Research On Path Planning And Collision Detection Method In Population Evacuation Simulation

In recent years, fires disaster and terrorist attacks which happened frequently have causedhuge property damage or casualties if people did not evacuate to safe area in limited time. Thecrowd how to reach evacuation shelter as soon as possible when emergencies occur has becomehot issue. Simulating and evaluating the movement of the crowd in emergency situations canprovide references for designing building and making emergency plans. It is difficult toaccurately describe evacuation movement with mathematical models because individual isaffected not only by its own psychological factors also the surrounding environment. Swarmintelligence algorithm is based on the group behavior in nature, such as bees, ants, birds, fish.Those swarm intelligence algorithms have the advantages of distribution, simplicity andself-organization than other models. Therefore, swarm intelligence algorithms provides newideas to simulate the behavior of crowd evacuation.Glowworm swarm optimization (GSO) algorithm is an emerging intelligent algorithm inSwarm Intelligence. The inspiration of the GSO algorithm is derived from group behaviors ofglowworms in nature. In glowworm model, each glowworm has sentient range which is similarwith the vision of person. The swarm of glowworms shows tendency congregation andself-organization, which is similar with the behavior of individuals in the process of evacuation.Therefore, the GSO algorithm is utilized for path planning in evacuation simulation in this paper.However, in most cases of evacuation simulation, particles are always lack of intelligence, whichresults in that they are not capable of detecting the change of environment self-adaptively andcollide with obstacles in the end. Thus in this paper, an effective collision detection algorithm isproposed in order to detect and avoid obstacles in time as well as promote the visualization ofevacuation simulation.The main innovations of this paper are listed below.1. The GSO algorithm has defects of slow convergence speed and low accuracy in solvingcomplex optimization problems. For this problem, it is improved by introducing the updatingmechanism of particle swarm optimization algorithm in which the global optimum solution isconsidered when updating position for each particle in each iteration. It is beneficial for thepromotion of solving complex problems.2. Propose a collision detection algorithm based on Cellular Automata. In this algorithm,the evacuation scene is compared to cellular space and then divided into many grids. Whether agrid is occupied or not is evaluated by obstacles’ positions. After that, collision avoidance ruleswill be to avoid collision with obstacles and improve reality of evacuation simulation.3. In the procedure of evacuation simulation, many individuals always arrive at the safe areathrough many exits. Therefore, an algorithm, which is derived from artificial bee colony algorithm, for the split of crowd is proposed in this paper in order to guarantee that there is nocongestion in each exit. In this algorithm, congestion degree and distance are taken intoconsideration simultaneously to promote the effectiveness of evacuation simulation. It canprovide significant basis for the design of buildings and reasonable evacuation scheme by takingadvantage of analyzing the relation between evaluation time and the population around each exitas well as the size of crowd.4. By utilizing VS2003, ACIS, HOOPS and SQL Server2005, an evacuation simulationsystem based on GSO algorithm is designed to simulate the process of crowd evacuation. Thealgorithms above are also applied in the scene of chemical plant in which each individual canarrive at the safe area without collision. In the end, the data of scene and path is import intoMaya in order to generate animation with good visualization.