Implementation Walkthrough of a Homegrown “Abstract State Machine” Style System in a Commercial Sports Game

Abstract

When I began working at Sony Computer Entertainment of America in 2002, the AI system they were using was very dated. Over the next few years, I designed and developed an almost completely data driven system that has proven to be very powerful, extremely extensible, and designer friendly. This system uses a homegrown data structure, the use of which in many ways resembles the software method of using Abstract State Machines for decomposing complex logical constructs iteratively. This paper will provide an overview of the construction and usage of the system, as well as the pros and cons of this type of game AI engine.