Multiagent Collaborative Search with Self-Interested Agents

This paper presents an experimental study on collaborative search by distributed autonomous agents. We consider a problem such that multiple agents search for target objects in a field, and communicate with each other to exchange information of objects. Agents have six different strategies: cooperative, skeptical, free rider, liar, skeptical liar and solitary. These strategies characterize behaviors of agents for transmitting/receiving information in different ways. We observe the effect of self-interested agents who act non-cooperatively or even act deceptively to increase their own profits. We identify situations in which self-interested agents have advantages and investigate conditions which are effective to suppress the development of self-interested agents.