Multi-hierarchy interaction control of a redundant robot using impedance learning

The control of robots with a compliant joint motion is important for reducing collision forces and improving safety during human robot interactions. In this work, a multi-hierarchy control framework is proposed for the redundant robot to enable the robot end-effector to physically interact with the unknown environment, while providing compliance to the joint space motion. To this end, an impedance learning method is designed to literately update the stiffness and damping parameters of the end-effector with a desired performance. Additionally, based on a null space projection technique, an extra low stiffness impedance controller is designed to provide compliant joint motion behaviour when interaction forces are acted on the robot body. With an adaptive disturbance observer, the proposed controller can achieve satisfactory performance of the end-effector control even under the external disturbances in the joint space. Experimental studies based on a 7 DOF Sawyer robot show that the learning framework is not only to update the target impedance model according to a given cost function, but also to enhance the task performance even when interaction forces are applied on the robot body.