Adaptive Impedance Control for an Upper Limb Robotic Exoskeleton Using Biological Signals

TLDR

The study aims to develop adaptive impedance control for an upper‑limb robotic exoskeleton driven by biological signals. It builds a calibrated musculoskeletal model, uses surface EMG to transfer operator stiffness, and applies an adaptive neural‑network controller with a high‑gain observer to compensate joint deadzone and unknown dynamics for trajectory tracking without velocity measurements. Experimental tests on a real exoskeleton with a human operator demonstrate the approach’s robustness.

Abstract

This paper presents adaptive impedance control of an upper limb robotic exoskeleton using biological signals. First, we develop a reference musculoskeletal model of the human upper limb and experimentally calibrate the model to match the operator's motion behavior. Then, the proposed novel impedance algorithm transfers stiffness from human operator through the surface electromyography (sEMG) signals, being utilized to design the optimal reference impedance model. Considering the unknown deadzone effects in the robot joints and the absence of the precise knowledge of the robot's dynamics, an adaptive neural network control incorporating with a high-gain observer is developed to approximate the deadzone effect and robot's dynamics and drive the robot tracking desired trajectories without velocity measurements. In order to verify the robustness of the proposed approach, the actual implementation has been performed using a real robotic exoskeleton and a human operator.