Abstract:Robot inspection is an effective way to replace manual inspection in mining field. It has the functions of continuous collection, monitoring and analysis of equipment production data. Due to the complex working conditions in the mineral processing plant, issues like poor adaptability and low positioning accuracy in the navigation of the mineral processing robot are affected, which impacts its industrial application significantly. Factoring into the special working conditions of the process plant, a VSLAM visual navigation technology is proposed in this paper. Via binocular camera and depth sensor, the real-time map construction and positioning function can be carried out, so that the robot can complete the inspection work quickly and efficiently, thus enhancing the inspection efficiency and intelligent management level.