Predictor‐based bipartite time‐varying formation control of nonlinear multi‐agents systems via disturbance observer

Abstract

AbstractFormation control is one of critical control issues in multi‐agent systems (MASs). In this article, we extend the traditional formation to a bipartite one in a structurally balanced signed graph and propose a bipartite formation control strategy for an MAS with unknown external disturbances and nonsymmetric input saturation constraints. Neural networks (NNs) are developed to approximate followers' dynamics, and their transient performance are improved by prediction errors. On this basis, an NN‐based nonlinear disturbance observer (NDO) is constructed to estimate generalized disturbances including external disturbances and NNs approximation errors. In order to reduce the number of learning parameters, a normalization method is introduced for NNs weights and applied to the proposed NDO. An auxiliary system is constructed to compensate for asymmetric input saturation, and it is not required to access unknown control gains. The stability analysis is presented, and it is proven that all signals in the MAS are bounded. A numerical simulation and a team of quadrotors examples corroborate the effectiveness and feasibility of the strategy.