Fully Distributed Finite-Time Formation Tracking of Heterogeneous Multi-Agent Systems Under Directed Networks
Binghe An, Huijin Fan, Bo Yuan Wang, Lei Liu · IEEE Transactions on Automation Science and Engineering · 2025
This paper considers the finite-time formation tracking of high-order heterogeneous multi-agent systems (HMASs), where the communication network among agents is directed and the leader’s input is unknown to all followers. An innovative fully distributed finite-time control strategy, independent of global topological information, is proposed. First, a fully distributed observer is constructed to estimate the leader’s state in finite time, despite leader’s unknown input. It is noted that an adaptive gain is wisely designed for the observer such that global topological information is no longer necessary. However, due to the asymmetry of the Laplacian matrix and the fact that the adaptive gain may not converge to its ideal value, the commonly used finite-time convergence criteria are not applicable to our case, and achieving the finite-time convergence of the observer becomes rather challenging. To overcome technical obstacles, a suitable Lyapunov function is specifically constructed, and then the desired finite-time convergence is established with the aid of a contradiction argument. Further, a disturbance observer is designed by utilizing the state transformation, which is able to estimate the composite disturbances consisting of both external disturbances and the leader’s unknown input. Finally, an observer-based formation controller is proposed to achieve the desired configuration. It is guaranteed that zero formation error, rather than just a neighbor of zero, can be realized in finite time through chattering-free control signals. Simulations are performed to verify the effectiveness of the proposed algorithm. Note to Practitioners—The formation tracking of HMASs is investigated in this paper, which has wide applications such as unmanned aerial vehicle (UAV)-unmanned ground vehicle (UGV) collaborative reconnaissance and multi-type UAV coordinated attack. Furthermore, the convergence time is very important for some urgent tasks, such as multi-UAV collaborative firefighting. Hence, the finite-time control approaches have been developed for multi-agent systems (MASs). However, most existing finite-time methods for general linear MASs require the global topological information, such as the eigenvalues of the Laplacian matrix or the number of agents, to determine the parameters. Obtaining such global information is difficult especially in large-scale networks. Therefore, the need for the global topological information reduces both the practicality of the algorithm and the scalability of the system. To address this problem, a fully distributed finite-time formation approach is proposed under directed networks, where the global topological information is successfully avoided through adaptive strategy. It is guaranteed that the formation error converges to zero in finite time, despite disturbances and the leader’s unknown input. In addition, instead of using a controller that generates high-frequency chattering to suppress uncertainties, our proposed robust controller produces a chattering-free output with the aid of a novel disturbance observer, thereby reducing actuator wear and enhancing the algorithm’s practicality.