Path-Constrained Admittance Control of Human-Robot Interaction for Upper Limb Rehabilitation.

In this paper, the problem of robotic rehabilitation of upper limbs is addressed by focusing attention on the control of a standard collaborative robot for those training activities that can be performed with the aid of an end-effector type system. In particular, a novel admittance control, that constrains the motion of the robot along a prescribed path without imposing a specific time law along it, has been devised. The proposed approach exploits the features of the arc-length parameterization of a generic curve to obtain a simple control formulation able to guide the patient in both a passive or an active way, with the possibility of supporting the execution of the task with an additional force or opposing the motion with a braking force. Being the method independent from the particular curve considered for the constraint specification, it allows an intuitive definition of the task to be performed via Programming by Demonstration. Experimental results show the effectiveness of the proposed approach.