Recognition of Drivers' Hard and Soft Braking Intentions Based on Hybrid Brain-Computer Interfaces.

Area of Science:

• Neuroscience
• Human-Computer Interaction
• Automotive Engineering

Background:

• Intelligent assistant driving systems require accurate prediction of driver intentions.
• Existing methods often lack the ability to reliably differentiate subtle driving maneuvers like braking.

Purpose of the Study:

• To develop and evaluate novel hybrid brain-computer interfaces (hBCIs) for classifying driver intentions (hard braking, soft braking, normal driving).
• To explore simultaneous and sequential fusion strategies for electroencephalography (EEG) and electromyography (EMG) signals.

Main Methods:

• Proposed simultaneous hBCIs with feature-level (hBCI-FL) and classifier-level (hBCIs-CL) fusion.
• Developed sequential hBCIs (hBCI-SE1 prioritizing EEG, hBCI-SE2 prioritizing EMG) for hard braking detection.
• Utilized spectral features and a one-vs-rest classification strategy.

Main Results:

• The sequential hBCI-SE1, prioritizing EEG signals with spectral features and a one-vs-rest strategy, demonstrated the highest performance.
• Achieved an average system accuracy of 96.37% for classifying driver intentions.
• Demonstrated the efficacy of hybrid BCIs in real-time driving intention recognition.

Conclusions:

• Hybrid brain-computer interfaces integrating EEG and EMG signals offer a promising approach for advanced driver assistance.
• The proposed hBCI-SE1 model provides a robust and accurate method for detecting critical driving events like hard braking.
• This research contributes to the development of human-centric intelligent driving systems, enhancing safety and comfort.