Racing AI

Started by
4 comments, last by mantra 21 years, 2 months ago
I am about to start a thesis to investigate the use of AI in racing games and am looking to build up a collection of information from fellow AI programming enthuiasts. The first thing I am investigating is the current state of racing AI, i.e. what techniques are currently being used and why - FSM''s, neural networks?, what techniques have been abandoned, etc. I have many articles on game AI in general but would like to filter these down to racing specific AI. Any help would be much appreciated.
Advertisement
quote:Original post by mantra
I am about to start a thesis to investigate the use of AI in racing games and am looking to build up a collection of information from fellow AI programming enthuiasts.

The first thing I am investigating is the current state of racing AI, i.e. what techniques are currently being used and why - FSM''s, neural networks?, what techniques have been abandoned, etc. I have many articles on game AI in general but would like to filter these down to racing specific AI.

Any help would be much appreciated.


Did you try google? I got the following when I did:

http://www.ebc.ee/~mremm/rars/rars.htm
http://rars.sourceforge.net
http://www.gamasutra.com/features/20010124/adzima_04.htm
http://www.gamasutra.com/features/19970905/ng_01.htm
http://www.gamasutra.com/features/20010324/brown_pfv.htm

I''ve also done the car driver AI for a Trans Am racing game and a Drag Racing game. However, I doubt you will find an article or paper that discusses the elusive "why" a technique was selected or not.

Eric
Thanks for the links, when I tried google I just got a load of reviews talking about how good/bad the AI is in games such as Gran Turisimo, etc.

Check out this interview with the AI programmer of "Colin McRae Rally 2":

http://www.generation5.org/hannan.shtml

You are also welcome to check my NeuroDriver project which demonstrates race car control by a neural-network.

http://gpdev.net/NeuroDriver.html



Here are two interesting papers.

Learning Coordinated Behaviors for Control of a Simulated Race Car
Learning to Race: Experiments with a Simulated Race Car

There are also a couple of chapters in the AI Wisdom book that describe a fairly typical way of implementing racing ai.
But sadly though, for most games I think a racing line is just mapped out manually before-hand and all the cars try to stick to it. Maybe I''m just being too cynical.

''Your theory of a donut shaped universe is intriguing Homer'' - Stephen Hawking

This topic is closed to new replies.

Advertisement