NRI: Novel Prosthetic Arm Control Based on a Low-Dimensional Internal Musculoskeletal Biomechanical (LIMB) Model

Upper limb amputation is a major cause of disability for nearly 160,000 Americans, many of whom could benefit from emergent sophisticated robotic, multifunctional prosthetic arms/hands. In these advanced prostheses, movements are typically controlled by interpreting the user's electromyographic (EMG) signals from residual or reinnervated muscles. State-of-the-art pattern recognition (PR) has been the most promising EMG control interface for multifunctional artificial arms. However, EMG PR-based control algorithms often require lengthy and frequent algorithm training and lack reliability when the external loading or arm posture changes. This is partly because EMG PR is data-driven and does not account for the behavior of the underlying neural or biomechanical system from which the EMG signals are sourced. The objective of this project is to develop a novel EMG control of multifunctional transradial (TR) prostheses based on a systematic study of neuromuscular control and biomechanical roles of residual muscles in TR amputees. This research can potentially enhance the health, function, and quality of life of upper limb amputees. This project's concept, methods, and frameworks for enhancing EMG-based prosthesis control may be extended to other assistive robotics to benefit other patient populations such as stroke survivors. This project will impact STEM education by promoting project-based cross-training among K-12, undergraduate, and graduate students in underrepresented groups including females, minorities, and students with disabilities. The research may also impact the neuroscience and movement science communities by elucidating the control mechanism of the arm/hand and unveiling new knowledge of neuroplasticity and the internal model in upper limb amputees.

At the core of the multifunctional prosthesis control is a musculoskeletal model of the missing limb that will be used to interpret intended joint motions from EMG signals. The intellectual merit of this project includes a new concept for the control of robotic, multifunctional prosthetic arms/hands. The PIs' musculoskeletal model-based interface is fundamentally different from existing data-driven, EMG PR-based control because it interprets EMG signals and decodes user movement intent in a more biological way. Additionally, this effort will result in new knowledge regarding the neuroplasticity, neuromuscular control, and perceived biomechanical roles of residual muscles in upper limb amputees, which has not been systematically investigated based on the investigators' knowledge. Ultimately, the project will result in a new prosthesis control that, compared to state-of-the-art EMG PR-based control, may require significantly fewer and shorter calibrations, provide more intuitive, robust control (against posture changes, external loading, etc.), and enable multi-joint coordinated prosthesis operations. The investigators expect that the research may transform the way in which upper limb amputees operate multifunctional prostheses in daily life.