Model-Predictive Optimization for Lane Keeping Assistance System with Exponential Decay Smoothing

Lane keeping assistance system is a widely used driving assistance system. When the vehicle enters and exits a curve, the responses related to lateral stability will show significant fluctuations. Smooth lane keeping is the cornerstone of the user's trust towards the lane keeping assistance, and it is important to smooth the responses related to lateral stability. Therefore, in this paper, a lane keeping control strategy that can achieve smooth steering operation is proposed. First of all, the lateral dynamics model that can be used for lane keeping control conveniently is established. In addition, with the model predictive control (MPC), the control strategy for lane keeping is designed. The main control objective is lateral stability, and performance variables are used to measure the performance during lane keeping process. What's more, the contribution of this paper is that a reference trajectory in the form of an exponential decay function is set for the performance variables, so that the performance variables approach the optimal value in a smooth manner. Finally, two simulation scenarios for lane keeping are set up, and the control strategy in this paper is verified by simulation experiments.