Active Fault-Tolerant Control for Steering Actuator Bias in Autonomous Vehicles Using Adaptive Sliding Mode Observer

Area of Science:

• Automotive Engineering
• Control Systems Engineering
• Robotics

Background:

• Steering actuator bias in autonomous vehicles causes path-tracking and stability issues.
• Faults like misalignment and degradation lead to steering errors, complicating diagnosis.
• Existing methods may require extra sensors or hardware redundancy.

Purpose of the Study:

• To develop an active fault-tolerant control framework for real-time detection, estimation, and mitigation of steering actuator bias.
• To enhance the robustness and practical implementation of fault detection in autonomous driving systems.

Main Methods:

• An adaptive sliding mode observer was designed using an extended four-state lateral error model.
• Adaptive observer gain tuning was employed to handle uncertainties from speed variations and model linearization.
• Simulations involved high-speed double lane changes with constant, drifting, and intermittent bias scenarios.

Main Results:

• The proposed framework reliably detected and estimated various steering bias types.
• Path-tracking accuracy was significantly improved compared to uncompensated systems.
• The method demonstrated effectiveness without additional sensors or controller switching.

Conclusions:

• The adaptive sliding mode observer offers a practical solution for mitigating steering bias in autonomous vehicles.
• The framework enhances vehicle stability and path-tracking performance under fault conditions.
• This approach is suitable for real-world implementation due to its sensor-agnostic nature.