An improved swarm model with informed agents to prevent swarm-splitting

The spatial aggregation of a large number of individuals and the coordination of individual behavior within the group are the two core characteristics of swarm behavior. Swarm-splitting blocks the information interaction between individuals, making it difficult for a swarm to stay together and achieve cooperation. In this respect, an improved distributed swarm model with a dual-adaptive feedback mechanism to prevent swarm-splitting and to improve the probability of reaching the target area is proposed. The first feedback mechanism is for informed agents to balance goal-oriented and social-oriented behavior, which helps informed agents maintain navigation accuracy while staying close to their neighbors. The second feedback mechanism is for followers to adjust their perception range adaptively, which helps the followers select appropriate neighbors based on the state of the nearby agents. Four metrics are provided to evaluate the swarm's performance, namely swarm connectivity, the average degree of temporal dependence, the average degree of temporal dependence, and the arrival rate. Simulation results show that the proposed swarm model outperforms the existing swarm models under the four metrics. The proposed model can be used for the distributed migration motion of large-scale unmanned swarms, such as navigation and target tracking.

Automated summary

The proposed paper introduces an improved distributed swarm model with a dual-adaptive feedback mechanism to prevent swarm-splitting and improve the probability of reaching the target area. The first feedback mechanism balances goal-oriented and social-oriented behavior of informed agents to maintain navigation accuracy while staying close to neighbors. The second feedback mechanism helps followers adaptively adjust their perception range and select appropriate neighbors based on nearby agents' state. Simulation results demonstrate the superiority of the proposed model compared to existing swarm models in four performance metrics. The proposed model has potential applications in the distributed migration motion of large-scale unmanned swarms, such as navigation and target tracking.