A Domain-Independent Multiplayer Architecture for Training

In this paper the two issues we address in developing an architecture for training are flexibility and ease of scenario generation. Flexibility is the need to make the architecture domain independent by investigating how the logic (dictating the scenario behaviour) is linked to the simulation environment (where the rendering, networking, etc., occurs). For the second issue we consider how to automatically generate training scenarios from expert systems (ES). Our architecture shows how an intermediary “events space” can be used to achieve flexibility by separating the logic from the simulation environment. To show the training suitability of the architecture we have developed a scenario on top of a prototype of the architecture to train new police recruits how to investigate vehicle accidents.