Driving a vehicle to the limits of the tires while maintaining stability is a challenging task. Models can provide a sense of what regions of the state space are stabilizable, but computing such stability limits for a physical vehicle is elusive due to a variety of unknown disturbances and modeling simplifications: it is difficult to predict exactly how hard a controller can push the vehicle until rotation becomes an unrecoverable spin or slide. We propose a method for simultaneously planning trajectories that avoid obstacles and that are minimally sensitive to disturbances, realized by minimizing open-loop velocity uncertainty. This allows the controller to take risks when necessary without being overly constrained by a conservative bound. We present experimental results that demonstrate the improvement of this approach over a baseline.
Robust Stabilization and Collision Avoidance through Minimizing Open-Loop Velocity Uncertainty
2020-10-19
1701754 byte
Conference paper
Electronic Resource
English
ROBUST STABILIZATION AND COLLISION AVOIDANCE THROUGH MINIMIZING OPEN-LOOP VELOCITY UNCERTAINTY
British Library Conference Proceedings | 2020
|