This is the authors' abstract. We don't add key points for this paper.
Robot grasping has always been a hot topic in the research field, but there are still a lot of problems to be solved. In this paper, we propose a general grasping method for the unknown objects in cluttered environment from a single-view. Our method is divided into three steps: sampling points selection, grasping pose sampling and grasping pose evaluation. Firstly, we generate a pixel-wise probability image based on the trained sampling affordance network from RGB images, and map the points with high probability to the 3D point cloud combined with the depth map. Then, we obtaine the candidate grasps by sampling on the proposal points, and finally we evaluate the candidate grasps through the trained grasp evaluation network. The network can capture the geometric structure represented by the point cloud information. The proposed method directly uses the original sensor data and can increases the sampling success rate.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Li et al. (2023) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: