Multi-Agent Cooperation Based on Interest Group

As a social entity, agent needs to collaborate with others to execute a task in most multi-agent systems. Agent has its own intention and interest as an individual, which is ignored in most systems and has great influence on cooperating with other agents. In this paper, referring to the principle of interest group in real life, a multi-agent cooperation scheme based on interest group to solve the problem effectively is proposed. Agent mind arguments: satisfaction degree and trust degree are introduced in the scheme, as important criteria to establish and maintain Interest Group, which are also used to evaluate agent´s task quality and choose agent of the highest aggregative indicator as cooperation agent. Taking formation conduct system as a test example, the simulated results show that using the scheme can reduce the cost of communication and improve cooperative efficiency.