摘要
无人驾驶汽车局部路径规划算法应确保避障的安全性、实时性和路径的平滑性等。本文提出了一种基于离散优化的局部路径规划算法,即采用代价函数分别评估离散生成的候选路径的安全性、平滑性等,再根据各代价函数加权计算获得局部最佳路径。针对障碍物移动随机性,设计了一种基于运动估计结合高斯卷积的移动障碍安全性代价函数;考虑候选路径曲率的变化及其连续性,设计了路径平滑性代价函数。使用了一种新的坐标转换计算方法将路径从s-ρ坐标系转换到大地笛卡尔坐标系,提高了实时性。最后,利用PreScan和Matlab软件进行联合仿真,并在"远飞"无人车实验平台上进行了真实道路场景的实车实验。实验结果表明:提出的路径规划算法不仅能使无人车安全、合理地规避静止和移动障碍,且完全满足局部路径规划算法对实时性的要求。
The local path planning algorithm of unmanned vehicle has certain requirements for the safety and real-time performance of obstacle avoidance,and the smoothness of obstacle avoidance path.In this paper,a local path planning algorithm based on discrete optimization is proposed,which uses cost function to evaluate the safety and smoothness of discretely generated candidate paths,and then obtains the local optimal path through the weighted calculation of each cost function.Aiming at the randomness of obstacles movement,a moving obstacles safety cost function is designed based on motion estimation combined with Gaussian convolution.Considering the curvature and its continuity of path,a path smoothness cost function is designed.A new coordinate transformation calculation method is adopted to convert the path from the s-ρcoordinate system to the earth Cartesian coordinate system,enhancing real-time performance.Finally,a PreScan/Matlab co-simulation and a real vehicle experiment on"Yuan Fei"unmanned vehicle experimental platform are both carried out.The results show that the path planning algorithm proposed not only enables the unmanned vehicle to safely and reasonably avoid the static and moving obstacles,but also fully meets the real-time requirements of local path planning algorithm.
作者
彭晓燕
谢浩
黄晶
Peng Xiaoyan;Xie Hao;Huang Jing(College of Mechanical and Vehicle Engineering,Hunan University,Changsha 410082)
出处
《汽车工程》
EI
CSCD
北大核心
2020年第1期1-10,共10页
Automotive Engineering
基金
国家自然科学基金(51575167、51775178)
湖南省自然科学基金(2017JJ2034)资助
关键词
无人驾驶汽车
避障
路径规划
代价函数
实车实验
unmanned vehicle
obstacle avoidance
path planning
cost function
real vehicle experiment
