Research on lane-change obstacle avoidance control strategies for in-wheel motor intelligent vehicles

Abstract

In-wheel motor-driven systems offer precise torque control and yaw moment compensation, enabling rapid vehicle state adjustments to facilitate lane-changing and obstacle avoidance maneuvers, thus providing an effective platform for vehicle dynamics control research. To address the prediction accuracy issues of dynamically interacting vehicle trajectories during high-speed lane-changing obstacle avoidance scenarios, as well as challenges including increased trajectory tracking errors and stability control degradation, this study developed a lane-changing obstacle avoidance control strategy based on LSTM interactive vehicle trajectory prediction. The system integrates: a lateral MPC trajectory tracking controller, a longitudinal LQR speed tracking controller, a coordinated sliding mode controller for front wheel angle-drive torque synchronization based on phase plane and stability determination coefficients, and a torque distribution controller based on tire adhesion optimization and multi-constraint optimization. Validation was performed through Prescan/Simulink/Carsim co-simulation under typical operating conditions. The results demonstrate, the trajectory prediction accuracy meets the requirements; the maximum lateral position deviation decreases from 0.29 to 0.11 m, and the peak deviation between the tracked centroid sideslip angle and its desired value reduces from 1° to 0.48°. Compared with the baseline control strategy, these metrics show 62% and 52% reductions respectively, achieving significant improvements in trajectory tracking precision and stability.