Objective:This study proposed an open-source myoelectric robotic hand orthosis prototype to support people with neuromuscular disorders in performing activities of daily living (ADLs). Methods:A user-tuned methodology was applied, designing the system’s usability and ergonomics with recommendations in literature (2017 to 2020), thus calibrating the prototype by user’s metrics. It has two modules: transmitter–interpreter system (TIS) and receiver–actuator system (RAS) that performs three hand pose: (i) pulp pinch, (ii) cylindrical grip, and (iii) resting hand. The system uses a linear discriminant analysis (LDA) classifier with real-time disruptive windowing (80 ms) and a majority vote algorithm (n=3) as post-processing. TIS classifies and transmits a hand pose within 300 ms to RAS, which executes the hand pose. Results:The LDA classifier was trained with ten datasets of surface myoelectric signals (sMES) recorded from a healthy volunteer. The system’s accuracy reached 90% in real-time tests with three everyday objects. Splints were 3D-printed with resistant material, so the fitting and range of motion were comfortable for the volunteer. Conclusion:The prototype achieved the recommendations in the literature, and it was the orthosis that most fulfilled said requirements compared to state-of-the-art (2017 to 2020) myoelectric robotic hand orthosis. However, it only assists with finger flexion. Significance:The open-source format provides other researchers a starting point to develop an orthosis for in-home rehabilitation that actively assists in ADLs. Additionally, production costs for the orthosis are R$856.35, or 2.98% of the cheapest myoelectric hand orthosis available on the market (PowerGrip), both quoted on November 2021.
Read full abstract