Related Experiment Video
Updated: May 10, 2025
06:37
Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
Published on: December 15, 2023
2.4K
Driver Head-Hand Cooperative Action Recognition Based on FMCW Millimeter-Wave Radar and Deep Learning
Lianlong Zhang1, Xiaodong Chen1, Zexin Chen1
1College of Electronic Engineering, South China Agricultural University, Guangzhou 510642, China.
Sensors (Basel, Switzerland)
|April 26, 2025
Summary
Millimeter-wave radar enhances driver safety by accurately detecting head-hand actions. A novel deep learning model achieved 99.16% accuracy, outperforming existing methods for driver status monitoring.
Area of Science:
- Automotive Safety
- Sensor Technology
- Artificial Intelligence
Background:
- Visual recognition methods for driver monitoring face limitations due to lighting, occlusion, and privacy.
- Millimeter-wave radar offers advantages like accuracy, cost-effectiveness, and robustness to environmental conditions for driver status detection.
Purpose of the Study:
- To develop an accurate and reliable system for driver action recognition using millimeter-wave radar.
- To address the limitations of existing driver action recognition methods in terms of accuracy and scope.
Main Methods:
- Utilized a 77 GHz millimeter-wave radar to collect data on seven types of driver head-hand cooperative actions.
- Developed a deep learning model (VGG16-LSTM-CBAM) that processes micro-Doppler spectrograms for action classification.
Main Results:
- The proposed VGG16-LSTM-CBAM network achieved a high classification accuracy of 99.16%.
- Demonstrated superior performance compared to existing Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) and AlexNet-LSTM networks.
Conclusions:
- The developed millimeter-wave radar system and deep learning model significantly improve driver action detection accuracy.
- This approach offers a promising solution for enhancing in-car safety systems.
