Abstract
The segmentation of 3D point clouds for ground plane can generate drivable area for robots' autonomous navigation. And Compared with lasers for generating 3D point clouds, cameras can provide more information and have higher scalability. However, in the process of using the camera to generate 3d point clouds of the ground, the ground lacks texture. Therefore, the ground is often be lacked in the 3D point clouds. A segmentation of 3D point clouds for weak texture ground plane method is proposed in this paper. Firstly, point cloud pretreatment process is designed by using down sampling methods. Secondly, Euclidean-clustering algorithm is used for segment of the point clouds. Thirdly, vertical projection and plane fitting based on RANSAC algorithm are proposed. Fourthly, feasible region refers to the area which the robot can safely pass is segmented. Finally, the method that proposed in this paper is verified using experiments.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call