密集障碍物下智能机器人避障路径自主选择研究  

Research on autonomous choice of obstacle avoidance path for intelligent robot under dense obstacles

作　　者：李可 张凌超 米捷[3] LI Ke;ZHANG Lingchao;MI Jie(Engineering Training Center,Henan University of Engineering,Zhengzhou 451191,China;SIPPR Engineering Group Co.,Ltd.,Zhengzhou 450007,China;College of Computer,Henan University of Engineering,Zhengzhou 451191,China)

机构地区：[1]河南工程学院工程训练中心,河南郑州451191 [2]机械工业第六设计研究院有限公司,河南郑州450007 [3]河南工程学院计算机学院,河南郑州451191

出　　处：《河南工程学院学报（自然科学版）》2023年第3期47-52,共6页Journal of Henan University of Engineering：Natural Science Edition

摘　　要：为避免智能机器人与移动障碍物发生碰撞,使其可以高效、自主地完成工作,研究了密集障碍物下智能机器人避障路径自主选择方法。采用激光雷达技术采集周边环境信息,并将采集到的障碍物信息输入BP神经网络,经模型训练后获取智能机器人周围密集障碍物分类结果。基于改进人工势场法,通过改进斥力势场函数,选择智能机器人路径方向;依据障碍物和目标点合力的大小,选取智能机器人避障路径,使智能机器人向局部目标点前进,重复此操作直到全局路径规划完成。实验表明,该方法可实现不同密集程度障碍物下的智能机器人避障路径自主选择,时间碰撞危险度和空间碰撞危险度分别大于期望值12 s和最小距离4.15 m,可有效避免智能机器人与移动障碍物发生碰撞。In order to avoid collisions between intelligent robots and moving obstacles and enable them to efficiently and independently complete their work,a method for autonomous selection of obstacle avoidance paths for intelligent robots under dense obstacles is studied.Firstly,LiDAR technology is used to collect surrounding environmental information,and the collected obstacle information is input into the BP neural network.After model training,the classification results of dense obstacles around the intelligent robot are obtained.Then,based on the improved artificial potential field method,the path direction of the intelligent robot is selected by improving the repulsive potential field function;based on the combined force of obstacles and target points,select the obstacle avoidance path distance of the intelligent robot to move towards the local target point,and repeat this operation until the global path planning is completed.The experiment shows that this method can achieve autonomous obstacle avoidance path selection for intelligent robots under different dense obstacles.The time collision risk and space collision risk are greater than the expected value of 12 s and the minimum distance value of 4.15 m,respectively,which can effectively avoid collisions between intelligent robots and moving obstacles.

关 键 词：密集障碍物 智能机器人 避障路径 人工势场 激光雷达 BP神经网络 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]