Torque-Vectoring Control of Autonomous Vehicles Considering Optimization of Vehicle Handling Characteristics

Distributed drive electric vehicles can apply the four-wheel differential drive to change the vehicle handling performance, which can make the connected and automated vehicles (CAV) more controllable. This article proposes a hierarchical scheme of the torque-vectoring controller (TVC), whose key parameters affecting the control objective are optimized from the human-vehicle closed-loop simulation test. First, the radial basis function (RBF)-based adaptive second-order sliding mode control (RASOSMC) for additional yaw moment generation is designed in the upper layer of the controller. The lower layer is the torque distribution strategy that takes into consideration the minimization of the tire load and the control error of the additional yaw moment and yaw rate. Afterward, the longitudinal and lateral driver model with the adaptive correction of preview time is established. A human-vehicle closed-loop test scheme for extracting the relationship between handling characteristics and handling quality is proposed. On that basis, the optimal value of the two key parameters ω_n and ζ, which influence the vehicle handling, can be quickly determined by a multivariate response surface method. In the simulation, the standard deviation of the tracking error for the desired yaw rate can be reduced by at least 49.1% compared with the first-order sliding mode control (FOSMC) and proportional integral derivative (PID).