LAUSR

Abstract In this paper, the bipartite state synchronization problem of heterogeneous multi-agent systems (MASs) with active leader under adversarial signals over signed digraph is studied. The non-zero control of active leader is bounded, and we can update this control to protect followers from possible danger. Firstly, the bipartite state synchronization problem over signed digraph is converted to a regular tracking synchronization problem over nonnegative digraph after state transformations for heterogeneous MASs with active leader under adversarial signals. Then, novel distributed observers are designed to estimate the state of leader, as well as prevent disrupted data of agent from propagating the communication networks. Observer-based H∞ optimal controllers are also designed for each follower to mitigate effects on disrupted agents that agents under attack, and the inhomogeneous algebraic Riccati equations (AREs) are obtained. Meanwhile, a data-based off-policy reinforcement learning (RL) algorithm is used to solve the AREs without requiring the dynamics of agents. At last, the effectiveness of the algorithm is demonstrated by a simulation example.