A Linear Time-varying MPC Method for Vehicle Path-following Assistance Based on Steering Torque

Lane keeping assistance is an important lateral control technology of autonomous driving system. A large number of autonomous steering control methods were given to deal with path-following and lane keeping problems by controlling steering wheel angle directly. In this paper, however, a linear time-varying model predictive control (LTV-MPC) method for path-following manipulating is proposed based on steering torque input. The electric power steering (EPS) system is integrated with vehicle dynamics system in order that the vehicle lateral motion is controlled autonomously through steering motor torque rather than road wheel angle. Local linearization approach is adopted to handle tire nonlinearity and the optimal control problem with constraints is then solved by quadratic programming. Finally, the control scheme is validated in Carsim/Simulink numerical simulation. The maneuver simulation results on different road surface reveals that the proposed path-following control scheme approach is effective. Due to the proposed steering torque control algorithm, path-following is realized by controlling steering torque rather than steering angle, which may be possibly conductive to design haptic feedback steering system for interactive lane keeping control in the future co-driving intelligent vehicle.