A robotic system for gait rehabilitation: nonlinear modelling and evaluation (original) (raw)

Research on Biomedical Engineering, 2022

Abstract

Purpose The precise control of an active ankle–foot orthosis (AAFO) based on a pneumatic artificial muscle (PAM) is a key aspect in gait rehabilitation applications. This brings the challenge of describing a priori the complex dynamics of the entire robotic system, highly dependent on both, the purely nonlinear behaviour of the PAM and the typical variable external load conditions throughout a gait cycle. The literature shows that this double dependence has been scarcely modelled for this system. To solve this difficulty and based on their predictive capacities, we propose to explore the use of artificial neural network models, such as the nonlinear autoregressive models with exogenous inputs (NLARX) and the Hammerstein-Wiener (HW) structures. Considering this, the objectives of this paper are first, to develop models describing the nonlinear dynamics of this type of gait rehabilitation robotic system under the load conditions considered in a gait cycle, and second, to evaluate the performance of these models. Methods The NLARX and HW structures were used to model the dynamics of an experimental prototype of AAFO driven by a PAM from the recording of input and output data. These models were then validated by comparing the simulated data and the experimental data. Results A comparison of the results of the two models evaluated in the highest demand condition within the gait cycle showed that the HW model describes more precisely the dynamics of the robotic system considered. Conclusions The HW model constitutes a valid alternative for the future design of a model-based control scheme and the evaluation of its performance as a gait-rehabilitation robotic system.

Claudia Lescano hasn't uploaded this paper.

Let Claudia know you want this paper to be uploaded.

Ask for this paper to be uploaded.