Risk Control in Multi-Agent Coordination by Negotiation with a Trusted Third Party

Abstract

In multi-agent coordination, the uncertainty may come from two major sources: the moves of the nature agent and the unpredictable behavior of other autonomous agents. The uncertainty may affect the expected payoff and the risk of an agent. A rational agent would not always play the strategy that gives the highest expected payoff if the risk is too high. To tackle the uncertainty in multi-agent coordination, a risk control mechanism is necessary in multiagent decision making. We assume agents may have different risk preferences, e.g. risk-averse, riskneutral, and risk-seeking, and separate the risk preference from the utility function of a given strategy. Taking agent's risk preference into account extends the notions of the dominant strategy, the Nash equilibrium, and the Pareto-efficiency in traditional game theory. We show how the risk control can be carried out by a negotiation protocol using communication actions of asking guarantee and offering compensation via a trusted third party. 1