Stand-Up, Squat, Lunge, and Walk With a Robotic Knee and Ankle Prosthesis Under Shared Neural Control.

Grace Hunt,Tommaso Lenzi,Sarah Hood

doi:10.1109/ojemb.2021.3104261

Grace Hunt, Tommaso Lenzi + Show 1 more

Open Access

https://doi.org/10.1109/ojemb.2021.3104261

Copy DOI

Abstract

Emerging robotic knee and ankle prostheses present an opportunity to restore the biomechanical function of missing biological legs, which is not possible with conventional passive prostheses. However, challenges in coordinating the robotic prosthesis movements with the user's neuromuscular system and transitioning between activities limit the real-world viability of these devices. Here we show that a shared neural control approach combining neural signals from the user's residual limb with robot control improves functional mobility in individuals with above-knee amputation. The proposed shared neural controller enables subjects to stand up and sit down under a variety of conditions, squat, lunge, walk, and seamlessly transition between activities without explicit classification of the intended movement. No other available technology can enable individuals with above-knee amputations to achieve this level of mobility. Further, we show that compared to using a conventional passive prosthesis, the proposed shared neural controller significantly reduced muscle effort in both the intact limb (21–51% decrease) and the residual limb (38–48% decrease). We also found that the body weight lifted by the prosthesis side increased significantly while standing up with the robotic leg prosthesis (49%–68% increase), leading to better loading symmetry (43–46% of body weight on the prosthesis side). By decreasing muscle effort and improving symmetry, the proposed shared neural controller has the potential to improve amputee mobility and decrease the risk of falls compared to using conventional passive prostheses.

Highlights

Above-knee amputation disrupts the natural coordination of biological legs, limiting the mobility of individuals with amputations [1]
After above-knee amputation, the knee and ankle joints are replaced by passive prosthetic joints that cannot perform the biomechanical functions of the missing biological leg joints [2]
The proposed shared neural controller was implemented in a robotic knee and ankle prosthesis (Fig. 1.a, Supplementary Fig. 1) and tested by two individuals with above-knee amputations, who performed sit-to-stand, stand-to-sit, squats, lunges, levelground walking, and transitions between activities

Summary

Introduction

Above-knee amputation disrupts the natural coordination of biological legs, limiting the mobility of individuals with amputations [1]. Controllers based on continuous use of EMG signals have been proposed as alternatives to pre-tuned controllers based on mechanical sensors [26] with the goal of improving adaptability to real world variability These controllers require subject- and session-specific training of the machine learning algorithm used to translate the EMG signals into effective commands for the prosthesis [27], [28]. Rather than explicitly classifying the user’s intended ambulation activity and enforcing a preplanned prostheses action, we provide users with continuous volitional control of a robotic knee and ankle prosthesis using EMG. The proposed shared neural controller enabled two individuals with above-knee amputations to stand up, sit down, squat, lunge, walk, and seamlessly transition between activities without explicitly classifying the ambulation mode intended by the user or enforcing pre-planned prostheses actions. A prosthesis controller with this functionality has the potential to improve the mobility of individuals with above-knee amputation in the real world

Objectives

Methods

Results

Discussion

Conclusion