In The Truman Show: Generating Dynamic Scenarios in a Driving Simulator

Abstract

All the devices, animals, and people make their decisions based on what you're doing, but you don't know it or even notice it. Your world is that of Truman Burbank, from the 1998 movie The Truman Show. With this idea in mind, we've taken the movie metaphor to implement a prototype simulation system where the user steps into Truman's shoes. The set of our "movie" is a driving simulator, and the user is learning to drive a car. During the driving lessons, users drive in a virtual world that lets them experience all kinds of traffic scenarios. The system generates the scenarios with the student as the focal point, and the other traffic entities respond to the student's behavior, without the student noticing. To control the traffic scenarios and make them more effective, our prototype employs an agent-based framework. In this framework, each entity in the simulator is an actor agent playing a role. The prototype also includes a hierarchy of directors that directs the main action and the behind-the-scenes activity. The advantage of the movie metaphor is that it helps separate scenario description from scenario playing. The agents can read their required information from a script and perform their actions based on that information. Using this framework lets us build software that's extensible, maintainable, and easy to understand