November 08, 2005

Great Expectation-Formations

I’ve been thinking a lot lately (and by “lately”, I mean “for the past 5 years or so”) about AI and expectation-formation. One of my original posts (on the FSM tyranny) dealt peripherally with the subject, discussing “Occupancy Maps”, which I spun at the time I was writing my thesis as a form of location-expectation (and which is the subject of my article for AIGPW3).

Clearly I talk a lot about the Media Lab’s Synthetic Characters group, mostly because I think we were onto a couple of really cool/important ideas there, that unfortunately never get explored as much as they could/should have (damn you, graduation). Expectation-formation was one of those subjects.

(For those who don’t want to wade through the nonsense below, there is an important question for game developers at the bottom of this post.)

Continue reading "Great Expectation-Formations" »

October 25, 2005

A New Beginning...

Greetings, my AI brothers. And apologies for the slow reaction to the ridiculous spam problem. As you'll see we have a new look, thanks to our recent upgrade to MovableType 3.2, featuring integrated spam control. Keyword-based spam control. Which does unfortunately mean that any valid threads containing references to a certain card-game of which Texas Hold'em is a variant, or to the establishments where one might go to play said game, OR to the male genitalia and any enhancement thereof, will be immediately deleted. Pity!

I've gone through and deleted all 12,000 or so spam comments -- unfortunately removing some real comments on our older entries in the process -- and set up a couple keyword filters. So god-willing our spam problems are at an end.

October 18, 2005

Dealing with spam?

I took the liberty of closing the comment sections on all the old posts, since they were getting seriously clogged with spam. I also tried mass deleting of spam on a couple of posts, but I think the ridiculous volume of spam overloaded things since it gives an internal error when "rebuilding" the files for the site.

Short of installing some plugins to screen comments (Damian?), I think we can make this blog usable again if the author of each post just makes sure to manually close out comments after a few days and manually deletes spam during that time.

Any thoughts?

September 01, 2005

Now that I'm rolling off of The Suffering: Ties That Bind and have my AIGPW3 article out the door, I figure it's time to make an intro post with obligatory bio:

Before getting into the game industry, I got my MS in CS at the University of Utah, where I once had the pleasure of flipping through Gouraud's and Phong's original dissertations in the library. I was then involved in a financially unsuccessful start-up that resulted in a nonetheless fun shareware game, StixWorld. That led me to Surreal Software where I've been for almost as many years as Duke Nukem Forever has been in development, developing AI for Drakan:OOTF, Drakan:TAG, LOTR:FOTR, The Suffering, and The Suffering: TTB. I presented a paper (slides) on the AI architecture used in FOTR and both Suffering games at the Challenges in Game AI workshop last year.

Now that I have time to start looking towards AI for next-gen consoles, I'm thinking that one of the biggest impacts will be the ability to have so many more characters on screen. On the PS2 and XBox, rendering and animation were enough of a bottleneck that we were limited to a small number of NPCs. But increasing the numbers of NPCs is more than just a quantitative change. As crowd density increases, navigation becomes a fundamentally different problem and any N^2 algorithms we've been using will fall apart. These changes will have effects rippling up into the AI decision making--it will be much less reasonable for an NPC to pick an "unoccupied" position, find a path, and then use simple steering behaviors to get to that exact spot, ignoring any possible crowd movements that might happen in the mean time.

That will keep us busy adapting for next-gen consoles and maybe another generation or two after that, but according to the crowd panic researchers, we're shooting for the fixed target maximum density of 10 people per m^2. Ouch.

I wonder how many scripters Carmack will have to hire then?