Adaptive Path Planning in a Dynamic Environment using a Receding Horizon Probabilistic Roadmap: Experimental Demonstration

We demonstrate the application of a receding-horizon motion-planning method based on probabilistic roadmaps (PRM) that uses sampling from motion primitive maneuver-automata, for the problem of adaptive path planning in the case of a partially unknown obstacle field. Specifically, we consider trajectories followed by a helicopter equipped with finite-range obstacle sensors in unknown or partially unknown terrain. We provide a functional summary of our planning approach and an overview of the algorithmic architecture. The planner functionality has been demonstrated through several softwarein-loop (SIL) scenarios including adaptation to newly discovered obstacles and vehicle deviation from planned paths. This paper presents experimental results from flight tests with an electric helicopter showing in-flight path adaptation to simulated obstacles.