基于机器视觉的四足机器人目标识别与跟随系统设计  

Design of target recognition and following system for quadruped robot based on machine vision

作　　者：陈瑾龙 徐哲壮[1] 黄平 江灏[1] 吴读桑 周泓宇 CHEN Jinlong;XU Zhezhuang;HUANG Ping;JIANG Hao;WU Dousang;ZHOU Hongyu(College of Electrical Engineering and Automation,Fuzhou University,Fuzhou 350108,China)

机构地区：[1]福州大学电气工程与自动化学院,福建福州350108

出　　处：《实验技术与管理》2022年第10期135-139,167,共6页Experimental Technology and Management

基　　金：福州大学本科教育教学改革研究项目(20213620);国家级大学生创新创业训练计划项目(202110386001)。

摘　　要：根据四足机器人竞赛项目要求,设计了一种基于机器视觉的四足机器人目标识别与跟随系统。该系统首先运用不同骨干网络的YOLOv3算法对目标图像数据集进行特征学习,生成目标识别模型,进而将模型部署至四足机器人的视觉感知主机,完成目标识别功能。再基于机器视觉获得的目标坐标信息,使用PID控制算法计算得到四足机器人的行进速度,实现对目标的实时跟随。最终测试结果表明,四足机器人能够成功完成竞赛项目要求的目标识别与跟随任务,具有较高的精度与实时性。According to the requirements of quadruped robot competition,a quadruped robot target recognition and tracking system based on machine vision is designed.Firstly,the system uses the YOLOv3 algorithm with different backbone networks to learn the characteristics of the target image dataset and generate the target recognition model,and then deploy the model to the visual perception host of the quadruped robot to complete the target recognition.Based on the target coordinate information obtained by machine vision,the travel speed of the quadruped robot is calculated by PID control algorithm to realize real-time tracking of the target.The final test results show that the quadruped robot can successfully complete the target recognition and tracking tasks required by the competition project,with high accuracy and real-time performance.

关 键 词：四足机器人 机器视觉 目标识别 目标跟随 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置] TP273.5[自动化与计算机技术—控制科学与工程]