University of Twente Student Theses
Validation of machine learning computed muscle activation with experimental EMG
Amsing, Yaela M. (2022) Validation of machine learning computed muscle activation with experimental EMG.
|
PDF
7MB |
| Abstract: | Objective: In biomechanics, muscle-actuated control is researched through the use of neuromusculoskeletal models, and movement tasks can be performed with help of optimization techniques or machine learning. However, the accuracy of the muscle control that is generated by the simulation methods is still in question. Approach: In this study, the muscle control that is simulated using a Myosuite musculoskeletal model during an elbow flexion/extension movement is validated with the help of experimentally collected EMG data. This paper describes how the experimental data is recorded, and the method for obtaining simulated muscle activation. The accuracy of the simulated muscle activation is evaluated by comparing it to the experimental data. In addition, the position data that is achieved by the muscle activation is evaluated for different reinforcement learning policies. Main results: Results demonstrate the possibility of simulating realistic muscle activation when an optimized reward function is used. The approach was shown to be effective to some extent for extrapolating movements with different speeds. However, more experimental data, in the form of more subjects and different movements, is needed for further validating the used models and policies. Significance: The ability to successfully simulate biomechanical tasks with accompanying realistic muscle control creates possibilities in developing and testing prosthetics and exoskeletons, along with rehabilitation opportunities. Simulation of biomechanical tasks enables us to better understand human-machine interaction. |
| Item Type: | Essay (Master) |
| Faculty: | ET: Engineering Technology |
| Programme: | Biomedical Engineering MSc (66226) |
| Link to this item: | https://purl.utwente.nl/essays/93841 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page