Onwards with another week of Game AI news in our fittingly named Game AI Roundup here at AiGameDev.com. Be sure to swing by The Game AI Forums for some stimulating discussion, and also don’t forget Alex’s Twitter account his random thoughts…
This roundup was written by Andrew Armstrong (site). If you have any news or tips for next week, be sure to email them in to <editors at AiGameDev.com>. Remember there’s a mini-blog over at news.AiGameDev.com (RSS) with game AI news from the web as it happens.
Richard Bull gives an explanation of the battle tactics the AI uses in Empire: Total War, showcasing the AI using certain tactics in the battle. Of course, he doesn’t lose, so obviously they’ve still got a way to go ;)
AI Changes In Total War
In an interview with Creative Assembly’s Kieran Brigden at Gamasutra, we find out about how the AI has changed from a state based system in Rome: Total War, to a more comprehensive and useful goal-based AI in Empire: Total War.
“Certainly that’s typified most in our AI system. We moved from a state-by-state AI system, which was like a chess game — do A then B then C — over to a goal-oriented action planner, which was essentially a very new kind of academic technique. It’s a constant list of priorities versus resources available. Those priorities are shifted like Post-It notes on somebody’s desk: “The number one thing I must do today is this.” Depending how increasingly important those tasks become, they move up and down the queue. Then, the resources are allocated to identify those tasks.
What that means is you get reactive gameplay. It makes a massive difference. In something like Rome, for example, the AI will say, “I have to break down the door or storm the wall to get into the city.” That is state A. To get to state B, where it’s fighting in the city, “I have to achieve that.” But if you can interrupt it anywhere along that path, it never reaches the next state. It just keeps trying to get there.
Now, you say, “All right, I’m going to outflank the enemy and go around the back, and he’s charging me down the center.” Then, suddenly the AI goes, “My general is in danger.” So, “protect the general” goes from priority six to priority one. Then, all of the resources are moved around doing that, and then the battle’s taking on a very different dynamic.”
Epic Frontiers Dev Blog
The Epic Frontiers Blog is explaining how the MMORPG is developing the NPC AI, notably NPC conversation systems. Might well be worth keeping an eye on developments of the game, since good AI R&D can come from anywhere and this seems to be on the right lines.
“One of the features I’m touting is the ability to have conversations (actual two-way conversations) with NPCs in Epic Frontiers. It’s not a new thing, and games like Zork have made deep interaction a priority from well over twenty years ago. Unfortunately, with the AAA game market leaning heavily on cutting-edge graphics, they have lost this ability to interact with NPCs outside of combat, and their ability to fund R&D; into this realm has become stunted. Of course, being an indie without a real budget to stay within, I have the advantage of not having anyone breath down my neck about addressing it by a specific date.”
GTA IV’s Pedestrian Dialogue
A quick quote from Dan Houser of Rockstar on how the AI is perceived to be better in each GTA iteration with more behaviors and lines being spoken.
“When we were doing production, we’d just get our friends to come in and do voices, about 10 lines each. And then by “IV” it split, but the big ones had hundreds of lines each but you keep adding new bits of A.I. or new behaviors and you want them to speak, because that’s what sells the A.I.”
— Dan Houser
“…you know how hard it is to convince someone that games don’t simply have to be about…”
Phil Carlisle writes about planning his AI GDC session, I hope he provides some insightful things to listen to at GDC!
“…you know how hard it is to convince someone that games dont simply have to be about killing?
Given that the Wii has expanded the market into new areas, you would think more experimentation in what is arguably the most profitable game segment ever (the simulation) you would expect more. But visit any game trailer site and look at the number of simulation games that aren’t based on war and you’ll see where I’m coming from.
Imagine trying to pitch a game involving the creation of the AI equivalent of a romantic comedy in the vien of “four weddings and a funeral” and you’ll maybe see my problem. I think games have the potential for social and emotional simulation that may well be a huge underinvestigated role for games as a medium, yet I’m just not entirely sure I can create something convincing enough that people will see what that might look like.”
Tutorial For AI In Flash Games
For anyone implementing Flash games with enemies, this tutorial might be a good starting point for adding AI to the mix.
All games need enemies…. okay, maybe not tetris. But if you’re in some game where you fly a spaceship it is more than likely that there’s going to be an enemy or two. So in this tutorial we’re going to learn to make our enemy class, making enemies randomly appear on the stage and move, and make them fire bullets at us.
Gamasutra publishes an article from a previous Game Developer with Tim Smith from THQ commenting on the AI in brawlers - such as God of War - with an in depth look at a variety of different ones. A good resource for researching the area, with some technical information and much design information:
AI attacks often fail to collide with the character even with no player input beyond blocking to avoid them (not moving, not dodging, not attacking), which feels odd. Not sure why this is happening-it seems like a bug. My best guess is it’s a sync issue with the block animation-block makes the player character crouch down a bit, and the attacks appear to go over his head.
Autonomous AI in student-made Feist
A notable technical achivement according to the makers of Feist was the AI, which is impressive given it’s a student made game. No more excuses from the big boys that AI is too difficult!
“GCG: Tell us one interesting technical achievement you had in making the game.
AS and FF: Because we wanted to be flexible during development, large parts of the mechanics are very adaptable and allow integration of further elements. The biggest part of that is probably the AI that is fully physics-driven and autonomous.
The player also inherits a lot of the AI code and therefore is also much more part of the game mechanics than in other games. This allows us to create additional characters and objects that interact dynamically with minimal work.”
Useful AI Resources
Some pages to bookmark - on the use of rand(), looking at the speed vs. randomness of different routines, and encouraging you to not use the default C++ library function, and a good page for general artificial intelligence links from around the web.
That’d be nice; Better A.I.
Finally, we have a rant about the progress of AI in relation to other areas of game development.
“While those have come many miles since their first incarnation, the A.I.’s progress seems to have all the progress of Sisyphus pushing his boulder up a mountain. Sure A.I. now is far superior than that of yesteryear but it’s progress doesn’t seem to be as important to game developers. Everytime a developer raves of their A.I. improvements it turns out to be little else than a more refined version of the marines from Half Life. Of course this is an unfair generalisation, but it still seems that the dream of facing off against A.I. that could actually challenge you (without cheating, i.e. catch up A.I., see mario kart and nearly every racing game) has been lost in favour of online play. Only RTS games seem to really get a clever A.I. (Galactic Civilization 2)while us dumb FPS gamers are stuck with A.I. that still sometimes run straight off cliffs or into fire. So I have devised a shopping list of things I would like to see used more in games.”