Emergency collision avoidance by steering in critical situations

Abstract

In this paper, an emergency collision avoidance system is proposed by including not only braking but also steering control actions. The minimum distance to avoid collision is calculated separately for braking and steering based on the relative motion to the surrounding vehicles and the lane information obtained through the vision sensor. For steering avoidance control, an optimal control input is calculated through the model predictive control that satisfies constraints such as safe avoidance region created by surrounding vehicles and capacity of the vehicle actuator. In particular, for avoiding collision by lane changing, the maximum lateral acceleration and the maximum angle of the trajectory are considered. In addition, the abrupt lateral movement in avoidance causes nonlinear characteristics in tires and, thus, tire parameters are estimated through EKF (Extended Kalman Filter) to improve model prediction accuracy. The control intervention time of avoidance maneuvering is determined for braking and steering, respectively. The simulation results demonstrate that the proposed algorithm of integrating AEB (Autonomous Emergency Braking) and AES (Autonomous Emergency Steering) can effectively avoid the collision in critical situations and that the host vehicle can still maintain the safety inside the road boundary.