Advertisement

AI Dissertation Ideas (RTS AI, preferably)

Started by July 04, 2008 05:43 AM
10 comments, last by Timkin 16 years, 4 months ago
I need some ideas for my final year university dissertation. I'd prefer to do something that's related to RTS AI, but I'm not overly fussed really. Something related to machine learning, pathfinding, that sort of thing? Any other topics? Any new research ideas? Cheers
You can do Card AI. There's a thread on AI for Magic the Gathering and it seems the AI to get that running is difficult from what I read.

Beginner in Game Development?  Read here. And read here.

 

Advertisement
If you're interested in RTS AI, why not look into strategic advisors as a form of 'decision support'?
A suggestion, you don't have to come up with some revolutionary new method. You could focus on optimizing existing methods. Also, keep your dissertation focus very specific. AI can slow performance much more than rendering or collision. AI optimization is especially in RTS since you have hundreds of units and each unit decision often depends on those units around it.

Maybe take some existing method and see how it scales to 1000+ units, then try to optimize. That would not only be useful for RTS, but also simulations.

Let us know what you come up with. =)
Quote: Original post by Timkin
If you're interested in RTS AI, why not look into strategic advisors as a form of 'decision support'?


I don't quite understand what you mean. Do you mean to create an AI that advises the player on what to do?

Quote: Original post by FippyDarkPaw
A suggestion, you don't have to come up with some revolutionary new method. You could focus on optimizing existing methods. Also, keep your dissertation focus very specific. AI can slow performance much more than rendering or collision. AI optimization is especially in RTS since you have hundreds of units and each unit decision often depends on those units around it.

Maybe take some existing method and see how it scales to 1000+ units, then try to optimize. That would not only be useful for RTS, but also simulations.

Let us know what you come up with. =)


Sounds interesting. I did have a thought on maybe creating a more "closer to human" AI, ie not totally scripted. The AI wouldn't have waypoints for pathfinding, it wouldn't know the map so it'd have to search around, and it wouldn't know where the enemy player would be. The only things it would know is what an average player would know: the various units, what counters what, how much gold required to build something, things like that. Then in order to counter the player it would have to scout them out (I wonder if it's possible for the AI to generate its own waypoints by scouting? That way they wouldn't have to keep calculating which way to go to get to a certain point), see what units they do have and can then build units in order to counter.

Keep the ideas coming :)
If you want to code AI for an AI player, read "The Art of War" by SunTzu, it's a must-have for RTS AI developers (pst.. the book is 2600 years old, but it's still very valid and very used).

Hope this helps
Dark Sylinc
Advertisement
Quote: Original post by Matias Goldberg
If you want to code AI for an AI player, read "The Art of War" by SunTzu, it's a must-have for RTS AI developers (pst.. the book is 2600 years old, but it's still very valid and very used).


Thanks for pointing this out. I had been considering getting it sometime anyway, but since I'm planning an RTS it's the perfect excuse to overstep my book budget, again [grin]

As for the OP, I think a 'more "closer to human" AI' would make for a very interesting subject. A bit of a novel idea might be to do a kind of Turing test with a game AI, but I'm not up to speed on AI publications so this may have been done to death already [smile]
Rim van Wersch [ MDXInfo ] [ XNAInfo ] [ YouTube ] - Do yourself a favor and bookmark this excellent free online D3D/shader book!
Quote: Original post by remigius
Thanks for pointing this out. I had been considering getting it sometime anyway, but since I'm planning an RTS it's the perfect excuse to overstep my book budget, again [grin]

You're welcome. If I'm not wrong it's public domain and it's price is low or can be downloaded from the net.
You surely will find this article very usefull

Cheers
Dark Sylinc

How about a politician simulator? You could have a speech module, a mudslinging module, a brown-nosing module, etc...
Go grab the AI source for Civ 4. It's floating around the Net someplace. (Legit)

Dave Mark - President and Lead Designer of Intrinsic Algorithm LLC
Professional consultant on game AI, mathematical modeling, simulation modeling
Co-founder and 10 year advisor of the GDC AI Summit
Author of the book, Behavioral Mathematics for Game AI
Blogs I write:
IA News - What's happening at IA | IA on AI - AI news and notes | Post-Play'em - Observations on AI of games I play

"Reducing the world to mathematical equations!"

This topic is closed to new replies.

Advertisement