Distributed Leader Following of an Active Leader for Linear Heterogeneous Multi-Agent Systems

TL;DR

This paper presents a distributed algorithm enabling heterogeneous linear followers in a directed acyclic graph to track an active leader's sampled states, even with limited information, ensuring energy-efficient control and formation maintenance.

Contribution

It introduces a novel distributed leader-following algorithm for heterogeneous LTI agents with limited leader information and sampled data, applicable to formation control.

Findings

Followers can accurately track the leader's sampled states.

The control inputs are proven to be minimum energy.

Homogeneous followers synchronize after the first sampling epoch.

Abstract

This paper considers a leader-following problem for a group of heterogeneous linear time invariant (LTI) followers that are interacting over a directed acyclic graph. Only a subset of the followers has access to the state of the leader in specific sampling times. The dynamics of the leader that generates its sampled states is unknown to the followers. For interaction topologies in which the leader is a global sink in the graph, we propose a distributed algorithm that allows the followers to arrive at the sampled state of the leader by the time the next sample arrives. Our algorithm is a practical solution for a leader-following problem when there is no information available about the state of the leader except its instantaneous value at the sampling times. Our algorithm also allows the followers to track the sampled state of the leader with a locally chosen offset that can be time-varying. When the followers are mobile agents whose state or part of their state is their position vector, the offset mechanism can be used to enable the followers to form a transnational invariant formation about the sampled state of the leader. We prove that the control input of the followers to take them from one sampled state to the next one is minimum energy. We also show in case of the homogeneous followers, after the first sampling epoch the states and inputs of all the followers are synchronized with each other. Numerical examples demonstrate our results.