Prediction of Step Length Using Neuro-Fuzzy Approach Suitable for Prosthesis Control

Abstract

Improper knee and ankle impedance during walking with the long steps may cause a loss of balance. In the case of powered prosthetic and assistive devices, a step length (SL) adaptive algorithm can help to maintain the required joint torques while walking with the long SLs. Moreover, it can reduce power consumption during walking with the short SLs. In this article, we propose a novel strategy for the prediction of SL that can help to achieve direct modulation of joint torques to walk comfortably at desired SL. For this purpose, the intent of the user toward walking with the desired SL is recognized with the help of an inertial sensor attached to the thigh of the subject and a neuro-fuzzy algorithm. The proposed method can be implemented in real-time. Along with an accurate prediction of SL, it helps to predict the SL before starting of the stance phase, which makes it suitable for the purpose of lower limb prosthesis control.