关键词: 自动驾驶；轨迹规划；运动走廊；贝塞尔曲线

Abstract: At present, unmanned driving is developing rapidly, and obstacle avoidance trajectory planning in the urban operating environment is one of the core issues. How to avoid corner collisions that may occur during the trajectory discrete optimization process has gradually become a research hotspot. This article compares the advantages and disadvantages of the coordinate system in the trajectory planning and selects the Cartesian coordinate system. It proposes a self-vehicle model that considers the shape of the vehicle. Based on the vehicle model, it combines a new obstacle boundary processing method to further investigate possible collisions and dangerous areas to simplify environmental information in order to achieve the effect of avoiding collision detection and reducing time-consuming planning. Combined with the positional relationship between the vehicle and the surrounding obstacles, the article proposes the use of motion corridors, combined with the Minimum Snap optimization theory for hard-constrained trajectory optimization, and finally proves that the method is feasible through simulation verification, and can effectively meet the safety and comfort requirements in obstacle avoidance scenarios.

Key words: Automatic driving; Trajectory planning; Motion corridor; Bezier curve