Controller for a knee-ankle prosthesis that can perform multiple actions.

No powered knee-ankle prosthesis allows for all five everyday tasks of sitting, standing, lunging, squatting, and transitions to walking via one controller. In current systems, each activity requires a dedicated controller to adapt to variations due to user preference or the variability of the environment. Even with the best machine learning algorithms, the action classification accuracy decreases over time, and retraining the machine learning algorithms can be difficult and unsafe for the user to perform at home.

University of Utah researchers have developed a shared neural controller in which a single EMG electrode on the residual hamstring muscle of an above-knee amputee is used to directly control the stand-up, sit-down, squat, lunge, and transition to walking actions of a powered knee prosthesis and indirectly the powered ankle. In this way, the device puts the user in control of all prosthesis actions instead of using machine learning classification algorithms to detect the user’s intention. This leads to more natural locomotion and increased user safety.

Pre-Clinical Validation

• Single controller.
• Enables more natural locomotion.
• Safer for the user.
• Offers direct user control of standing, squatting, lunging, transitions to walking, and sitting-down.

Huy Tran
(801) 581-7792
huy.tran@utah.edu