Novel Gripper-like Exoskeleton Design for Robotic Grasping based on Learning from Demonstration

Abstract

Learning from demonstration (LfD) has been developed and proved to be a promising method for transferring skill knowledge from human to robot. It is desired to have a demonstration device that can effectively map demonstrations to the robot’s motion to compensate for the correspondence problem in LfD. Thus, we presented a novel design of a gripper-like exoskeleton for robotic grasping based on Learning from Demonstration. The exoskeleton collected the displacement of its grippers, position, and posture information in the demonstration. This paper first presented the mechatronic design of the exoskeleton and then described the experiment for data validation. We illustrated the preliminary functionality of the exoskeleton by reproducing the demonstration trajectory on the Franka Emika robot.