Noncooperative Formation Control in the Presence of Malicious Agents: A Game-Based Strategy

This article investigates noncooperative formation control in the presence of malicious agents who disseminate negative information to their neighbors, resulting in misbehavior. A local cost function for agents is introduced to transform the formation control problem into a noncooperative game problem such that the Nash equilibrium solution corresponds to the agents' decisions in the desired formation. The resilient game strategy is proposed, comprising an event-triggered game algorithm and a threshold-based malicious agents detection algorithm, enabling each agent to determine the malice of their neighbors at each trigger moment and selectively interact with them, which guarantees that all normal agents can achieve Nash equilibrium, and the stability is analyzed by Lyapunov theory. Finally, the effectiveness of the algorithms is verified through a multi-unmanned aerial vehicles formation control experiment.

I. Introduction

The formation control problem of multi-agent systems (MAS) has received extensive attention because it shows the coordinated control ability between agents and is widely used in practical systems, such as multi-unmanned aerial vehicles (UAVs) [1], [2], [3], multimobile robots [4], [5], [6], [7], [8], and multi-autonomous surface vehicles [9], [10], [11]. According to the formation intention of the agent, formation control can be divided into static formation [5], [6], [12], [13], [14], [15], [16], and dynamic formation [1], [2], [3], [4], [7], [8], [9], [10], [11], [17], [18], which is also considered as formation deployment and formation tracking. The latter requires the agent to maintain a desired formation tracking dynamic trajectory, whereas the former refers to a time-invariant fixed formation configuration.