Fine-tuning Myoelectric Control through Reinforcement Learning in a Game Environment
Artikel i vetenskaplig tidskrift, 2025
Methods: The starting point of our method is a SL control policy, pretrained on a static recording of electromyographic (EMG) ground truth data. We then apply RL to fine-tune the pretrained classifier with dynamic EMG data obtained during interaction with a game environment developed for this work. We conducted real-time experiments to evaluate our approach and achieved significant improvements in human-in-the-loop performance.
Results: The method effectively predicts simultaneous finger movements, leading to a two-fold increase in decoding accuracy during gameplay and a 39% improvement in a separate motion test.
Conclusion: By employing RL and incorporating usage-based EMG data during fine-tuning, our method achieves significant improvements in accuracy and robustness. Significance: These results showcase the potential of RL for enhancing the reliability of myoelectric controllers, which is of particular importance for advanced bionic limbs. See our project page for visual demonstrations: https://sites.google.com/view/bionic-limb-rl.
Reinforcement learning
Electromyography
Human computer interaction
Deep Learning
Prosthetic limbs
Författare
Kilian Tamino Freitag
Chalmers, Elektroteknik, System- och reglerteknik
Yiannis Karayiannidis
Lunds universitet
Jan Zbinden
Chalmers, Elektroteknik, System- och reglerteknik
Rita Laezza
Chalmers, Elektroteknik, System- och reglerteknik
IEEE Transactions on Biomedical Engineering
0018-9294 (ISSN) 15582531 (eISSN)
Vol. In PressÄmneskategorier (SSIF 2025)
Robotik och automation
Datavetenskap (datalogi)
DOI
10.1109/TBME.2025.3578855