Online Learning Agents for Group Cooperative Control Systems

This paper introduces a novel approach to group cooperative control in MASs by designing an online learning neural network for follower agents. The proposed neural network is capable of dynamically adjusting its structure and parameters in real-time based on the agents’ current states and input data. This feature allows the network to accurately approximate and continuously learn the complex nonlinear components present in the control systems, thereby achieving higher accuracy and stronger robustness in consensus tracking. Unlike traditional control methods, our approach leverages the adaptive capabilities of online learning to effectively handle dynamic changes and uncertainties, significantly enhance the overall performance and stability of the cooperative control system. Simulation results and experimental validations confirm the effectiveness of the proposed method, demonstrating its superiority in tracking accuracy and robustness.