Abstract: In this paper a multi-sensor information fusion of LiDAR and industrial camera for obstacle detection is designed for a hexapod wheel-legged mobile robot. Firstly, based on voxel downsampling and Euclidean clustering, the obstacle target recognition of LiDAR point cloud data is realized. Secondly, the internal parameter calibration of mobile robot camera is carried out. For the problem of inaccurate target recognition under complex weather, multi-scale retinex with color restoration algorithm is used to do image enhancement processing on the collected data, and the image-based target detection is realized by using YOLOv5. Then, the fusion detection of two sensors is realized by using Kuhn-Munkres (KM) algorithm, and the effectiveness of the algorithm is verified by simulation, and finally the experimental verification is carried out in the hexapod wheel-legged mobile robot platform. The experimental results show that the scheme can effectively realize the accurate perception of the surrounding environment, and can improve the environmental adaptability and perception ability of robot. The scheme has application potential in the field of mobile robot.