Hierarchical neural affine formation maneuver control of air-bearing robots with prescribed performance and collision avoidance

This paper proposes a hierarchical neural network control framework with prescribed performance for affine maneuver formation of air-bearing robots (ABRs), addressing actuator nonlinearities, false data injection (FDI) attacks, and collision avoidance. Firstly, a virtual system is constructed to establish links between leaders and followers, which calculates and map the leader’s affine formation for each follower while preventing fault propagation in followers. Additionally, a novel prescribed performance controller considering collision between follower ABRs is proposed, integrating a neural network with an extended state observer (ESO) for FDI attacks and actuator nonlinearities. In particular, the comparison between control inputs and saturation thresholds is used for performance boundary design, thus achieving prescribed performance without vulnerability to input saturation. Theoretical analysis guarantees system stability, and experimental results demonstrate the method’s effectiveness. • Hierarchical neural framework for ABR affine formation with prescribed performance. • Adaptive PPC boundaries prevent vulnerability under actuator saturation. • Confidence factors and virtual systems block FDI attack propagation. • Obstacle Lyapunov function ensures collision avoidance among followers.