Beyond Gait: Seamless Knee Angle Prediction for Lower Limb Prosthesis in Multiple Scenarios

Knee angle estimation plays a crucial role in the development of lower limb assistive devices, particularly prostheses. Current research in this area primarily focuses on stable gait movements, which limits applicability to real-world scenarios where human motion is far more complex. In this paper, we focus on estimating the knee angle in a broader range of activities beyond simple gait movements. By leveraging the synergy of whole-body dynamics, we propose a transformer-based probabilistic framework, the Angle Estimation Probabilistic Model (AEPM), which offers precise knee angle estimation across various daily movements. AEPM achieves an overall RMSE of 6.83 degrees, with an RMSE of 2.93 degrees in walking scenarios, outperforming the current state of the art with a 24.68% improvement in walking prediction accuracy. Additionally, our method can achieve seamless adaptation between different locomotion modes. Also, this model can be utilized to analyze the synergy between the knee and other joints. We reveal that the whole body movement has valuable information for knee movement, which can provide insights into designing sensors for prostheses.

I. Introduction

Lower limb amputation profoundly affects both the physical capabilities and emotional health of individuals, diminishing their quality of life [1]. In 2005, 1.6 million Americans experienced limb loss, projected to double by 2050 [2]. This issue is not only significant in the United States but also represents a global concern. Most cases of limb loss involve the lower limbs. According to an investigation by Dillingham et al. [3], dysvascular amputations account for 82% of limb loss-related discharges, with 97% of them occurring in the lower limbs, predominantly at the transfemoral or transtibial levels. To address the prevalent issue of lower limb loss, prosthetic devices are preferred among assistive technologies, as they enable individuals to perform daily activities more naturally and significantly enhance the quality of life for those with limb disabilities [4]. Active lower limb prostheses, equipped with a power source, provide greater support for amputees in tasks such as climbing stairs or slopes, and standing up from a seated position, compared to passive or semi-passive prostheses [5]. The improved mobility provided by these active devices depends on advanced prediction and control systems, which can accurately predict knee angles and finely adjust the prosthesis to accommodate the user's various activities.