Agent-based Simulation System For Multi-robot Cooperation

The conception of agent is derived from DAI(Distributed Artificial Intelligence). The relevant theory and technology, specially MAS(Multi-Agent System) and AOP(Agent Oriented Programming), has been applied to the fields of Knowledge Representation, DAI, Internet Data Mining etc, which have closely-related with AI(Artificial Intelligence). MRS(Multi-Robotic System) based on MAS is a concrete application of Multi-Agent conception. This paper will study Multi-Robotic Cooperative Technology on the basis of MAS.First, we research the structure model of Agent and the technology of MAS step by step. Second, an architecture model is proposed which not only is appropriate for an agent alone to undertake tasks but also support cooperative requirement. At the same time implementation flow that multiple robots accomplish cooperatively a task is described in detail. Third, we take example for a simulation model about robots patrolling street district and put emphasis on path planning of Multi-Robotic Cooperative Technology. According to the speciality of simulation system, we bring Dynamic Programming, Dijkstra algorithm and other graph theory of operation research into the resolution of route optimization and path collision.Finally, we carry out an experiment to simulate multiple robots to patrol street district in virtual reality environment. The results prove that the route optimization strategy in view of dynamic programming and graph theory we propose is reasonable.