NPC Vehicular A.I. Driving Model using Statecharts

Emulating realistic open world NPC vehicles has been problematic for many years. NPC also known as Non-Playable Characters in a world help to add a level of realism and believability to a virtual world, but when they go wonky it can sometimes ruin an experience. This is typically handled by a decision tree or a behavior map which both control how how the NPC may act given a particular situation, things like an enemy being alerted and investigating a noise. The problem with these two different implementations is that they can become very complex and difficult to debug if they start going “off the rails” so to speak which makes hard to fix and usually get left in as a random player experience. Statecharts however allow for a more controlled Artificial Intelligence or A.I. by breaking down the NPC into different finite states restricting the character from doing anything else at the time, making repairs a little easier should they arise due to the smaller scope of possible bugged code that is handling the character in a given state. Statecharts in of themselves are easier to debug due to most every action only being possible in a specific state, which can be thought of as a garage door being in one of many states but usually only one at a time such as the door can’t both be open and closed, but if the door won’t open we know to look at the code responsible for opening the door in the opening state.