Detection of Driving Capability Degradation for Human-Machine Cooperative Driving.

Area of Science:

• Automotive Engineering
• Human-Computer Interaction
• Control Systems

Background:

• Current cooperative driving systems face challenges in determining optimal driver assistance timing.
• Driver state-based algorithms have limitations in accurately assessing driver readiness for intervention.
• Human drivers remain critical in public transportation systems due to technological and cost constraints.

Purpose of the Study:

• To develop a novel index for evaluating driver capability beyond traditional driver state monitoring.
• To establish a reliable method for detecting driver capability degradation in real-time.
• To enable seamless and safe transitions between manual and automated driving modes.

Main Methods:

• Proposed a new metric: the correction ability of the driver, combined with driving risk.
• Introduced a degraded domain (DD) to detect driver capability degradation.
• Utilized log-normal distribution for DD boundary modeling and an online algorithm for interactive parameter updates to identify individual driving styles.

Main Results:

• Bench validation confirmed the DD boundary identification algorithm converges effectively.
• The algorithm accurately reflects individual driving characteristics and styles.
• The proposed degradation detection method successfully identifies the optimal switching time for driver assistance.

Conclusions:

• The novel degraded domain (DD)-based system enhances cooperative driving safety.
• This approach provides a robust mechanism for determining when driver assistance is required.
• The system facilitates safe vehicle operation by managing transitions between manual and automatic driving.