Objective: We aimed to develop a system for people with amputation that non-invasively restores missing control and sensory information for an ankle-foot prosthesis. Methods: In our approach, a wrist exoskeleton allows people with amputation to control and receive feedback from their prosthetic ankle via teleoperation. We implemented two control schemes: position control with haptic feedback of ankle torque at the wrist; and torque control that allows the user to modify a baseline torque profile by moving their wrist against a virtual spring. We measured tracking error and frequency response for the ankle-foot prosthesis and the wrist exoskeleton. To demonstrate feasibility and evaluate system performance, we conducted an experiment in which one participant with a transtibial amputation tracked desired wrist trajectories during walking, while we measured wrist and ankle response. Results: Benchtop testing demonstrated that for relevant walking frequencies, system error was below human perceptual error. During the walking experiment, the participant was able to voluntarily follow different wrist trajectories with an average RMS error of 1.55 degrees after training. The ankle was also able to track desired trajectories below human perceptual error for both position control (RMSE = 0.8 degrees) and torque control (RMSE = 8.4%). Conclusion: We present a system that allows a user with amputation to control an ankle-foot prosthesis and receive feedback about its state using a wrist exoskeleton, with accuracy comparable to biological neuromotor control. Significance: This bilateral teleoperation system enables novel prosthesis control and feedback strategies that could improve prosthesis control and aid motor learning.

中文翻译：

---

腕足假肢与脚踝假肢的遥操作

目的：我们旨在为截肢患者开发一种系统，该系统可无创地恢复踝足假体缺失的控制和感觉信息。方法：在我们的方法中，腕关节外骨骼允许截肢患者通过远距手术控制并从假肢踝关节接收反馈。我们实施了两种控制方案：在手腕处通过脚踝扭矩的触觉反馈进行位置控制；扭矩控制，允许用户通过将手腕靠在虚拟弹簧上来修改基线扭矩曲线。我们测量了脚踝假体和腕部外骨骼的跟踪误差和频率响应。为了演示可行性并评估系统性能，我们进行了一项实验，其中一名行胫骨截肢的参与者在行走过程中跟踪了所需的手腕轨迹，同时我们测量了手腕和脚踝的反应。结果：台式测试表明，对于相关的步行频率，系统误差低于人的感知误差。在步行实验期间，参与者在训练后能够自愿遵循不同的腕部轨迹，平均RMS误差为1.55度。对于位置控制（RMSE = 0.8度）和扭矩控制（RMSE = 8.4％），脚踝还能够跟踪低于人类感知误差的期望轨迹。结论：我们提出了一种系统，该系统允许截肢的用户使用腕部外骨骼控制踝足假体并接收有关其状态的反馈，其准确性可与生物神经运动控制相媲美。意义：