基于深度学习的远距离汽车充电口定位研究  

Research on Vehicle Charging Port Location of Long Distance Based on Deep Learning

作　　者：张晓勇 全朋坤 赵凌宇 ZHANG Xiao-yong;QUAN Peng-kun;ZHAO Ling-yu(Beijing Huashang Sanyou New Energy Technology Co.,Ltd.,Beijing 100000 China;Mechanical and Electrical Engineering,Harbin Institute of Technology,Harbin 150000 China)

机构地区：[1]北京华商三优新能源科技有限公司,北京100000 [2]哈尔滨工业大学机电工程学院,黑龙江哈尔滨150000

出　　处：《自动化技术与应用》2023年第3期40-44,共5页Techniques of Automation and Applications

摘　　要：随着电动汽车和自动驾驶技术的发展,电动汽车充电逐渐迈向自动化,充电口的识别定位是实现自动充电口的基础。以单目视觉为基础,提出一种远距离下电动汽车充电口目标识别方法。利用yolov5目标识别算法,建立复杂环境下远距离电动汽车充电口图像数据集,得到充电口的卷积神经网络识别模型,测试不同距离下充电口的识别定位效果,总体识别定位成功率为98.7%。可以更好的实现远距离识别定位的要求。With the development of electric vehicle and automatic driving technology,electric vehicle charging is gradually moving towards automation.The identification and positioning of charging port is the basis of realizing automatic charging port.Based on monocular vision,a target recognition method of electric vehicle charging port in long distance is proposed in this paper.Using yolov5target recognition algorithm,the image data set of long-distance electric vehicle charging port in complex environment is established,the convolution neural network recognition model of charging port is obtained,and the recognition and positioning effect of charging port at different distances is tested.The overall recognition and positioning success rate is 98.7%.long-distance recognition and positioning can be better realized in complex environment.

关 键 词：电动汽车充电口 远距离定位 深度学习 单目视觉 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]