Probabilistic, Real-Time, Reactive Planner for Aggressive 3D Aircraft Maneuvers

The ultimate goal is to build a Path Planner (PP) that can rapidly plan collision-free routes for a variety of UAVs in highly dynamic environments with large numbers of threats. Our Phase I includes Stanford University''s Professor Jean-Claude Latombe, a pioneer in autonomous path planning, at the forefront of rapid path planning with moving obstacles and complex environments, and creator of the Probabilistic Roadmap approach proposed here and the basis for much of the work in the cited references. In Phase I, the feasibility of our approach will be proven through the design, prototype implementation, testing, and analysis of the 3D Path Planner proposed here. Specifically in Phase I we will gather the requirements for a militarily useful UAV path planner, develop a prototype PP that operates on simulated sensor input and plans aggressive, collision free 3D routes in complex simulated environments in close proximity with moving objects, and optimized to avoid enemy fire while fulfilling a large number of constraints. Further, we will test the implemented PP in simulated scenarios, analyze the expected performance of the PP and compare that analysis with experimental results, and design the Phase II system. As shown in the proposal, our approach can be readily adapted to different types of UAVs and different types of planning constraints.