Online Finger Control Using High-Density EMG and Minimal Training Data for Robotic Applications

A hand impairment can have a profound impact on the quality of life. This has motivated the development of dexterous prosthetic and orthotic devices. However, their control with neuromuscular interfacing remains challenging. Moreover, existing myocontrol interfaces typically require an extensive calibration. We propose a minimally supervised, online myocontrol system for proportional and simultaneous finger force estimation based on ridge regression using only individual finger tasks for training. We compare the performance of this system when using two feature sets extracted from high-density electromyography (EMG) recordings: EMG linear envelope (ENV) and non-linear EMG to muscle activation mapping (ACT). Eight intact-limb participants were tested using online target reaching tasks. On average, the subjects hit 85%+/- 9% and 91%+/- 11% of single finger targets with ENV and ACT features, respectively. The hit rate for combined finger targets decreased to 29%+/- 16% (ENV) and 53%+/- 23% (ACT). The nonlinear transformation (ACT) therefore improved the performance, leading to higher completion rate and more stable control, especially for the non-trainedmovement classes (better generalization). These results demonstrate the feasibility of proportional multiple finger control in intact subjects by regression on non-linear EMG features with a minimal training set of single finger tasks.