Page 1 of 1

An interesting university AI project

Posted: 29 Jun 2005, 08:22
by el_muchacho
NERO is the result of a joint project between the Digital Media Collaboratory (DMC) and the neuroevolution group at the Department of Computer Sciences at the University of Texas at Austin (UTCS). NERO utilizes state-of-the-art academic artificial intelligence research in order to demonstrate its effectiveness in a game environment. The game uses AI to allow simulated robotic agents to cope with changing environments and situations, and form adaptive tactical solutions. The end result is a game that adapts itself to the strategies desired by the player, while still allowing the AI-controlled entities to operate as autonomous agents. NERO introduces a new genre of video game that is only possible with machine learning technology: The player takes the role of trainer and teaches a team of novice soldiers the skills for battle. When sufficient skills have been acquired, players can take their teams to battle mode where their skills are tested against teams trained by other players.
http://nn.cs.utexas.edu/NERO/index.php

It would be interesting to adapt this to Spring. 8)

Posted: 29 Jun 2005, 09:10
by [K.B.] Napalm Cobra
Go on then.

Posted: 29 Jun 2005, 09:53
by AF
oo sounds system resource intensive. Eitherway I'd rather pummel the AI into the ground than carefully train it to pummel me.

Eitherway I shall download, read, and adapt. See if you could get any of this academic research they talk about?

Posted: 29 Jun 2005, 12:52
by Min3mat
coolness...if you could make this i don't think you would ever have to program a AI again...it would adapt, wow this would be awesome but seems a bit out of spring's league :(

Posted: 01 Jul 2005, 12:29
by AF
Now I've started reading stuff on this, NEAT looks promising, providing such things as effective ways of controlling resource collectors and scouts, and adaptive backup strategem for fi ever the AI's preprogrammed tactical agents arent capable of dealing with the situation.

http://nn.cs.utexas.edu/area-view.php?R ... D(Areas)=7

Lots and lots of neuroevolution papers, very interesting.

http://nn.cs.utexas.edu/soft-list.php

And numerous implementations of Neural networks including sourcecode to the NEAT implementation in delphi, java (JNEAT & ANJI), c#, C++, matlab among other things.

lol, I should ahve realised university research groups where the best place to find these things. Although it does make em realise that an adaptive AI as described by me in the original SSAI thread is possible, it neednt have the great swathes of data on what unit is bets against which and how likely is it to appear, although such an AI would take a long time to program and a long time to develop to a stage where it can put up a decent fight.

Posted: 07 Jul 2005, 07:35
by High Caliber
And then if we could adapt it to upload/download new strategies to/from the Spring server...

Posted: 07 Jul 2005, 09:46
by AF
Yah redownload and upload different generations and collect dominant strategies.

However downloading and uploading entire generations though accelerating the development process might be a little unprecedented, and may leave the AI vulnerable to easy attacks which may take further generations to eliminate or adapt to.