基于YOLO_v3的工件抓取机械臂研究  

Research on Workpiece Gripping Robotic Arm Based on YOLO_v3

作　　者：朱花[1] 陈璐 阳明 刘正超 ZHU Hua;CHEN Lu;YANG Ming;LIU Zhengchao(School of Mechanical and Electrical Engineering,Jiangxi University of Science and Technology,Ganzhou 341000,Jiangxi,China)

机构地区：[1]江西理工大学机电工程学院,江西赣州341000

出　　处：《实验室研究与探索》2023年第9期132-138,共7页Research and Exploration In Laboratory

基　　金：国家自然科学基金项目(52205528)。

摘　　要：为解决目前传统机器视觉在多样本环境下识别、抓取不准确的问题,基于机器视觉设计一套识别、抓取工件机器人实验装置。该装置通过改进YOLO_v3目标检测算法抓取5种不同种类的工件放于指定位置。算法依据数据增强自制数据集样本扩充,提升算法的鲁棒性,为实现模型轻量化,将YOLO_v3原始骨干网络替换为Mobile Net-v3,提升模型检测性能。结合工件的分类和位置信息,以三自由度机械臂为执行系统,通过逆运动学得到角度值;同时通过串口传入单片机控制机械臂末端完成任务。实验表明,基于Mobile Net-v3结构下不同工件的识别准确率达到了85%以上,可用于工业生产和自动化实验的实践教学,具有一定的应用价值。In order to solve the problem that the traditional machine vision is not accurate in the recognition and grasp of various environments,based on machine vision this paper designs a set of robot experiment equipment to recognize and grasp the workpiece.The purpose of the experimental setup is to capture 5 different kinds of workpieces by improving YOLO_v3 object detection algorithm and place them in the specified positions.Firstly,the algorithm expands the sample of the self-made data set based on data enhancement to improve the robustness of the algorithm.On this basis,in order to realize the lightweight of the model,the original YOLO_v3 backbone network is replaced by Mobile Net-v3 network to improve the model detection performance.Then,combined with the classification and position information of the workpiece,a three-degree-of-freedom manipulator is used as the execution system,and the angle value is obtained by inverse kinematics.At the same time,by the serial port,the MCU control manipulator end is controlled to complete the task.The experiment shows that the recognition accuracy of different workpieces reaches more than 85%.It can be used in the practical teaching of industrial production and automation experiment,and has certain application value.

关 键 词：机器视觉 YOLO_v3算法 目标检测 深度学习 

分 类 号：TP241.2[自动化与计算机技术—检测技术与自动化装置]