Model Predictive Steering Control with Flexible Lane Crossing Considering Nearby Vehicles

Abstract

Various lane-keeping assist systems have been studied. To avoid collisions, automated lane crossing is sometimes required. In this paper, we propose model predictive steering control with flexible lane crossing for obstacle avoidance. In the proposed method, collision avoidance is guaranteed by hard constraints, and inappropriate lane boundary crossing is suppressed by soft constraints. Experiments show that the proposed controller generates the optimal trajectory for autonomously crossing a lane boundary while suppressing lane deviation and preventing collisions.