Pattern Recognition Control of a 2 Degree-of-Freedom Prosthetic Device

For individuals with transradial limb loss or difference, myoelectric prostheses providing advanced functionality are often controlled through two control strategies: Amplitude and Threshold Control (ATC) and Pattern Recognition (PR). While ATC, a conventional method which is sometimes called direct control, is commonly used, it can be less intuitive for multi-degree-of-freedom tasks. PR, leveraging machine learning to classify muscle activation patterns, accesses movements using physiologically appropriate contractions. This study evaluates a Reduced-Contact PR (RC-PR) system, against ATC with rate-sensitive switching. Four individuals with transradial limb loss participated in a randomized crossover trial using a two-degree-of-freedom prosthesis. Functional outcomes, manual dexterity, and user perception were assessed using the Box & Blocks Test, Clothespin Relocation Test, AM-ULA, ACMC, and a UserPerceived Control Questionnaire. RC-PR demonstrated superior performance across all metrics and was preferred by participants for its intuitive operation. These findings suggest that RC-PR offers significant advantages in functional outcomes and user satisfaction, presenting a promising alternative to traditional ATC systems.