基于机器视觉的机器人环境感知技术研究
Research on Robot Environment Perception Technology Based on Machine Vision
黄晓清 1朱荣钊2
作者信息
- 1. 厦门华天涉外职业技术学院 信息技术学院,福建 厦门 361016
- 2. 湖北大学 计算机与信息工程学院,湖北 武汉 430062
- 折叠
摘要
环境感知技术是巡检机器人场景应用的核心技术,对传统单目感知技术存在的环境适应性差的问题,提出了融合雷达感知和视觉感知的双目环境感知技术.通过雷达外参、相机外参、相机内参建立了 5 个坐标系之间的转换关系,实现了雷达传感与视觉传感之间的有效融合.将其应用于矿井用巡检机器人环境感知中,采用 Canny 边缘检测算法检测图像边缘,SVM 实现图像语义分割.通过样机实验验证了雷达传感与视觉传感融合的双目感知技术对环境的感知能力比较强,感知结果明显优于单独的视觉感知和雷达感知.这对巡检机器人在复杂、恶劣环境下的应用提供了技术支撑.
Abstract
Environment perception technology is the core technology of inspection robot scene application.To solve the problem of poor environment adaptability of traditional monocular per-ception technology,binocular environment perception technology combining radar perception and visual perception is proposed.The conversion relationship between five coordinate systems is es-tablished through radar external parameters,camera external parameters and camera internal pa-rameters,and the effective fusion between radar perception and visual perception is realized.It is applied to the environment perception of mine inspection robots.Canny edge detection algorithm is used to detect image edge,and SVM is used to realize image semantic segmentation.Through the prototype experiment,it is verified that the binocular perception technology combining radar perception and visual perception has a strong ability of environment perception,and the percep-tion result obviously surpasses those of separate visual perception and radar perception.This pro-vides technical support for the application of inspection robots in complex and harsh environ-ments.
关键词
机器视觉/环境感知/巡检机器人Key words
machine vision/environment perception/inspection robot引用本文复制引用
基金项目
福建省教育厅课题项目(JAT220777)
出版年
2024
NETL
NSTL
万方数据