A robot grasping method considering adjacent interference of sorting targets

Abstract

The application discloses a robot grabbing method considering adjacent interference of sorting targets. The application comprises the following steps: firstly, identifying and extracting the sorting workpieces from the RGB image of the grabbing task, then combining the up-down state of the sorting workpieces and the depth image to determine the final workpiece to be grabbed, the corresponding sorting target area, the potential interference area and the compensation height of the gripper grabbing point; secondly, determining the grabbing point search area and the actual interference area of the gripper; thirdly, setting the search constraint condition to detect the optimal grabbing pose, and then controlling the gripper of the robot arm to realize the grabbing of the workpiece to be grabbed; and finally, continuously collecting images and grabbing the corresponding workpiece to be grabbed in each image until the workpieces in the image are grabbed. The grabbing pose of the sorting workpiece obtained by the application can evade the adjacent interference workpieces in the robot sorting task while considering the easy grabbability of the sorting target, and has good engineering practical value.

Images