A 3D pickup estimation method for industrial parallel robots based on point cloud simplification and registration

In industrial environments, cameras are heavily influenced by light and perspectives, making traditional vision-based parallel robots struggle with object-sorting tasks. Two-dimensional vision lacks depth information, which limits parallel robots to pick up objects at various heights. This paper proposes a 3D pickup estimation method for parallel robots based on point cloud simplification and registration for multi-objective pickup tasks. Firstly, a point cloud segmentation method based on the spatial domain is proposed to separate incomplete point clouds of object from the robot's workspace. The object dataset is generated by scanning complete point clouds of object using a 3D scanner. Secondly, according to the fast point feature histogram (FPFH) and the weight locally optimal projection (WLOP) algorithms, fusing the FPFH and WLOP (FF-WLOP) method is proposed to simplify the incomplete point cloud and extract more distinctive edge features of objects. The complete point cloud from the dataset is aligned with the simplified incomplete point cloud, and the calculated barycenter's coordinate information is given to the incomplete point cloud. Next, a dynamic weight singular value decomposition (D-SVD) hand-eye calibration method and an optimal projection point strategy are proposed to transform the barycenter coordinates of the object to the optimal pickup coordinates. Experimental results show that the point cloud registration error is 0.38 mm, the pickup rate is 92%, and the robot positioning error is 4.67 mm, which meets the basic pickup requirements.