Decentralized Nonlinear Control of Redundant Upper Limb Exoskeleton with Natural Adaptation Law

TL;DR

This paper introduces a simplified adaptive decentralized control method for a 7-DoF upper-limb exoskeleton, using a natural adaptation law to reduce tuning complexity and ensure stability, validated through experiments.

Contribution

It proposes a novel adaptive VDC scheme employing a natural adaptation law, eliminating the need for multiple gain tuning and parameter bounds in high-DoF robotic control.

Findings

Reduced adaptation gains to one, simplifying tuning.

Ensured physical consistency of estimated parameters.

Demonstrated excellent experimental performance.

Abstract

The aim of this work is to utilize an adaptive decentralized control method called virtual decomposition control (VDC) to control the orientation and position of the end-effector of a 7 degrees of freedom (DoF) right-hand upper-limb exoskeleton. The prevailing adaptive VDC approach requires tuning of 13n adaptation gains along with 26n upper and lower parameter bounds, where n is the number of rigid bodies. Therefore, utilizing the VDC scheme to control high DoF robots like the 7-DoF upper-limb exoskeleton can be an arduous task. In this paper, a new adaptation function, so-called natural adaptation law (NAL), is employed to eliminate these burdens from VDC, which results in reducing all 13n gains to one and removing dependency on upper and lower bounds. In doing so, VDC-based dynamic equations are restructured, and inertial parameter vectors are made compatible with NAL. Then, the NAL adaptation function is exploited to design a new adaptive VDC scheme. This novel adaptive VDC approach ensures physical consistency conditions for estimated parameters with no need for upper and lower bounds. Finally, the asymptotic stability of the algorithm is proved with the virtual stability concept and the accompanying function. The experimental results are utilized to demonstrate the excellent performance of the proposed new adaptive VDC scheme.