Optimized Bipartite Consensus Tracking Control of Low-Altitude Autonomous Swarms Subject to Global Actuator Constraints and Nonfully Periodic Disturbances

This paper investigates the optimal bipartite consensus tracking control issue of unmanned aerial vehicle (UAV) and unmanned ground vehicle (UGV) swarms under global actuator constraints and non-fully periodic disturbances. A novel UAV-UGV swarm bipartite consensus tracking control framework is established. Multiple gusts in the environment or irregular electromagnetic oscillations in the equipment can generate non-fully periodic disturbances in the system. Meanwhile, some physical or electromagnetic uncertainties cause unknown persistent actuator saturation that should not occur, which indirectly affects the observation and compensation of composite disturbances, and this can be addressed by the control input increment. Hence, a composite anti-disturbance control method based on a multiple harmonic disturbance observer and robust theory is proposed, which observes and compensates for periodic disturbances, and suppresses the residual disturbance through the improved robustness of each system. A reinforcement learning (RL) algorithm under the gradient descent method is utilized to construct an actor-critic-identifier (ACI) structure, which decouples the actuator constraints in the multiple harmonic disturbance observer, and obtains the ideal control input increment compensation. Additionally, to mitigate potential excessive trajectory amplitude caused by initial actuator constraints in the swarm, where control input compensation remains incomplete during the startup phase, an improved decay function is proposed to ensure trajectory error convergence under actuator constraints from the initial movement, while offering the advantage of adaptive self-tuning. Then, an adaptive ACI structure within the decay function is designed, and an optimal bipartite consensus tracking anti-disturbance control strategy under actuator constraints is proposed for the UAV-UGV swarm. Finally, simulations and experimental results are performed to demonstrate the efficacy of the proposed approach.