Shepherding Algorithm Based on Variant Agent Detection for Heterogeneous Flock

The problem of guiding a flock of agents using repulsion forces exerted by a smaller number of agents is called the shepherding problem. The objective of this problem is to design shepherd agent methods so that the shepherd agent guides sheep agents to the destination area in a short time. Although various shepherding methods have been proposed, most of these methods assume that the sheep agents have the same dynamics. However, this assumption does not necessarily hold in reality. Therefore, in this paper, we propose two shepherding methods for a flock of sheep agents including variant agents. In the proposed method, the shepherd agent estimates the trajectory of the sheep agents and discriminate the type of agents based on the degree of deviation from the predicted trajectory. Then, the shepherd agent guides sheep agents discriminated to be normal. Numerical simulations showed that the proposed methods can guide more agents than the conventional method. The shepherd agent guided average 84% of normal sheep by the proposed methods while 64% by a conventional method.

Automated summary

This paper proposes two shepherding methods for guiding a flock of sheep agents, including variant agents. By estimating the trajectory of the sheep agents and discriminating their types based on the degree of deviation, the proposed methods can guide more sheep agents than conventional methods.