Design of the Autonomous Path Planning System for Mining Robots Based on Stereo Vision

Abstract

Rock crushing task for excavators has harsh working environments. With the development of vision sensors and intelligent algorithms, it is possible to achieve autonomous intelligent mining for unmanned excavators. This paper presents an autonomous path planning system for mining robots based on stereo vision. For semantic obstacles such as excavators and loaders in the quarry environment, a recognition network for construction vehicles is trained. The positions and sizes of the construction vehicles are determined by stereo vision. To avoid the rollover hazard that may occur, a gradient map is derived from the elevation map. The traffic cost considering the energy consumption is adopted to build a graph, where the Dijkstra graph search algorithm is used to obtain the path with the lowest cost. Finally, ROS and Gazebo are utilized to simulate and verify the autonomous path planning system. The results show that the proposed system based on stereo vision can complete the autonomous intelligent navigation task in the quarry environment.