Simplifying Kinematic Parameter Estimation in sEMG Prosthetic Hands: A Two-Point Approach
Liu, Gang, Wang, Zhenxiang, He, Ziyang, Guo, Shanshan, Zhang, Rui, Yao, Dezhong
–arXiv.org Artificial Intelligence
Regression-based sEMG prosthetic hands are widely used for their ability to provide continuous kinematic parameters. However, establishing these models traditionally requires complex kinematic sensor systems to collect corresponding kinematic data in synchronization with EMG, which is cumbersome and user-unfriendly. This paper presents a simplified approach utilizing only two data points to depict kinematic parameters. Finger flexion is recorded as 1, extension as -1, and a near-linear model is employed to interpolate intermediate values, offering a viable alternative for kinematic data. We validated the approach with twenty participants through offline analysis and online experiments. The offline analysis confirmed the model's capability to fill in intermediate points and the online experiments demonstrated that participants could control gestures, adjust force accurately. This study significantly reduces the complexity of collecting dynamic parameters in EMG-based regression prosthetics, thus enhancing usability for prosthetic hands.
arXiv.org Artificial Intelligence
May-1-2024
- Country:
- Asia > China > Henan Province (0.15)
- Genre:
- Research Report > New Finding (1.00)
- Industry:
- Technology: