智能电动车弯曲道路场景中的避障路径规划

为研究智能电动车在弯曲道路场景下进行避障规划的有效性, 提出了一种将笛卡尔坐标系转换为曲线坐标系的方法, 利用5次贝塞尔曲线对弯曲道路场景中的车道线进行逼近得到参考路径, 通过对参考路径进行弧长参数化, 以弧长为横坐标, 横向偏移为纵坐标的方法建立曲线坐标系, 根据车辆和子目标点在曲线坐标系中的位置关系, 采用3次多项式实时生成候选路径, 利用序列二次规划算法对候选路径进行优化; 为验证所提算法的有效性, 以某智能电动车为平台, 利用单目相机、64线激光雷达、工控机等设备搭建试验车, 通过Apollo平台对车辆在弯曲道路场景中的避障算法进行在线仿真, 在园区实车试验中对避障算法进行了GPS位置误差和航向角累计误差分析。研究结果表明: 在曲线坐标系中进行车辆弯曲道路场景下的避障路径规划, 能有效地描述规划路径曲率半径、车辆中心位置偏移车道线距离等信息, 容易确定自身车辆的可行驶区域、前方障碍物位置信息, 从而生成最优路径; 在园区场景的避障过程中, GPS位置误差发生在初始点、转弯点以及避障点, 最大误差为0.15 m, 航向角累计误差为12°, 突然增大的弯道位置误差主要由车辆姿态瞬时改变及障碍物匹配过程引起, 但是误差都能够很好地控制在一定范围之内, 利用曲线坐标系解决弯曲道路场景中的避障路径规划是可行的。 

Obstacle avoidance path planning of intelligent electric vehicles in winding road scene

In order to study the reliability of obstacle avoidance planning for intelligent electric vehicles in the winding road scene, a method of converting the Cartesian coordinate system into the curvilinear coordinate system was proposed. The quintic Bézier curve was used to approximate the lane line in the winding road scene to obtain the reference path. Through the arc-length parameterization of reference path, a curvilinear coordinate system was established by using the arc-length as abscissa and the lateral offset as ordinate. According to the position of the vehicle and the sub-target points in the curvilinear coordinate system, the candidate paths were generated by the cubic polynomial in real time, and the candidate paths were optimized by using the sequence quadratic planning method. In order to verify the reliability of the proposed algorithm, an electric vehicle was used as a platform to build a test car with monocular cameras, 64-line laser radar, industrial control computers and other equipments. The online simulation of vehicle obstacle avoidance algorithm in the winding road scene was implemented based on Apollo platform. During the real vehicle experiment in the zone, the GPS position error and heading angle cumulative error of the obstacle avoidance algorithm were analyzed. Research result shows that the obstacle avoidance path planning of vehicle in the winding road scene can effectively describe the information of path curvature radius, the offset distance of vehicle center from the lane line, etc, and it is easy to determine the driving area of the vehicle and the obstacle position information ahead, so as to generate the optimal path. During the obstacle avoidance in the zone, the GPS position error occurs at the initial point, turning point and obstacle avoidance point, the maximum error is 0.15 m, and the heading angle cumulative error is 12°. The sudden increase of curve position error is mainly caused by the instantaneous change of vehicle posture and the matching process of obstacles, but the error can be well controlled within a certain range. So it is feasible to solve the obstacle avoidance path planning in the winding road scene by using the curvilinear coordinate system.