Robust formation maintenance of heterogeneous multi-vehicle systems: a neural network strategy under communication delays

Formation control strategies for multi-vehicle systems are fundamental to their practical applications. Existing research primarily relies on real-time state information and higher-order derivative data for effective formation maintenance. This paper introduces a neural network-based formation control method for heterogeneous multi-vehicle systems that demonstrates remarkable robustness when subjected to communication delays. The proposed approach employs a consensus protocol that requires only state information and state error data exchange between vehicles, substantially reducing communication overhead. The neural network implementation, coupled with an adaptive tuning mechanism, eliminates dependency on precise vehicle model parameters, enabling the controller to effectively estimate diverse nonlinear dynamics present in heterogeneous vehicle systems. Rigorous analysis demonstrates that the approximation errors diminish over time, ensuring that each vehicle's state remains within a solvable compact set. Notably, although the controller was not explicitly designed to handle communication delays, extensive simulation results reveal its inherent robustness against delayed information exchange. The findings show that the proposed strategy successfully maintains formation stability across various vehicle configurations even when subjected to significant communication delays, highlighting the potential of neural network strategies in addressing real-world challenges for heterogeneous multi-vehicle systems operating in non-ideal communication environments.