A novel energy-motion model for continuous sEMG decoding: from muscle energy to motor pattern

2020-06-11T16:35:03Z (GMT) by Gang Liu Lu Wang Jing Wang

Background: At present, the gesture recognition using sEMG signals requires vast amounts of training data or limits to a few hand movements. This paper presents a novel dynamic energy model that can decode continuous hand actions with force information, by training small amounts of sEMG data.

Method: As activating the forearm muscles, the corresponding fingers are moving or tend to move (namely exerting force). The moving fingers store kinetic energy, and the fingers with moving trends store potential energy. The kinetic and potential energy of fingers is dynamically allocated due to the adaptive-coupling mechanism of five-fingers in actual motion. At this certain moment, the sum of the two energies is constant. We regarded energy mode with the same direction of acceleration of each finger, but likely different movements, as the same one, and divided hand movements into ten energy modes. Independent component analysis and machine learning methods were used to model associations between sEMG signals and energy mode, to determine the hand action, including speed and force adaptively. This theory imitates the self-adapting mechanism in the actual task; thus, ten healthy subjects were recruited, and three experiments mimicking activities of daily living were designed to evaluate the interface: (1) decoding untrained configurations, (2) decoding the amount of single-finger energy, and (3) real-time control.

Results:(1) Participants completed the untrained hand movements (100 /100, p < 0.0001). (2) The test of pricking balloon with a needle tip was designed with significantly better than chance (779 /1000, p < 0.0001).(3) The test of punching a hole in the plasticine on the balloon was with over 95% success rate (97.67±5.04 %, p <0.01).

Conclusion: The model can achieve continuous hand actions with force information, by training small amounts of sEMG data, which reduces trained complexity.