Multimodal data-based human motion intention prediction using adaptive hybrid deep learning network for movement
1Advanced Manufacturing Institute, King Saud University, Riyadh, 11421, Saudi Arabia. mabidi@ksu.edu.sa.
Scientific Reports
|December 24, 2024
Summary
This study introduces an Adaptive Hybrid Network (AHN) for predicting human motion intention using electroencephalogram (EEG) and electromyography (EMG) signals. The novel approach enhances personalized assistance for rehabilitation robots, outperforming traditional methods.
Area of Science:
- Biomedical Engineering
- Robotics
- Artificial Intelligence
Background:
- Increasing social demand for improved quality of life among the elderly and disabled.
- Need for advanced assistive and rehabilitation robots.
- Challenges in human-machine interaction for personalized assistance.
Purpose of the Study:
- To implement an Adaptive Hybrid Network (AHN) for effective human motion intention prediction.
- To enhance personalized assistance in assistive and rehabilitation robots.
- To fuse biomedical signals and sensor data for accurate prediction.
Main Methods:
- Collected multimodal data: electroencephalogram (EEG)/electromyography (EMG) signals and sensor measures.
- Utilized AH-CNN-LSTM for EEG/EMG spectrograms and AH-CNN-Res-LSTM for sensor data.
- Optimized AHN parameters using the Improved Yellow Saddle Goatfish Algorithm (IYSGA).
Main Results:
- The proposed AHN model demonstrated superior performance compared to standard models.
- Effective prediction of human motion intention was achieved.
- The integration of AH-CNN-LSTM and AH-CNN-Res-LSTM proved successful.
Conclusions:
- The developed Adaptive Hybrid Network (AHN) offers a promising solution for human motion intention prediction.
- This technology can significantly improve personalized assistance for rehabilitation and assistive robots.
- Further research can explore broader applications of this fused multimodal approach.
