Crowd modelling and simulation using fine grid cellular automata

Crowd modelling and simulation have been used for different purposes such as evacuation and crowd safety inspection, performance improvement of crowd systems such as buildings and stadiums, animations, and finally training and serious games. Cellular automata model has been widely used in modelling crowd. In this model, each pedestrian is restricted to a single cell with the size of a pedestrian body. In this research, the model is enhanced using a method called fine grid cellular automata whereby smaller cells are used and a pedestrian body of various sizes may occupy several smaller cells. This model provides more realistic and accurate results and with this model it is possible to simulate pedestrians with different body sizes, speeds and specifications.

In this research, two varieties of fine grid cellular automata are proposed namely the straight line movement towards a target and circular movement around an object. With this model, it is also possible to simulate more accurately and realistically movement of wheelchairs and similar wheeled equipments such as push-chairs and shopping trolleys, and the effect of slope on the movement of the pedestrians. Therefore, the model is extended to support the simulation of a scenario involving a mixture of wheelchairs and pedestrians, and movement of pedestrians on slope. A model for simulation of movement of groups of pedestrians such as families and friends using cellular automata is also proposed. Instead of using a probabilistic model for the cellular automata transitions, an alternative approach using machine learning is also proposed.

A simulation platform is created in order to implement, evaluate and validate the proposed models. It has been shown that the fine grid cellular automata model is able to produce smoother movements and more accurate resemblance of the empirical data based on speed-density graphs and fro some aspects based on visual observation. Overall, the proposed fine grid automata model has been shown to produce more accurate and realistic results including its varieties and extensions.

There will also be a short presentation on the School of Computer Sciences, Universiti Sains Malaysia and another presentation on an overview of Long Term Research Grant on Crowd Comfort and Safety in the Hajj Pilgrimage.