Urban Road Traffic Simulation System Based On Agent

The content of text is concentrated on the assessing system for traffic road constructing in the city. On one respect it needs to appraise to design of up-coming construction of road facilities and to be forecasted to future traffic capability. On another respect it also needs to appraise to optimized approach of controlling traffic, which is already exist, and to be forecasted the approximate result for the tiny change of traffic facilities. One of the most effective methods is the traffic micro-simulation in the study.It' s an effective approach to partition comprehensive system using Agent-Oriented methodology, and it' s suitable for distributed environments. Agent expresses the real world actually more than object, so using Agent-Oriented methodology to implement traffic simulation is wise method.The main research of the text includes collision detection algorithm, collision-free path planning algorithm, and the implement of distributed simulation system, which are less involved in other literature about traffic simulation. The design of such kind of algorithms must base on the special system characteristic, balance between the accurate and the algorithm complexity with the other requests like effective and complicacy, and integrate other methodology. The collision of convex polygon is a very suitable abstract model for vehicle emulation, and a suitable situation for computational geometry method that has same low bound as intersection of line segments. The key of such optimize method is how accelerate algorithm to improve their efficiency under average meaning, because the probability of worst cases is very little, especially in traffic simulation. Collision-free path planning is the natural extension of collision detection, so they can take the similar way. Its main problem is how to raise efficiency under certain accuracyand tolerance. Distributed simulation system involves how to parallelize above-mentioned algorithms and , and how to imply the basic characteristic of Agent-Oriented system which alt least consist of autonomyNreactivity^social ability and pro-activeness. Difficult points of implement of Agent-Oriented system are complicatedness and parallelity, therefore it must be considered systematically how to simplify models and enhance performance.The collision detection algorithm and collision-avoiding algorithm in this study, which use are based on Voronoi diagram, not only accelerate the execution speed with substituting the Voronoi box for actual object , but also can get a lot relevant information about nearest objects, contribute to improving the artificial accuracy. There are some strategies that face the actual application. The realization of the simulation platform has adopted the Agent-Oriented thought, has set up effective message mechanism which solved the cooperation among Agents and suit traffic simulation, and also has designed the solution for agent' s mobility on the same hardware and software platform.