Transfer of abstract control skills to prosthesis use
AbstractComputer interface tasks have shown that motor learning based control schemes enable multi-grip myoelectric control with only two electrodes. However, it is unclear if this control transfers to prosthesis use. Here, we test if training abstract control with delayed feedback transfers to prosthetic control in a 7-session experiment. Two participants completed five 1-hour training sessions in between a pre- and post-test. The abstract decoding scheme ensured participants had access to five grips (power, tripod, point, lateral, and hand open), and the prosthetic tests included a grip matching task, the modified box and blocks task, and a pick and place test. Both participants increased their grip matching score, reaching a classification accuracy of 93.33% and 98.33%. They also increased the amount of blocks they relocated in the modified box and blocks test, completed the pick and place test faster, lowered the amount of objects they dropped, and increased the accuracy of the grips they selected during the pick and place test. These results show that a motor-based training strategy of abstract control transfers to prosthetic use, enabling five grips with only two electrodes.
How to Cite
S. Dupan, S. Stuttaford, K. Nazarpour, and M. Dyson, “Transfer of abstract control skills to prosthesis use”, MEC Symposium, Aug. 2022.
Myoelectric Control Algorithms