Implementation of Mass-Independent Impedance Control for RFSEA Using a Linkage Arm

Kyeongmin Kim,Young Sam Lee

doi:10.1109/access.2019.2932091

Abstract

When using the powered lower-limb prosthesis, it is important to obtain varying amputee's weight to prevent an excessive knee flexion in the early stance phase of finite state impedance control. This paper proposes a method for implementing an impedance controller that can operate independently of the varying load mass for reaction force sensing elastic actuator (RFSEA) with a linkage arm. The proposed controller estimates the load mass using recursive least square. We apply a Kalman filter and a disturbance observer to improve the accuracy of the measurement and the tracking performance of the force controller, respectively. The results of an experiment applying a lab-developed RFSEA system with a linkage arm validate the feasibility of this approach. It is also expected that the proposed system reduces the size and development costs of the prosthesis due to the advantages of using the RFSEA.

Highlights

Compared to a traditional passive lower-limb prosthesis, recent powered knee prostheses enable the reproduction of a more natural locomotion and provide various functions such as walking, running, sitting, and stair climbing [1], [2]
We propose a method of implementation that allows the change in mass to be estimated using the recursive least squares (RLS) algorithm in reaction force sensing elastic actuator (RFSEA) with a linkage arm and uses the estimated mass as the input to the force and impedance controllers such that the entire system operates independently of the mass
In this paper, we proposed a method applied to a lab-developed RFSEA system that prevents a change in the dynamic behavior of the impedance model set based on the initial load through a mass estimation when the load mass changes

Summary

Introduction

Compared to a traditional passive lower-limb prosthesis, recent powered knee prostheses enable the reproduction of a more natural locomotion and provide various functions such as walking, running, sitting, and stair climbing [1], [2]. Assuming that locomotion is periodic, many powered knee prostheses use an impedance controller [3]–[5]. To use the impedance controller, many of the parameters need to be selected by clinicians. The adaptive control system used in a previously developed magnetorheological prosthetic knee in [6] creates the appropriate parameters by storing the gait cycle and axial force applied to the prosthesis after the amputee has performed the prescribed gait step. The learning system proposed in [7] produces desired joint trajectories corresponding to the amputee’s physical characteristics by tracking the invariant trajectories and estimates controller parameters by

Objectives

Methods

Findings

Conclusion