Real-time Detection of Travelable Path Based on Image Point Cloud

In the process of moving, the robot first needs to perform ground detection, and the currently feasible area is analyzed to achieve the purpose of safe driving. Using image information and point cloud data, a system for detecting feasible regions by processing three-dimensional information is proposed. The system can quickly identify the approximate travelable area by detecting the edge of the image, and then pass the recognized image information to the corresponding depth image (point cloud date). By performing local normal vector calculation, clustering and segmentation on the depth image, a safe area that the mobile robot can travel is obtained.