Abstract: Trajectory planning, as a key module of behavioral decision-making and bottom-level control, determines whether the automatic driving command of intelligent vehicles can be successfully executed. In review of the problem of current trajectory planning algorithm that it didn’t consider the actual driving characteristics of the vehicle, such as the maximum angle of the front wheels at different speeds and the condition whether the trajectory of the vehicle was smooth when avoiding obstacles, an improved artificial potential field method was proposed. Kinematic constraints based on two degree of freedom model of vehicle were added, so that the path planning algorithm could plan the path adapted to different vehicle speeds. The vehicle movement space was set as a secondary highway, and the road constraint with lane keeping function was set. In order to make the curve smoother and could guide the vehicle to avoid obstacles, the traditional circular obstacle model was changed to the irregular elliptical obstacle model based on a safe distance model of overtaking. Finally, MATLAB was used to simulate. The comparison experiment shows that the improved scheme can better complete the overtaking path planning of vehicles on structured roads, and the path is smoother when vehicles make obstacle avoidance and lane change behavior.

Key words: vehicle engineering; intelligent vehicle; overtake obstacle avoidance; local path planning; obstacle model; artificial potential field

摘要： 轨迹规划作为行为决策和底层控制的关键模块，决定了智能汽车自动驾驶命令能否顺利执行。当前路径规划算法大多未考虑车辆实际的行驶特性，如未考虑不同车速下的前轮最大转角、车辆避障时轨迹是否平滑等。针对当前的轨迹规划算法存在的问题，设计了一种改进人工势场法，增加了基于车辆二自由度模型的运动学约束，使得路径规划算法能够规划出适应不同车速的路径；将车辆运动空间设定为二级公路，设定了具有车道保持功能的道路约束；为了使得避障时的曲线更加光滑，便于引导车辆进行避障，将传统的圆形障碍物模型改为基于安全距离的不规则椭圆障碍物模型。最后利用MATLAB进行仿真，通过对比实验说明改进的方案能较好的完成车辆在结构化道路中的超车路径规划，车辆做出避障变道行为时的路径更加光滑。

关键词: 车辆工程；智能车辆；超车避障；路径规划；障碍物模型；人工势场法