Distributed Formation Splitting and Merging for Multiple UAVs With Actuator Faults Under Cooperation-Antagonism Interactions

Abstract

In this paper, a formation splitting and merging problem for multiple quadrotor UAVs with actuator faults is investigated. A fully distributed neural fault-tolerant formation splitting and merging control algorithm is designed. Under the leading-following framework, the formation splitting and merging maneuver of multiple UAVs can be achieved by assigning cooperation or antagonism between any two UAVs, without the need to plan paths for all agents. Firstly, a distributed finite-time estimator is proposed for some follower UAVs that can't obtain the leader's velocity. Then, the introduction of adaptive dynamic gain avoids the controller's dependence on the global topology information, while also addressing the impact caused by partial loss-of-effectiveness faults. Additionally, adaptive mechanisms and RBF-NN with a single parameter learning law are used to deal with unknown but bounded control input of the leader, disturbance, bias faults, and nonlinearity in the system dynamics. Finally, the effectiveness of the proposed formation control scheme is verified by simulation.