Abstract: In recent years,with the rapid development of robot technology and artificial intelligence technology,more and more mining enterprises begin to use robots for inspection and monitoring,in order to improve production efficiency and reduce the cost of manual inspection. However,due to the complex and changeable mine environment,there are a large number of obstacles and dangerous factors,traditional inspection robots often cannot meet the actual demand,it is necessary to introduce some advanced scene understanding and machine vision technology,in order to improve the robot′s navigation and obstacle avoidance ability. In this paper,a navigational obstacle avoidance method of mine inspection robot based on integrated scene understanding and machine vision is studied. First of all,the mine environment is identified and positioned by machine vision technology to obtain the position and surrounding scene information of the current robot. Then,the scene understanding technology is used to analyze the mine scene,including identifying various ores,equipment and personnel,and detecting and classifying obstacles in the mine environment. Finally,the integrated scene understanding and machine vision information are applied to the navigation and obstacle avoidance control of the robot. The experimental results show that the proposed method can effectively improve the robot′s navigation and obstacle avoidance ability in complex mine environment,have certain significance for promoting the construction of intelligent mine,improving mine production efficiency and reducing inspection cost.

Key words: mine inspection robot,intelligent mine,scene understanding,machine vision,navigational obstacle avoidance