Research On Collaborative Mechanism In Multi-agent System And Its Application

Multi-Agent System (MAS) makes agent with simple functions to reach complex task goal byimplementing a collaborative mechanism. In the real world, many applications can be implementedby MAS, e.g. intelligent transportation model system, multi-robot soccer, multi-UAV cooperates todetect or combat with enemy and so on. Collaborative mechanism is one of the most importantaspects of MAS, so research on collaborative problem in MAS is of significant value in theory andapplication.This paper research on dynamic collaborative mechanism under the circumstance that agents aredistributed, the main research work includes:1) By analyzing the cooperation problem in MAS,design a multi-agent collaboration model based on planning and negotiation;2) Research on somemulti-agent planning methods and improve existed partial planning, propose a planning methodcombined with prediction and design corresponding algorithm;3) Give an evaluation method to helpagent estimate the quality of planning result, it can be used in the process to solve conflicts betweenseveral agents;4) Design a multi-agent collaborative planning framework. For conflict resolutionproblem, propose a conflict detection method based on monitoring and design negotiation algorithmto solve conflict. Then, verify the conflict resolution method is effective by typical simulations.This paper implements planning algorithm and conflict resolution algorithm with Java languageunder Eclipse platform. And apply them in the situation that multiple unmanned vehicles collaboratewith each other to reach a common goal. Simulation result shows that agent state prediction method,path planning method and conflict resolution method can work very well; they all verified theeffectiveness of models and algorithms designed in this paper.In this paper, we give a novel idea that combine prediction with improved partial planning methodwhich increases agent’s intelligence and reaction rate. Meanwhile, give a conflict resolution modelwhich improves agent’s conflict solving ability and design core negotiation algorithm with anevaluation method.