Fast terminal conditions-free model predictive control and its application to intelligent autonomous vehicles

In this paper, fast terminal condition-free model predictive control (Fast TCF-MPC) is investigated for a constrained linear discrete-time system with a series of incremental coefficients. The terminal penalty term and terminal constraints in classical finite prediction horizon MPC are removed to simplify application processes. The series of incremental coefficients is added to a cost function to gradually boost penalty strength within the prediction horizon. The optimal series of increasing coefficients is determined based on the maximum ratio of the incremental coefficients at adjacent update moments. The prediction horizon is related to both the recursive feasibility of an optimization problem and the uniform exponential stability of the constrained discrete-time system. Numerical examples and experiments with autonomous vehicles demonstrate the effectiveness of Fast TCF-MPC.