Data-driven modeling and adaptive event-triggered secure control for autonomous vehicles subject to sensor attacks.

Area of Science:

• Control Engineering
• Cybersecurity
• Autonomous Systems

Background:

• Autonomous vehicles face vulnerabilities from sensor attacks, compromising safety and performance.
• Traditional control methods struggle with the complexities of real-world sensor attacks and data-driven modeling.

Purpose of the Study:

• To develop a data-driven secure control strategy for autonomous vehicles against sensor attacks.
• To enhance communication efficiency and control performance using an adaptive event-triggered mechanism.

Main Methods:

• Dynamic Mode Decomposition (DMD) for data-driven lateral vehicle model identification.
• Adaptive event-triggered control scheme to optimize communication and performance.
• Sliding-mode-like control to counteract sensor attacks.
• Lyapunov theory and Linear Matrix Inequalities (LMIs) for stability analysis.

Main Results:

• Successful identification of autonomous vehicle dynamics from data using DMD.
• Development of an adaptive event-triggered scheme that balances communication load and control effectiveness.
• Demonstration of effective mitigation of sensor attacks through the proposed control strategy.
• Validation of the control scheme's effectiveness via comparative simulations.

Conclusions:

• The proposed data-driven approach effectively addresses sensor attacks in autonomous vehicles.
• DMD simplifies model identification, while the adaptive event-triggered control enhances system efficiency.
• The secure control scheme offers a robust solution for enhancing the safety and reliability of autonomous driving.