(D-130) Advancing Bidirectional Neuroprosthetics with the Open Source Leg

Individuals with lower-limb loss often report diminished balance confidence and an increased risk of falls compared to their able-bodied peers, primarily due to absence of sensory feedback from their missing limb and the lack of intuitive control over their prosthesis [1]. A bidirectional neuroprosthesis would restore sensory feedback from foot-floor interactions and capture user intent for intuitive motor control. Our team aims to mitigate these challenges by developing a bidirectional neuroprosthesis designed to interface with the PNS in the residual limb. While commercially available advanced prostheses come equipped with sensors and motors, their proprietary design does not support interfacing with the nervous system [2]. In contrast, open-source platforms such as the Humotech Open Source Leg (OSL) developed by the University of Michigan offer robust customization and full access to the control system of the device. This work outlines our preliminary steps toward preparing the OSL for use as a bidirectional neuroprosthesis. We successfully demonstrated the ability to assemble, access internal sensor data, and program the embedded OSL controller system. Additionally, we made custom modifications to the OSL for use with two participants, one with transtibial and the other with a transfemoral limb loss.

We utilized the Open Source Leg v2.0 as the prosthesis platform in this project. This self-contained two degree-of-freedom powered prosthesis offers a 120 and 60 degree range-of-motion in the knee and ankle joints, respectively, and features a belt drive transmission ratio of 45.0 across both joints. The OSL is equipped with two inertial measurement units (IMUs), a six-DOF loadcell, dual joint encoders, and actuators from Dephy Inc. Our testing and data acquisition setup employed a Raspberry Pi 4 as a single-board microcomputer which communicated with the sensors and actuators via a digital bus. The functional integrity of the joint subassemblies was ascertained by utilizing pre-developed scripts by Dephy Inc. Within the OSL’s three-tiered control system, we focused on testing the low-level controller responsible for servo motor control. To verify its performance, we implemented an impedance control strategy, enabling the motor to reach a predetermined stiffness at its initial position and a desired damping upon release to a position offset. This control strategy, conceived to actuate one joint at a time, was implemented during benchtop testing while the joint was clamped in a vise. To customize the prosthesis size, we recorded measurements from mid-tibia to foot and mid-patella to foot for the transtibial and transfemoral amputees, respectively. Based on these measurements, we ordered and custom cut titanium pylons (Bulldog Tools, Inc.). We also identified necessary modifications to prosthetic joint housings and electronics hardware to improve the functionality of the prosthesis and facilitate individual operation of the joints during benchtop testing.

The OSL assembly was completed successfully, as confirmed by validation tests conducted using pre-developed controller scripts. The internal tolerance of the joint housings and placement of the magnets for the joint encoders impeded the disassembly and maintenance. To resolve this, we sanded the housing walls and adjusted the magnet placement. We chose to interface the strain gauge amplifier, connected to the six-DOF loadcell, directly to the Raspberry Pi via  I²C instead of through the knee-mounted actuator. This alteration provided a more direct pathway for controller scripts to access the loadcell sensor data. Bench testing demonstrated successful engagement with the low-level controller system by adjusting the stiffness and damping coefficients between 400-600 and 200-300, respectively, at each joint. Furthermore, we were able to stream joint angle from the integrated sensors at a 100 fps framerate and thereby verify the reliability of the connection with the Raspberry Pi. With the successful assembly, internal sensor access, and validation of the low-level control system, we have established the foundation for future development of mid and high-level control systems and integrating the OSL into our neurally integrated bidirectional lower limb prosthesis. Future work will focus on clinical testing of the OSL’s low-level control system with our two participants. Subsequently, we will develop the interface between the OSL and the nervous system, specifically through the implementation of a myoelectric controller and generating proprioceptive neural stimulation based on the OSL’s loading profile and joint movements.

This project was supported in part by the Department of Defense under Awards No. W81XWH-18-1-0321 and W81XWH-20-1-0802. This work was also supported by Wen H. Ko Summer Internship Program sponsored by the Advanced Platform Technology Center at Louis Stokes Cleveland VA Medical Center. In addition, this submission is the result of work supported with resources and the use of facilities at the Louis Stokes Cleveland VA Medical Center.

1. Wong CK, Chen CC, Blackwell WM, Rahal RT, Benoy SA. J Rehabil Med. 2015 Jan;47(1):80-6. 2. Yildiz, K.A., Shin, A.Y. & Kaufman, K.R. J NeuroEngineering Rehabil 17, 43 (2020)