Maneuver-based trajectory planning with fast collision check for autonomous vehicles in a low-speed road environment

For autonomous vehicles driving in a low-speed road environment like complex campus, there are several peculiarities such as two-way lanes, break-in obstacles, smaller safety margin, etc. In this paper, a novel maneuver-based trajectory planning scheme is designed for autonomous vehicles driving in a low-speed environment with uncertain dynamic obstacles. First, a quintic polynomial is adopted to generate candidate trajectories and we develop a fast collision check algorithm to guarantee safety. In contrast to the existing methods, our collision check can reduce computation time with less conservatism. To balance the conflict between overtaking and lane-keeping, varying maneuvers are proposed to determine different weights in cost function for selecting the optimal trajectory. Simulations with uncertain low-speed obstacles are conducted to evaluate the performance of our proposed method.