Urban pedestrian simulation

Abstract

In this Final Year Project we are simulating the flow of pedestrians in and around streets or sites that were purposely designed by the user. It can predict the flow of pedestrians should certain areas or pedestrian walkways close. The simulation can be used for crowd dynamics by councils, civil engineers and urban planning companies when planning events that may block or redirect the flow of pedestrian traffic. The simulation is divided into two main areas, the area representing the city or site, and the pedestrian movement. The A* pathfinding algorithm gives a best or shortest path for the pedestrians to follow from the start to the target location which is selected by the user. Each pedestrian is an independent individual that will roam about in the given area. No external control is given by any user during the simulation but the deviations and collision detection are automatically done by the simulation itself. When running the simulation the pedestrians will move from the start to the target location avoiding other pedestrians on their way using the collision avoidance algorithm by implementing heuristics that approximate behaviour and priority rules. These rules will solve the collision problems in the crowd simulation. When a pedestrian deviates from its original path, after avoiding the other pedestrian it will eventually return to its original path. There are mainly three different types of collision that a pedestrian can encounter, which are Toward collision (when two pedestrians walk toward each other and collide), Away collision (when two pedestrians are walking in the same direction but at different speeds and collide), and Glancing collision (when two pedestrians collide in a side-on collision).