无人驾驶矿卡局部避障路径规划算法研究  

Research on Local Obstacle Avoidance Path Planning Algorithm for Autonomous Mining Trucks

作　　者：黄帅 王佳 黄佳德 黄鹏 吕亮 HUANG Shuai;WANG Jia;HUANG Jiade;HUANG Peng;LYU Liang(CRRC Zhuzhou Institute Co.,Ltd.,Zhuzhou,Hunan 412001,China)

机构地区：[1]中车株洲电力机车研究所有限公司,湖南株洲412001

出　　处：《控制与信息技术》2024年第6期19-26,共8页CONTROL AND INFORMATION TECHNOLOGY

基　　金：国家重点研发计划项目(2021YFB2501805)。

摘　　要：针对目前无人驾驶矿用卡车(简称“矿卡”)在矿区环境下局部避障路径规划存在的近距离识别、车地通信延时、地面无法实时规划等问题,文章提出一种基于降尺度优化的局部避障路径规划方法。其首先通过引入Frenet坐标系简化障碍物和参考路径信息,并提出了一种障碍物聚类策略以降低可行路径空间的离散性;接着,采用边界转换策略将避障路径规划中的非凸约束转换为凸约束,进而构建了一个以横向偏移量、偏移速度及偏移加速度为变量的二次规划模型;最后,通过求解优化函数,实现了矿卡在遇到障碍物时的自主局部路径规划。仿真与实车测试结果表明,采用所提方法能够直接利用全局路径信息快速生成有效的避障路径,规划时间短于40 ms;与基于A*搜索和动态规划(DP)的局部路径规划方法相比,该方法提升了70%~80%的规划效率,提高了无人驾驶矿卡在矿区环境中的避障能力。Autonomous mining trucks currently face several challenges in local obstacle avoidance path planning within mining environments,including difficulties in close-range identification,delays in vehicle-to-ground communication,and the absence of real-time planning capabilities on the ground.Based on a downscaled optimization strategy,this paper presents a local obstacle avoidance path planning methodology.Following the initial simplification of obstacle and reference path information based on the Frenet coordinate system,an obstacle clustering strategy is introduced to reduce the discreteness of potential path space.The subsequent utilization of a boundary transformation strategy enables the conversion from non-convex constraints in obstacle avoidance path planning into convex constraints,leading to the construction of a quadratic planning model with variables including lateral offset,offset velocity,and offset acceleration.Through solving the optimization function,autonomous local path planning is initiated for mining trucks encountering obstacles.Both simulations and real vehicle experiments confirmed the effectiveness of the proposed method in rapidly generating effective obstacle avoidance paths using global path information,with a planning time of less than 40 milliseconds.Compared to other local path planning methods based on A*search and dynamic programming(DP),this method could improve planning efficiency by 70%-80%,enhancing the obstacle avoidance capability of autonomous mining trucks in mining environments.

关 键 词：无人驾驶矿卡 障碍物聚类 降尺度优化 局部路径规划 避障路径 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]