An Autonomous Platform Simulator (APS)

The development of an intelligent autonomous vehicle, that can perform high risk missions or operate in environments too hazardous for humans, has been a long standing quest of the military community. The Autonomous Platform Simulator APS uses the flexibility and power of realistic graphical simulation to provide a low cost testbed for the study of real-time path planning algorithms and control strategies without the commitment of resources involved in building a prototype system. It is a bridge between the theoretical study of an abstract AI path planning problem and applied research, producing concrete performance measurements under realistic conditions. APS consists of one or more vehicle simulators, each implemented on a Silicon Graphics IRIS4D- 70GT graphics workstation. One vehicle simulator is linked with an AI agent path planner, implemented on a pair of Symbolics AI workstations using the Automated Reasoning Tool development shell. System trails demonstrated that APS was able to achieve real-time path planning and guidance of a realistically depicted ground vehicle navigating using digitized data of actual terrain. Communication bottlenecks currently limit the ability to make direct comparisons between human and machine control, but the system holds promise to fill the gap as a pre- prototype autonomous platforms simulator. Keywords Visual simulators, Real-time graphics, Distributed processing, Line-of-sight. Theses. aw