Ant-inspired Decentralized Task Allocation Strategy in Groups of Mobile Agents

Abstract

This paper presents a hybrid task allocation strategy for decentralized groups of mobile agents. The work is strongly inspired by the biological behavior of social insects. Agent based modeling approach has been used to model the behavior of agents. Environment of the model comprises of set of tasks and heterogeneous groups of mobile agents. Each group of agents follows respective local rules to communicate with neighboring agents and with the environment in their vicinity. The agents can not only carry out their own tasks but can also switch tasks to meet colonial demands. This paper presents an extension of a series of models presented earlier. In earlier models, agents reacted only to stimulus. Although evidence exists of task preferences among castes (bias) in many ant species, little work has been done in investigating the benefits for the colony of such preferences. This paper addresses the role of bias in the colony performance under different environmental circumstances. Experiment results indicate that the model presented in this paper is highly efficient, accurate and consistent with biological equivalents. The model with bias within a group is novel and has not been done before to the best of my knowledge.