2024/12/12 by Leonard Jung, Jung, Leonard, Alexander Estornell +3 · 1 citation
Computer Science · #AI-based Problem Solving and Planning #Logic, programming, and type systems #Logic, Reasoning, and Knowledge
paper · pdf · doi:10.48550/arxiv.2412.09777
For safety, autonomous systems must be able to consider sudden changes and\nenact contingency plans appropriately. State-of-the-art methods currently find\ntrajectories that balance between nominal and contingency behavior, or plan for\na singular contingency plan; however, this does not guarantee that the\nresulting plan is safe for all time. To address this research gap, this paper\npresents Contingency-MPPI, a data-driven optimization-based strategy that\nembeds contingency planning inside a nominal planner. By learning to\napproximate the optimal contingency-constrained control sequence with adaptive\nimportance sampling, the proposed method's sampling efficiency is further\nimproved with initializations from a lightweight path planner and trajectory\noptimizer. Finally, we present simulated and hardware experiments demonstrating\nour algorithm generating nominal and contingency plans in real time on a mobile\nrobot.\n