Networked Model Predictive Vehicle Race Control

This paper proposes a Networked Model Predictive Control (NMPC) approach for vehicle racing scenarios with multiple participants. The control algorithm incorporates a trajectory optimization problem for each vehicle, which is solved in a receding horizon way. Sequential Quadratic Programming (SQP) is used for optimization. The individual vehicle controllers are coupled via collision avoidance constraints or a blocking objective term, depending on the race situation, leading to a Networked Control System (NCS). We demonstrate our method in simulations. It produces realistic racing situations, showing overtaking and blocking maneuvers while avoiding collisions. With optimization times in the order of a few milliseconds, real-time applications seem possible.