Safe Planning and Control Under Uncertainty for Self-Driving

Motion planning under uncertainty is critical for safe self-driving. This paper proposes a unified obstacle avoidance framework that deals with 1) in ego-vehicle motion; and 2) prediction of dynamic obstacles from the environment. A two-stage traffic participant trajectory predictor comprising short-term long-term used layer to generate but not over-conservative trajectories ego-vehicle. The module cooperates well existing approaches. Our work showcases its effectiveness Frenet frame planner. robust controller using tube MPC guarantees execution presence state noise model uncertainty. Gaussian process regression on-line identification uncertainty's bound. We demonstrate effectiveness, safety, real-time performance our CARLA simulator.