Rapidly Tuning the PID Controller Based on the Regional Surrogate Model Technique in the UAV Formation

Area of Science:

• Robotics and Control Systems
• Aerospace Engineering
• Computational Intelligence

Background:

• Leader-follower formations are crucial for unmanned aerial vehicle (UAV) coordination.
• Tuning proportional-integral-derivative (PID) controllers for UAV formations is typically empirical and time-consuming.
• Existing surrogate models for controller tuning can be inaccurate due to singular points.

Purpose of the Study:

• To propose a novel regional surrogate model technique (RSMT) for accurate and efficient PID controller tuning in UAV formations.
• To address the limitations of global surrogate models in handling singular points.
• To enhance the speed and precision of controller parameter optimization.

Main Methods:

• Development of the regional surrogate model technique (RSMT) utilizing regional information entropy.
• Implementation of a classifier to screen out failed samples, allowing RSMT to focus on successful ones.
• Evaluation of RSMT by comparing Pareto fronts with traditional simulation models and global surrogate models.

Main Results:

• The proposed RSMT accurately reconstructs the simulation model for controller tuning.
• RSMT reduces the runtime for tuning PID controllers by an order of magnitude compared to global surrogate models.
• RSMT improves the accuracy of surrogate models by dozens of orders of magnitude, mitigating singular point issues.

Conclusions:

• The regional surrogate model technique (RSMT) offers a significant advancement in tuning UAV formation controllers.
• RSMT provides a faster and more accurate alternative to existing global surrogate modeling approaches.
• This method enhances the practical application of sophisticated control strategies in autonomous systems.