Abstract
This paper proposes a new distributed finite-time differentiator (DFD) for multi-agent systems (MAS) under directed graph, which extends the differentiator algorithm from the centralized case to the distributed case by only using relative/absolute position information. By skillfully constructing a Lyapunov function, the finite-time stability of the closed-loop system under DFD is proved. Inspired by the duality principle of control theory, a distributed continuous finite-time output consensus algorithm extended from DFD for a class of leader-follower MAS is provided, which not only completely suppresses disturbance, but also avoids chattering. Finally, several simulation examples are given to verify the effectiveness of the DFD.
Introduction
In the coordination of multi-agent systems (MAS), such as drone swarms or autonomous underwater vehicles, knowing the leader's velocity is often a prerequisite for effective control. However, equipping every agent with high-precision velocity sensors is frequently impractical due to constraints in cost, weight, and space. While centralized differentiators for single systems are well-established, they are insufficient for distributed networks where agents only communicate with immediate neighbors.
This research addresses the challenge of estimating a leader's velocity using only local position data under the constraints of a directed communication graph. The authors introduce a Distributed Finite-time Differentiator (DFD) that enables a network of agents to reach consensus on a leader's signal and its derivative within a guaranteed, finite time.
Methodology
The proposed DFD comes in two variants: DFD-R (Relative) for scenarios using relative position measurements (e.g., via onboard cameras) and DFD-A (Absolute) for scenarios with global coordinates (e.g., GPS). The core second-order algorithm for each agent $i$ is defined as:
$$\dot{\hat{p}}_i = \hat{q}_i - k_1\lfloor y_i \rceil^{1/2}$$ $$\dot{\hat{q}}_i = -k_2\lfloor y_i \rceil^0 + u_i$$
where $y_i$ is the neighborhood error. The inclusion of fractional powers and signum-like terms ensures finite-time convergence, a significant improvement over standard asymptotic methods.
A key innovation is the algorithm's robustness on directed graphs. Unlike many studies assuming bidirectional communication, this DFD works as long as at least one agent can observe the leader, allowing information to propagate through the network even with asymmetrical signal strengths. Furthermore, by applying the duality principle, the researchers developed a continuous consensus controller that eliminates "chattering"—high-frequency oscillations that can damage mechanical hardware in traditional sliding mode control.
Results and Discussion
The effectiveness of the DFD was validated through numerical simulations involving complex, time-varying signals and external disturbances. The results demonstrate rapid convergence, with estimation errors for both position and velocity dropping to zero in less than one second.
The system proved robust against bounded external disturbances, modeled as periodic oscillations. The extended consensus controller achieved precise tracking without the jagged control inputs typical of non-continuous methods. However, the system's performance relies on the network remaining "strongly connected" and the leader's acceleration being bounded.
Conclusion
This work provides a robust framework for velocity estimation in distributed networks using only position data. By extending centralized differentiator theory to directed graphs and ensuring finite-time stability, the authors offer a practical solution for MAS coordination. Future research will focus on extending this framework to higher-order systems and validating the algorithm on physical robotic hardware.
Summary updated