Adaptive distributed observer design for nonlinear multiagent systems

Distributed state estimation is crucial for the leader-following control problem of multiagent systems (MASs). In this paper, adaptive distributed observers (DOs) are designed for a nonlinear autonomous leader system using only its output. Via system transformation, the DO design problem of the origin system is converted to that of a canonical system with lumped dynamics. When the lumped dynamics is parametric, an adaptive DO is developed to reconstruct the state and the unknown parameters under an undirected topology, which also addresses the distributed state/parameter estimation problem of an uncertain autonomous system with its dynamics in a parametric form. Then, the DO framework is extended to the case of non-parametric uncertainties, and a neural network (NN) DO is designed for reconstructing the state/lumped dynamics over a strongly connected digraph Finally, the effectiveness of the proposed DOs is demonstrated via numerical simulations.