基于深度学习及嵌入式视觉系统的球童机器人  

Ball-picking robot based on deep learning and embedded vision system

作　　者：黄林琪 卢颖仪 黎盛业 关兆基 郜志航 李德荣 HUANG Linqi;LU Yingyi;LI Shengye;GUAN Zhaoji;GAO Zhihang;LI Derong

机构地区：[1]广东海洋大学机械与动力工程学院,广东湛江524088 [2]广东海洋大学电子与信息工程学院,广东湛江524088

出　　处：《现代机械》2021年第5期15-19,共5页Modern Machinery

基　　金：广东省大学生科技创新培育专项(编号:pdjh2020 b0274);广东海洋大学大学生创新创业训练计划项目(编号:CXXL2020202);广东海洋大学本科生创新创业团队项目(编号:CCTD201810)。

摘　　要：为解决球场运动后捡球工作量大的问题,设计了一款自动捡球的球童机器人。在对其主要结构进行应力分析后,符合安全要求的基础上,该机器人通过嵌入式控制,首先利用深度学习YOLO-v3算法对球类进行识别,继而利用激光雷达配合UWB定位实时监测周围环境,实现自动避障,最后通过驱动四爪机械手结构来捡球。此款机器人集识别、捡球、储球等功能于一体,实现了高度自动化、智能化,其捡球效率高,能有效减少人工成本,具有广阔的应用前景。In order to solve the problem of heavy workload of picking up balls in ball games,an automatic ball-picking robot is designed.The stress of the main structure is analyzed to meet the safety requirements,and an embedded system is adopted for the control of the robot.Firstly,the ball is identified by YOLO-v3 algorithm.Then the laser radar and the UWB positioning are used to monitor the surrounding environment in real time,so that automatic obstacle avoidance is realized.Finally,the robot drives a four-claw manipulator to pick up the ball.The robot integrates the functions of recognizing,picking up and storing balls,and is highly automatic and intelligent.It has high efficiency in picking up balls and can effectively reduce the labor cost,which has good application prospect.

关 键 词：球童机器人 深度学习 定位避障 图像识别 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]