Synergy-based Knee Angle Estimation on Stairs and Slopes for A Powered Transfemoral Prosthesis

Lifen Mo, Fei Gao, Yiou Sun, Zhipeng Tang, Jing Jie, Zhenhua Song, Wei-Hsin Liao, Fengyan Liang, Ming Yin

Abstract: Above-knee amputees have difficulties in locomotion on stairs and slopes. One of the most challenging issues in controlling powered transfemoral prostheses is the design of adaptive knee trajectories for different users and terrains. The synergy-based trajectory generation approach is becoming a promising approach to solving this issue. It estimates a user-adaptive knee trajectory from the motion of the residual limb in amputees for the prosthesis, based on the synergy. In a previous study, we have validated the feasibility of this approach in level-ground walking. Herein, we aim to evaluate the performance of this approach in scenarios of slope descent/ascent and stair descent/ascent. Results indicate that the sequence-to-sequence (Seq2Seq) model achieved RMSE values of 2.71°, 2.38°, 2.29°, and 2.15° for predicting knee angles during ascent and descent on slopes and stairs, respectively. It surpasses other state-of-the-art methods in overall performance. Our study shows the promise of the synergy-based approach for above-knee prostheses on slopes and stairs by generating adaptive trajectories, eliminating the need for manual adjustments of prosthetic parameters based on terrains or users.