摘要
针对果园环境复杂,障碍物种类繁多,传统动力无法有效进入作业的情况,提出一种基于障碍物分类识别的果园机器人自主避障方法。首先对果园中存在的障碍物进行识别分类,然后针对不同种类的障碍物采用不同的避障方法来完成避障动作。通过搭建基于ROS的避障试验台,对果园机器人上搭载的视觉传感器和激光雷达传感器进行标定,在ROS功能包中植入避障算法,然后进行验证。试验结果表明:在同等条件下,相比传统的避障方法,本文的方法避绕圆形障碍物时用时少1.7 s,所用路程少0.31 m,具有一定的优势;而在避绕不规则障碍物时,本文的方法虽然所用时间和路程比传统方法分别多1.7 s和0.41 m,但机器人能够更加贴近障碍物进行避障,对林下中耕施肥和除草等作业有很大帮助,具有很好的实际应用价值。
Aiming at the situation where the orchard environment is complex,the obstacles are numerous,and traditional power cannot effectively enter the work,an orchard robot autonomous obstacle avoidance method based on obstacle classification and recognition was proposed.First,the obstacles in the orchard were identified and classified,then different obstacle avoidance methods were used to complete obstacle avoidance actions for different types of obstacles.By building a ROS-based obstacle avoidance test bench,the vision sensors and lidar sensors mounted on the orchard robot were calibrated,obstacle avoidance algorithms were embedded in the ROS function package,and then verified.The test results showed that:under the same conditions,compared with the traditional obstacle avoidance method,the method in this paper took 1.7 s to avoid round obstacles and used 0.31 m less distance,which had certain advantages.While avoiding irregular obstacles although the method in this paper took 1.7 s and 0.41 m more time and distance than the traditional method.The robot can avoid obstacles closer to the obstacles,which is very helpful to the cultivation,fertilization and weeding under the forest,it has very good practical application value.
作者
耿乾
毛鹏军
李鹏举
黄传鹏
方骞
张家瑞
Geng Qian;Mao Pengjun;Li Pengju;Huang Chuanpeng;Fang Qian;Zhang Jiarui(College of Mechanical and Electrical Engineering,Henan University of Science and Technology,Luoyang,471003,China;College of Agricultural Engineering,Henan University of Science and Technology,Luoyang,471003,China;Gansu Institute of Mechanical Science Co.,Ltd.,Lanzhou,730000,China)
出处
《中国农机化学报》
北大核心
2020年第8期170-177,共8页
Journal of Chinese Agricultural Mechanization
基金
河南省重大科技专项(181100110100)
河南省高校科技创新团队支持计划(19IRTSTHN021)。
关键词
果园
卷积神经网络
视觉传感器
激光雷达传感器
避障
orchard
convolutional neural network
vision sensor
lidar sensor
obstacle avoidance
