Neural Network-Based Control for Affine Formation Maneuver of Multi-Agent Systems with External Disturbances

This article proposes a distributed control law with a neural network to achieve leader-follower affine formation maneuver (AFM) of multi-agent systems (MASs) subjected to time-varying external disturbances. The leaders determine the desired collective formation maneuvering such as scaling, shearing, translation, and rotation. The distributed controller ensures that the followers are tracking the maneuvers of the leaders at all times despite the disturbances. The disturbances are approximated and compensated with the aid of neural networks. A Lyapunov candidate function is used to highlight that the closed-loop system is uniformly ultimately bounded. Finally, in order to validate the proposed control method, it is applied to a multi-agent system of quadrotors. The simulation results have shown that the proposed control protocol is able to maintain the collective maneuvering of the quadrotors in the presence of time-varying disturbances.