Lateral Path Optimization for Autonomous Driving in Highly Constrained Environments

This paper presents a lateral path optimization method to avoid collisions for autonomous driving in highly constrained environments. Firstly, the spatial model in road coordinate is established to describe the kinematic relationship between the vehicle and the reference road. The nonlinear optimization based on vehicle kinematic model in space domain is employed to smooth the reference road. Secondly, the multilayer searched method along the smoothed reference road is used to find the optimal node by using the designed cost function and then established the safety driving corridor in current passable region. Furthermore, the lateral path planning is formulated as the optimization problem based on the spatial model to improve driving safety and comfort. The proposed method considers the constraints of road boundaries and the constraint of curvature in order to provide the optimal path for the collision avoidance maneuver. Finally, the effectiveness of the proposed method is verified through simulation in different highly constrained scenarios.