Plug-and-Play Distributed Fencing of Multi-Agent Systems for an Uncooperative Target Without Velocity Measurements

This paper investigates the target fencing problem of multi-agent systems (MASs) for an uncooperative, freely-moving target whose position is accessible solely to a subset of agents. A plug-and-play distributed target fencing strategy is proposed, comprising a finite-time target acceleration estimator, a distributed target motion observer, and a label-free target fencing controller. Under the proposed fencing strategy, all agents can asymptotically observe the target position and acceleration. Moreover, without needing velocity measurement, the target is asymptotically fenced with the velocity convergence and the avoidance of inter-agent collisions. Furthermore, the plug-and-play feature of the proposed strategy ensures that each agent can join or leave the MAS during the fencing process, and the activated agents can still fence the target via self-organization without the need to adjust any parameters of the activated agents. The results are proved by rigorous theoretical analysis and verified by numerical and real-world experiments. Note to Practitioners—This paper focuses on the target fencing problem of multi-agent systems (MASs), which has potential applications in target protection or assault for multiple vehicles. Unlike existing approaches that concentrate on a target with simple movement patterns, this paper considers a freely moving target whose position is measurable solely by a portion of agents, which is more practical in real fencing scenarios. A distributed fencing strategy is proposed to enable all agents to estimate the target position and acceleration and to cooperatively fence the target without measuring their own velocity. Under the proposed strategy, each agent self-organizes its behavior by following identical rules. Each agent can join or leave the MAS while the target remains fenced via the self-organization of remaining agents without modifying any agent’s parameters. This effectively reduces the complexity of configuring numerous agents, such as multiple vehicles, in engineering applications under dynamic situations. The strategy can be applied to multiple vehicles for uncooperative target assaults in complex situations where vehicles are damaged at unknown instants and supported when needed.