Objects Grasping of Robotic Arm with Compliant Grasper Based on Vision

Abstract

To deal with the problem of slow recognition and poor adaptability in grasping for home service robots. In this paper, a robust robotic arm system based on vision is proposed. The system uses an OpenMV machine vision module to recognize and output the measured pose information of the AprilTag, which is attached on the corresponding object. Then a 4-DOF robotic arm with two-fingers compliant grasper is designed to grasp different objects with various shapes and sizes according to the certain trajectory planning from compliant grasper to the target. Moreover, the experiments are performed to validate the whole system. Results show that the system can recognize and locate the target objects quickly and accurately, as well as grasp and drop the objects in the expected position reliably.