Loading...

Designing a Meta-Learning Algorithm of Knee Joint Angle Prediction for Lower Limb Exoskeleton

Mortazavi, Hassan | 2024

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 57927 (08)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Vossoughi, Gholamreza
  7. Abstract:
  8. This research aims to design and evaluate a generalized algorithm for predicting knee joint angle during gait. To achieve this, electromyography (EMG) sensors were attached to muscles involved in knee joint motion to record their activity. By analyzing and extracting features from these signals, a machine learning model was developed to establish a relationship between the EMG signals as input and the knee joint angle as output. The study prioritizes a model that can be quickly adapted to new users with minimal data, making meta-learning the core approach. This method shares similarities with transfer learning and was trained using two datasets, each containing data from over 10 subjects. The trained model was evaluated in various scenarios, including predictions for unseen individuals within the same dataset and individuals from a different dataset. In the first scenario, the model required minimal recalibration for unseen users within the same dataset and delivered accurate predictions. In the main scenario, the trained model, calibrated on data collected from a new user in the laboratory, was evaluated under varying amounts of adaptation data. When a larger amount of user-specific data was available, transfer learning achieved an average error of 6% at higher speeds, while meta-learning resulted in an 11% error. However, in a low-data setting (few-shot scenario), meta-learning outperformed transfer learning with an average error of 11% compared to 14% at higher speeds. This highlights the effectiveness of the proposed meta-learning approach for adapting knee angle prediction models to new users with limited data
  9. Keywords:
  10. Exoskeleton ; Machine Learning ; Electromyography ; Metalearning ; Knee Rehabilitation ; Motion Prediction ; Knee Angle

 Digital Object List

 Bookmark

No TOC