Study on robustness and feasibility of MPC based vehicular Adaptive Cruise Control system

Abstract

Model Predictive Control (MPC) framework is becoming more and more attractive in designing vehicular Adaptive Cruise Control (ACC) system. However, benefiting from its advantage, e.g. close-loop optimality, MPC algorithm must overcome some of its practical problems, among which low robustness to model mismatch and computing infeasibility of control law are most critical. Aiming at MPC based vehicular ACC algorithm, this paper studies its robustness and computing feasibility issues for the purpose of applying the algorithm into vehicle products. In order to enhance its robustness to model mismatch, feedback correction method is adopted to compensate the predictive error of vehicular following model and improve its predictive precision on the system state. Constraint management method is employed to revise the cost function and soften the I/O constraints of predictive optimization problem, avoiding the computing infeasibility of control law caused by larger tracking errors. Series of simulations with a commercial truck model indicate that the adopted methods can effectively solve the low robustness and computing infeasibility problems of vehicular MO-ACC algorithm, laying a foundation for its implementation on vehicle products.