Predictive Control for Autonomous Driving With Uncertain, Multimodal Predictions

We propose a stochastic model predictive control (SMPC) formulation for path planning with autonomous vehicles in scenarios involving multiple agents with multimodal predictions. The multimodal predictions capture the uncertainty of urban driving in distinct modes/maneuvers (e.g., yield and keep speed) and driving trajectories (e.g., speed and turning radius), which are incorporated for multimodal collision avoidance chance constraints for path planning. In the presence of multimodal uncertainties, it is challenging to reliably compute feasible path planning solutions at real-time frequencies ( ${\geq }10~\mathrm {Hz}$ ). Our main technological contribution is a convex SMPC formulation that simultaneously 1) optimizes over parameterized feedback policies and 2) allocates risk levels for each mode of the prediction. The use of feedback policies and risk allocation enhances the feasibility and performance of the SMPC formulation against multimodal predictions with large uncertainty. We evaluate our approach via simulations and road experiments with a full-scale vehicle interacting in closed loop with virtual vehicles. We consider distinct, multimodal driving scenarios: 1) negotiating a traffic light (TL) and a fast, tailgating agent; 2) executing an unprotected left turn at a traffic intersection; and 3) changing lanes in the presence of multiple agents. For all these scenarios, our approach reliably computes multimodal solutions to the path-planning problem at real-time frequencies.