction.
17 THE NEED FOR LEADERSHIP AND A CHAMPION The demonstrator program recommended by the committee in its original report would be a necessary first step in the Army's exploitation of the tremendous potential that current technology offers. However, the committee finds â¢that the Army's demonstrator program is funded at a level considerably lower than recommended in its original report. In fact, the current funding of this program appears to fall short of the critical mass of resource concentration required to demonstrate and exploit current technology before it becomes obsolete. Only in FY 1987 did the Army appear to have moved toward centralizing responsibility for coordinating its robotics and AI programs. Two technology base groups--one for robotics, one for AI--were established at the Labora- tory Command. The Army is to be commended for this increase in the visibility and coordination of these programs. The preparation by TRADOC and AMC of an Army master plan for robotics provides an important tool for coordina- tion. It will allow comprehensive management and direc- tion in the application of robotics on the battlefield. Although the plan appears to be developing from the ground up rather than by edict from the top down, this type of creative work is to be encouraged. It should involve the assignment of a capable person to oversee, coordinate, evaluate, plan, manage, and promote Army robotic systems. The demonstrator program needs strong leadership at the departmental level, both to integrate the Army's research and to justify the program to Congress. Vigorous, articu- late leadership of the program at the Department of the Army level is essential to encourage Congress to supply the risk capital required to exploit new technology. The potential benefits of the near-term demonstrator program justify stronger central management and leadership of this program. FUNDING OF APPLICATIONS Engineering development--intention to produce full scale (6.4)--can be a very expensive phase of a program. If 6.4 funding is not carefully planned and given high- level support in advance, there is a danger that research and development (6.1, 6.2, 6.3, and 6.3a) will be reduced to obtain the 6.4 funds. If this happens, then the Army
IS will be "eating its own seed corn" for the future. Investing in projects that integrate the most promising technology-based programs with user needs in the field seems to be the most prudent strategy at this time, provided that requirements documentation is started early and completed in time for early initiation of full-scale engineering development. Another consideration, particularly with regard to large research or leapfrog areas, is that field-use experience will keep the research and development activities moving in the proper direction. However, this, type of field testing must be done carefully. People other than the researchers (Army, Congress, and others) must be given enough background information to understand that the testing and experimentation is for knowledge of field use and not that a proven product is being tested. INDUSTRIAL APPLICATIONS It is interesting to note that the Japanese are currently engaged in a $100 million, 5-year program to develop prototypical robotic systems for service applica- tions (not for factory automation). Construction, fire fighting, security, hazardous material handling, outdoor painting, and institutional cleaning are some of the applications envisaged. Although these are not simple applications, their requirements are considerably simpler than those for military service. Yet the funding level is far greater and more realistic than the Army's pro- jections. It is thus very likely that many commercial Japanese robots will be performing functions in outdoor environments years before a similar robotic system is generally available in the U.S. Army. In all advanced economies, private industry has clearly demonstrated the increased quality and productivity that can result from automation. Unlike tactical battlefield applications, which require enormous R&D expenditures by the Army to meet unique needs and operating conditions, the Army's industrial (internal manufacturing and material handling) applications resemble those of the private sector. Consequently, the time, money, and risk necessary to realize the expected benefits are relatively low. The committee was therefore surprised by the Army's apparently low usage of industrial robots. Several calls to various parts of the Army pointed to the Manufacturing and Methods Technology (MMT) program as the primary source of
19 industrial robotic technology. This program has funded few robotic projects in the past and funds even fewer today. If other programs are not addressing this issue, we recommend increased use of the available technology within Army plants, depots, and arsenals. In addition to industrial robots funded by the Army, robots developed independently by companies could be adapted for Army needs. The legged robots described in the previous chapter are one example; the FMC Corpora- tion's research drawn on by the AGVT program is another. It is worthwhile for the Army to monitor such industrial developments. Although increased use of industrial robots is warranted by the expected productivity improvements alone, the robots are also an important adjunct to the Army's education and training programs. Novices who have received formal AI or robotics training can successfully augment their understanding by implementing these applications--thus broadening the skill base within the Army--as they acquire the additional in-depth knowledge generally required for the more sophisticated tactical applications. Having completed a lower-risk project, this cadre of trained specialists can advance to other projects, leaving a spin-off group of technicians and operators behind, multiplying the training effect. Implementing such a progression plan might include establishing one or more sites to demonstrate the solution $o a generic, frequently occurring Army industrial problem. Serving first as a hands-on training aid, the site would also be a model for other installations throughout the Army. Careful selection of demonstration applications can also include the related technologies (vision, force sensing, etc.) which are the components of the corresponding tactical systems under development.
