An Agent Based Simulation Environment for Public Order Management Training

This paper 1 discusses an object-oriented discrete-state simulation environment to support the creation and execution of agent based simulations. The system allows the graphical speciication and monitoring of simulations where interactions between agents and their environment are used to model real world systems. The exibility and clarity of operation provided by graphical tools enabled a rapid prototyping approach to be adopted; which is particularly important where the domain to be modelled is not well understood. A prototype encompassing these ideas has been built to assist in training police oocers in the techniques necessary to manage public order incidents. The trainee attempts to manage a crowd (march) of interacting agents moving around a simulated world drawn from digital maps. The map data is augmented with additional details relevant for the simulation (eg. premises information). The graphical interface of the prototype uses object-oriented techniques to control the aspects of the simulation which are visible to the trainee. The diiculty of the task may be adjusted by altering the amount of information available about the state of the simulation and by scaling the incident size. The development environment assisted in creating a simulation for this poorly understood domain, and by supporting an iterative design cycle which has lead to a realistic well tuned model.