Research Article
A Comparative Study of Sensing Modalities for Classifying Active and Passive Motion
@INPROCEEDINGS{10.1007/978-3-032-38657-1_8, author={Haruki Higa and Momoko Oshiro and Satoshi Shimabukuro and Tomo Akamine and Emi Tamaki}, title={A Comparative Study of Sensing Modalities for Classifying Active and Passive Motion}, proceedings={Body Area Networks. 20th EAI International Conference, BodyNets 2025, Okayama, Japan, December 2--4, 2025, Proceedings}, proceedings_a={BODYNETS}, year={2026}, month={10}, keywords={Proprioception Active and Passive Movement Surface Electromyography Optical Myography Movement Classification Machine Learning}, doi={10.1007/978-3-032-38657-1_8} }- Haruki Higa
Momoko Oshiro
Satoshi Shimabukuro
Tomo Akamine
Emi Tamaki
Year: 2026
A Comparative Study of Sensing Modalities for Classifying Active and Passive Motion
BODYNETS
Springer
DOI: 10.1007/978-3-032-38657-1_8
Abstract
Proprioceptive input and output for computers hold strong potential in gaming and rehabilitation. Active and passive movements differ fundamentally and yield distinct interactions. In this study, fourteen right-handed adults performed elbow flexion and extension under active and passive conditions. Data were recorded from agonist muscles using Surface Electromyography (sEMG) and Optical Myography (OMG). Clips were classified with Random Forest and Support Vector Machine. Across classifiers, sEMG showed consistently higher accuracy than OMG in our sample. Subject-specific (personal) training also performed better than cross-subject (overall) evaluation. These findings indicate that sEMG is preferable for classifying active and passive states from agonist sensing, especially with few channels.