MPC-based Lateral Motion Control for Autonomous Vehicles Through Serially Cascaded Discretization Method Considering Path Preview

Autonomous vehicles have gained popularity over the past few years. In this paper, an MPC-based lateral motion controller for autonomous vehicles through serially cascade discretization method is developed. A single discretization method in the entire prediction horizon is usually adopted in MPC. This article seeks to cascade discretization methods of different levels of precision within a single prediction horizon to realize a tradeoff between path tracking accuracy and real-time performance. Leveraging the nature of MPC, a high-fidelity plant model obtained by orthogonal collocation discretization method in the first part of the prediction horizon continuously provides a high quality of control, while the rest of the horizon can be extended by Euler method at a low computational cost. For the specific high-speed driving condition, a prediction horizon expansion strategy considering path preview is combined with MPC to further improve the real-time performance on the premise of maintaining the path tracking performance. The proposed controller is validated in four cases on the MATLAB/Simulink and CarSim co-simulation platform. On average, the tracking error is reduced by 31.2% compared with the controller based on single Euler discretization and the online computing time is 27.3% less than the controller discretized by single orthogonal collocation method. The results show its effectiveness in improving path tracking accuracy and real-time performance. In addition, The viability of the proposed controller in the obstacle avoidance scenario is demonstrated.