Model Predictive Control of Autonomous Vehicles Using Sequence Convex Programming

A planner for autonomous vehicles is presented in this paper. The purpose of this algorithm is to plan a collision-avoidance trajectory with cheap computation. We consider model predictive control (MPC) of the vehicle in the presence of obstacles, in which the optimization usually admits nonlinear programming. By introducing sequential programming (SCP), the nominal trajectory dependent by convexification is updated in time, then the non-convex optimization problem is approximately solved. The approach is validated through simulation compared with regular nonlinear MPC formulation solved by numeric solver.