Time-Varying Optimal Formation Control for Second-Order Multiagent Systems Based on Neural Network Observer and...

Area of Science:

• Robotics and Control Systems
• Artificial Intelligence
• Networked Systems

Background:

• Multiagent systems (MASs) face challenges in achieving coordinated formations due to uncertain nonlinear dynamics and sensor limitations.
• Existing optimal formation control methods struggle with unknown system dynamics and unmeasurable states, particularly in time-varying scenarios.

Purpose of the Study:

• To develop a distributed, time-varying optimal formation control protocol for second-order uncertain nonlinear dynamic MASs.
• To address challenges posed by local information constraints, unmeasurable states, and unknown system dynamics.

Main Methods:

• Utilized an adaptive neural network (NN) state observer combined with a backstepping method.
• Employed simplified reinforcement learning (RL) for Hamilton-Jacobi-Bellman optimization.
• Ensured stability using Lyapunov stability theory.

Main Results:

• Successfully achieved desired second-order formation configurations with guaranteed stability (SGUUB errors).
• The proposed protocol effectively handles unmeasurable states and unknown dynamics.
• Demonstrated ease of updating critic and actor components in the RL framework.

Conclusions:

• The developed optimal control scheme provides a robust solution for time-varying formation control in uncertain MASs.
• Validated the theoretical framework through rigorous Lyapunov stability analysis and digital simulations.
• Offers a practical approach for real-world applications with sensor constraints.