Academic research papers on AI in RTS games

[NeuronExMachina]

I'm a grad student that does AI-related research, and I recently stumbled across TA Spring. Being a fan of the original TA, this excited me tremendously. I haven't actually been able to run TA Spring yet (my personal computer is out of commission), but I have been skimming through the wiki and forums about the various AIs people have implemented for the game. I'm particularly intrigued by the Group AIs which can support human players, as it reminds me of Garry Kasparov's Advanced Chess proposal, where AIs and computers form cooperative teams.

In any case, I thought I'd share a few links to relevant research papers and sites, in the hope that some of you might find them interesting/useful. I've also posted the following to the wiki.

* Forbus et al 2001, "How qualitative spatial reasoning can improve strategy game AIs"

* Michael Buro's lab: Much of his lab's research is focused on "real-time planning and learning AI with applications to RTS games." They've actually created an open-source RTS called ORTS, in order to experiment with various AI techniques. With any luck, they might even be persuaded to switch their focus to TA Spring, or perhaps try implementing some of their algorithms. Also, it might be worthwhile to use elements of ORTS in TA Spring.
**M. Chung, M. Buro, and J. Schaeffer 2005, "Monte Carlo Planning in RTS Games"
**M. Buro 2004, "Call for AI Research in RTS Games": Gives a nice overview of the current state of AIs in RTS games
**M. Buro and T. Furtak, "RTS Games and Real-Time AI Research"

I'm not sure yet how useful such research will be right now, but hopefully as TA Spring proceeds aspects of such research will become more relevant.

[AF]

wow you rediscovered gabbas research! That first pdf has heavily influenced future plans in my AI projects and research a lot the last few months since gabba posted it. So far I've outlined the necessary plan to integrate path finding and such described by that first pdf into my next AI project and I intended on doing something similar to improve path finding in the summer.

The others I've come across but I don't remember reading them (looks at the mountain of research material one reads through and the even more immense amount stacked up yet to read).

However I hope you stay here and become productive or at least advise on those of us which paths would be the greenest. The last person who said he had already gotten experience and tried to help ended up leaving after an old fiasco reared its head again(i dare not speak of those things and I advise you leave the past alone lest you resurrect more than gabbas research).

*AF gives NeuronExMachina a cookie*

[Veylon]

Impressive!

I wasn't aware that any serious work was being done on real-time AI. I would like to see how a "real" AI would handle all the various confusions and mods in Spring.

[NeuronExMachina]

First off, I posted a few more paper links to the [http://taspring.clan-sy.com/wiki/AI#Relevant_research wiki], including a 164-page Master's thesis on using game trees to determine optimal unit placement and target selection in RTS games. There's also a neat paper on an extension of A* that deals with moving targets.

wow you rediscovered gabbas research! That first pdf has heavily influenced future plans in my AI projects and research a lot the last few months since gabba posted it. So far I've outlined the necessary plan to integrate path finding and such described by that first pdf into my next AI project and I intended on doing something similar to improve path finding in the summer.

Cool!

However I hope you stay here and become productive or at least advise on those of us which paths would be the greenest.

I'd love to help however I can, but being a grad student my primary focus is on my research, so I'm not sure how much time I can give. Also, my area of expertise is in computer vision and machine learning, which is probably of limited usefulness to you guys. Still, if there's ways I can competently advise, I'm glad to do so. :)

*AF gives NeuronExMachina a cookie*

Yum!

[NeuronExMachina]

[quote="NeuronExMachina"]First off, I posted a few more paper links to the [http://taspring.clan-sy.com/wiki/AI#Relevant_research wiki], including a 164-page Master's thesis on using game trees to determine optimal unit placement and target selection in RTS games. There's also a neat paper on an extension of A* that deals with moving targets.

Whoops, botched link:

http://taspring.clan-sy.com/wiki/AI#Relevant_research

[SinbadEV]

Groovey... my friend did some reasearch on Neural Networks Based Image Analysys for one of his courses... is that the kind of thing you are doing? I wonder if he ever got around to publishing it...

[NeuronExMachina]

Groovey... my friend did some reasearch on Neural Networks Based Image Analysys for one of his courses... is that the kind of thing you are doing?

A little, although I generally prefer using support-vector machines (SVMs) instead of neural nets, as they're theoretically more sound, perform much better, and aren't as prone to random-stuffs as neural nets are. Lately I've been doing most of my work in face recognition and face clustering.

[NeuronExMachina]

Another quick FYI... I just came across a course website for a class on "Topics on AI and Computer Game Programming":

http://www.cse.lehigh.edu/~munoz/CSE497/

I haven't had a chance to look through all the lecture slides yet, but it seems like quite an interesting read. Several of the lectures are dedicated to topics that come up in AI for RTS games.

[Dragon45]

Yay, more material to parse through...

Now if only i could be arsed to actually code


<---- Is a non-coding coder (its a paramount sin around here )