Genetic-Algorithm-Assisted Self-Scheduled Multidelay PIR Control: Experiments in a Car-Like Vehicle System

Area of Science:

• Control Systems Engineering
• Robotics
• Automotive Engineering

Background:

• Intelligent vehicle path-tracking is challenged by high-frequency measurement noise.
• Existing proportional-integral-derivation (PID) controllers struggle with noise attenuation.

Purpose of the Study:

• To propose a novel multidelay proportional-integral-retarded (PIR) control strategy for intelligent vehicle path-tracking.
• To effectively attenuate high-frequency measurement noises while ensuring accurate path tracking.

Main Methods:

• Modeling the vehicle as a linear parameter-varying (LPV) system with car position as the scheduling variable.
• Designing a multidelay PIR controller incorporating a retarded term for noise attenuation.
• Utilizing linear matrix inequalities (LMIs) and Taylor's expansion for controller parameter tuning.
• Formulating a self-scheduled tracking controller as a weighted sum of convex subcontrollers.

Main Results:

• The proposed multidelay PIR controller effectively reduces tracking errors and attenuates high-frequency measurement noises.
• The self-scheduled controller adapts to varying operational conditions through adaptive weight functions.
• Real-time experiments on a laboratory vehicle validated the controller's performance.

Conclusions:

• The multidelay PIR control offers a robust solution for intelligent vehicle path-tracking in the presence of measurement noise.
• The LPV framework and LMI-based tuning provide an effective method for designing adaptive controllers.
• This approach enhances the reliability and performance of autonomous driving systems.