结构化道路下智能汽车自主换道轨迹规划研究  

Lane-changing Trajectory Planning for Autonomous Vehicles on Structured Roads

作　　者：刘鹏[1,2] 贾寒冰 张雷 王震坡[1,2] LIU Peng;JIA Hanbing;ZHANG Lei;WANG Zhenpo(National Engineering Laboratory for Electric Vehicles,Beijing Institute of Technology,Beijing 100081;Collaborative Innovation Center of Electric Vehicles in Beijing,Beijing 100081)

机构地区：[1]北京理工大学电动车辆国家工程实验室,北京100081 [2]北京电动车辆协同创新中心,北京100081

出　　处：《机械工程学报》2023年第24期271-281,共11页Journal of Mechanical Engineering

基　　金：科技部重点专项资助项目(2017YFB0103600)。

摘　　要：自主换道系统是智能汽车关键技术之一。针对自主换道系统的轨迹规划,提出基于路径-速度分解的分层换道轨迹规划方法。在路径规划层,基于改进二维正态分布函数建立道路、固定障碍物、周围车辆势场并生成环境总势场,采用五次多项式曲线构建换道路径簇,并基于环境总势场确定车辆的最优换道路径;在速度规划层,考虑换道效率、平顺性、安全性、动力学响应、换道时间窗、路径约束等多重影响因素,提出基于凸优化的速度规划方法并对方法有效性进行验证;最后,建立Prescan-Simulink联合仿真环境,在不同场景下对所提出的轨迹规划方法进行仿真验证。结果表明,该方法能够有效处理换道过程中的复杂约束,保证换道过程的安全性、舒适性与换道效率。Automated lane changing plays a crucial role in the advancement of autonomous driving technology.A layered trajectory planning method is present that separates path planning and speed planning into independent processes.The path planning phase involves establishing potential fields for the road,static obstacles,and surrounding vehicles,followed by generating path clusters using the quintic polynomial method.The environmental potential field is determined to derive the optimal lane-changing path.The speed planning process simultaneously considers influencing factors such as lane change efficiency,ride comfort,safety,vehicle dynamics response,time window,and road constraints,and a convex optimization-based method is proposed.To evaluate the proposed scheme,a Prescan-Simulink co-simulation environment is established,and the trajectory planning algorithm is tested under diverse scenarios.The results demonstrate the efficient handling of complex constraints during the lane-changing process using the proposed method,while simultaneously ensuring safety,ride comfort,and lane change efficiency.

关 键 词：路径规划 多项式曲线 环境势场 速度规划 凸优化 

分 类 号：TP216[自动化与计算机技术—检测技术与自动化装置]