Vehicle steering maneuvers with direct trajectory optimization

Abstract

Steering control systems have been used to develop vehicle automated lane change maneuvers or evasive maneuvers for collision avoidance. Most of these systems have used predetermined desired trajectories to perform the required maneuvers. In this study, an optimal trajectory is found while ensuring minimization of lateral acceleration throughout the maneuver. Collocation technique was used to numerically solve the nonlinear programming problem. The results show a near optimal trajectory can be achieved. The generated trajectory is compared to that of a fifth-order polynomial. The resultant trajectory was substantially better than the polynomial one, with both a lower peak and the overall lateral accelerations.