摘要
针对传统的矿用巡检机器人视觉伺服多是在标定条件下实现,标定参数精度较差,无标定视觉伺服控制系统因其具有更好的适应性、灵活性,越来越得到广泛的认可。基于YOLO-V4算法,采用CSPDarknet53网络结构,对矿用巡检机器人无标定视觉伺服控制系统进行设计实验,并与YOLO-V3,SSD,AlexNet这3种模型进行分析比较。结果表明:基于YOLO-V4的矿用巡检机器人无标定伺服控制技术切实可行,其识别精度较高,实时性良好,满足当前矿业监测的需求,能够进一步提升作业的安全性。
For the traditional mine inspection robot,the visual servo system is mostly implemented under calibration conditions,and the calibration parameters have poor accuracy.The non-calibration visual servo control system is more and more widely recognized because of its better adaptability and flexibility.Based on YOLO-V4 algorithm and using CSPDarknet53 network structure,designs and tests a non-calibrated visual servo control system for mining patrol robot,and compares it with three models:YOLO-V3,SSD and AlexNet.The results show that the non-calibration server control technology of the mining patrol robot based on YOLO-V4 is feasible,its recognition accuracy is high,its real-time performance is good,it meets the current requirements of mining monitoring,and it can further enhance the safety of operation.
作者
任百峰
REN Baifeng(Tangshan Polytechnic College,Tangshan 063299,China)
出处
《煤炭技术》
CAS
北大核心
2022年第10期216-218,共3页
Coal Technology
基金
2018年河北省教育厅青年基金项目(QN2018322)。
