基于机器视觉的图书馆机器人障碍识别自动检测系统设计  

Design of Automatic Detection System Based on Machine Vision

作　　者：张馨月 ZHANG Xinyue(Yulin University,Yulin Shaanxi 719000,China)

机构地区：[1]榆林学院,陕西榆林719000

出　　处：《自动化与仪器仪表》2022年第12期205-210,共6页Automation & Instrumentation

基　　金：榆林市产学研合作项目《区域经济与农村发展研究中心建设—基于产学研协同创新的图书馆学科信息服务效能提升研究》(CXY-2020-015-11)。

摘　　要：针对图书馆机器人障碍识别准确率低的问题,提出基于MobileNetV3-YOLOv4算法的障碍识别和D-starLitc+DWA的避障算法,并搭建图书馆机器人障碍识别检测系统。实验结果表明,相较于改进前的YOLOv4算法,提出的MobileNetV3-YOLOv4算法的目标识别准确率提升8.8%,且能够将每秒传输帧提高17PFS;在图书馆障碍仿真环境中,基于D-starLitc+DWA的避障平均距离为40.02m,少于单一的局部避障。实验证明,该检测系统可以明显提升机器人的障碍识别准确性和避障能力,满足图书馆机器人的使用设计需求。According to the problem of low library robot obstacle identification accuracy,the obstacle identification algorithm based on MobileNetV3-YOLOv4 algorithm and D-starLitc+DWA is proposed,and the library robot obstacle identification and detection system is built.The experimental results show that compared with the improved YOLOv4 algorithm,the target recognition accuracy of the proposed MobileNetV3-YOLOv4 algorithm is improved by 8.8%and increased by 17 PFS per second,and the average obstacle avoidance distance based on D-starLitc+DWA is 40.02 m,less than a single local obstacle avoidance.Experiments have proved that the detection system can significantly improve the accuracy of obstacle recognition and obstacle avoidance ability,and meet the use and design needs of library robots.

关 键 词：障碍识别 图书馆机器人 YOLOv4算法 D-starLitc路径规划 DWA避障算法 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]