DASH∞-RRT: Dynamics-aware Safe Motion Planning under Adversarial Disturbances
Nariman Niknejad, Ramin Esmzad, Hamidreza Modares · 2025
This paper presents an adversarial-disturbance-aware motion planning technique that generates collision-free paths using invariant sets. The proposed planner computes a series of invariant sets within which closed-loop trajectories adhere to safety and disturbance attenuation criteria. Each computed performance-aware invariant set is centered around randomly generated waypoints, which form a corridor of connected invariant sets. For each waypoint, an optimization problem identifies the largest invariant region and designs a robust controller to ensure safety. The algorithm, named Dynamics-aware Safe H∞Motion Planning (DASH∞-RRT), integrates the notion of the worst-case adversarial disturbance rejection between waypoints, minimizing the need for frequent re-planning in disturbed environments. The effectiveness of this approach is validated through simulations and real-world trials involving an omnidirectional wheeled robot navigation that is tasked with obstacle avoidance.