Abstract: Aimed at the problem that automated guided vehicle (AGV) cannot effectively avoid 3D obstacles in a lightless environment, an AGV positioning and navigation technology based on multi-sensor information fusion is proposed. The 3D point cloud and 2D laser data were fused through the point cloud distance screening mechanism, and the multi-sensor fusion positioning and 2D grid map construction were realized based on the graph optimization algorithm. The adaptive obstacle weight was introduced into the A* algorithm, and optimized the path inflection point. The distance function observed by the depth camera was add into the dynamic window approach (DWA) algorithm to realize the dynamic path planning of the AGV. The experimental results show that the positioning accuracy of the system is 3.34cm, and it can effectively avoid obstacles that cannot be observed by 2D lidar, which shortens the path planning time by 37.4% and improves the operation efficiency of AGV.