Game AI Mega-Roundup Week #49-50 ’08: 11 Stories, 1 Video, 1 Demo

Novack on December 15, 2008

This column at is dedicated to rounding up smart links from the web relating to artificial intelligence and game development. This week, as always, there are some good articles and blog posts for you to read. Remember, there’s also lots of great content to be found in the forums here! (All you have to do is introduce yourself.) Also don’t forget the Twitter account for random thoughts!

This post is brought to you mostly by Marcos Novacovsky (aka “Novack”). If you have any news or tips for next week, be sure to email them in to editors at Remember there’s a mini-blog over at (RSS) with game AI news from the web as it happens.

Random Steering - 7 Components for a Toolkit

A very good article on the always fascinating place to where Craig Reynolds opened the door years ago: steering behaviors and emergent activities.

“Random steering is often a useful for simulating interesting steering motion. In this post we look at components that make up a random steering toolkit. These can be combined in various ways to get agents to move in interesting ways.”

ICIDS’08: Comments

On the Monique de Haas blog, we found an interesting post about the ICIDS’08 ICIDS08 conference, developed on past november.

I am very pleased to have attended the ICIDS08 conference in Erfurt, Germany. The conference brings together writers of interactive stories with the computational AI (Artificial Intelligence) scientists of story engines and applications. For more then 20 years there has been research in the domain of interactive storytelling. One side of this domain is the AI; Artificial Intelligence domain. Here the goal is (and I am using a liberal explanation here) to dissect stories into abstract units that can ideally and intelligently create stories as systems. Progress has been difficult, due to the reality that stories are probably too human to be generated by computers. There is a lot of pre-conceived implicit knowledge at the level of the writer, (knowledge of the world, knowledge of people, creative human exploration) which is hard to dissect into abstract units.

Empire: Total War Developer Interview

Via PS3 News we found an interview to Kieran Brigden -Creative Assembly’s communications manager-, made by WorthPlaying early this month. Some comments on the game’s AI systems.

WP: You’ve changed how the AI works for Total War. Can you explain that a little bit?
KB: Yeah. That’s a massive overhaul. We scrapped the AI system and wrote it again from the start. We used to use what was called a state-based AI approach, which is very much like a chess game. It does A, then B, then C. Now we’re using a system of goal-oriented action planning, or GOAP, which basically means that the AI is constantly looking at the status of forces on the battlefield, which of course, is resources, and it’s got a list of jobs. Like, imagine you’ve got a list of post-its on your desk, and this is the number one thing I’ve got to do today, and this is the second. It moves those things around based on what’s happening. So let’s say its general is safely at the back of its army, right? And you’re fighting very well and suddenly you outflank him. You come right around from the back, and there’s a real chance that its general is in danger. The post-it note that says “Protected the general” goes from seventh to first, and then it looks at its resources and goes, “I’ve got artillery crews, I’ve got cavalry, I’ve got infantry. How do I achieve that objective now?” So what that means is you get a lot more dynamic backward and forward gameplay with the AI reacting to the things you’re doing in real time, rather than going “I’ve got to do A, then I’ve got to do B,” and if you interrupt it anywhere in that cycle, it doesn’t get to the next step. Now, it’s constantly looking at what its priorities should be to achieve its objectives and basing its movements and its actions around that exact thing.

WP: So it’s more like fighting a real human.
KB: Really. Yeah, yeah. Much, much more so. In some cases, more difficult. Fighting dumb people is easier than fighting the AI, from some perspectives.

Emotional Virtual Faces

ScienceDaily brought us news about a research team at the Balearic Islands University, developed a system that generates virtual faces able to display emotions and moods according to personality traits.

The researchers believe that this model could be applied in both educational environments (virtual tutors and presenters with personality traits) and in video game characters or interactive stories that have their own emotional motor.

HowStuffWorks “Intelligent Virtual Agents”

Novamente’s “intelligent virtual agents” would use online games and virtual worlds as a development zone, where they will grow, learn and develop by interacting with humans. The company said that it will start by creating virtual pets that become smarter as they interact with their (human-controlled) avatar owners. (An avatar is the character or virtual representation of a player in a virtual world.) More complex artificially controlled animals and avatars are expected to follow.

Behind the Scenes with Killzone 2

Rob Heald, among other personnel from Guerrilla Games, commented on the upcoming Killzone 2, in a “behind the scenes” interview (Alex posted the video yesterday). He said this, regarding the game’s AI:

A separate team worked on the artificial intelligence that powers the assorted Helghast enemies as well as the buddies who help the player through the journey. Rob Heald, AI designer, said the PS3 offers each AI a much wider range of on-the-fly options.

“The advanced processing power results in more intelligent AI across the board,” said Helad. “As a result, players will see a greater range of animations. Each individual enemy is able to do things like jump over objects, use doors and ladders and have intelligent reactions to dynamic things that happen on the battlefield.”

Long Snaps: The Sports Game A.I. Frontier published an interesting article on the AI on sport games.

One trend I’m noticing is that while graphics are improving to near photo-realism with each new console generation, artificial intelligence still lags way behind. We see it repeatedly in almost every sports game from 2008. Almost all of the sports games have incredible, jaw-dropping graphic quality, while the AI of the CPU has done nothing to make us believe we are playing against an adaptable human opponent.

Call for Papers: IJCAI-09

The theme of the 21st International Joint Conference on Artificial Intelligence is “The Interdisciplinary Reach of Artificial Intelligence,” with a focus on the broad impact of artificial intelligence on science, engineering, medicine, social sciences, arts and humanities.

The IJCAI-09 Program Committee invites submissions of technical papers for IJCAI-09, to be held in Pasadena, CA, USA, July 11-17, 2009. Submissions are invited on significant, original, and previously unpublished research on all aspects of artificial intelligence.

A* Search: A Nice Tool for AI Game Developers

Another article, this time on the blog Nadim is Simple, on one of the most ubiquitous topics in game AI: pathfinding, and more specifically A*.

If a game is not developed with Artificial Intelligence (AI) then the characters in that game seem brainless which we don’t like. We want games seem realistic although the whole thing is virtual. The most important problem in AI is searching. Let me describe with an example. Suppose in an action game there is a maze like one shown in the image below. A man is walking in this maze and there is a bug that tries to attack that man. Now how can that bug reach to that man? In brute-force fashion we can try all the possible paths between the bug and the man but in a large world brute-force search is not feasible which requires both huge CPU cycles and memory. Moreover in brute-force means the bug will not show any intelligence. Also the bug needs to avoid obstacles shown by black blocks.

Genetic Algorithm Evolving a Car on Bumpy Terrain

An awesome flash demo project featuring Genetic Algorithms.

“Hello, thanks for looking at my program. It’s not finished, so please forgive its inadequacies! I’ll try and find some time to make a better version, and release the code.”

GDC09: First Details of AI Summit

This year, GDC is hosting an AI Summit to take place on Monday and Tuesday, which regulars help organize and are participating in. Develop has more details:

Key speakers at the summit, which ‘promises to give attendees an inside look at key AI architectures and issues within successful commercial games’, include Soren Johnson of EA Maxis, Rockstar Leeds’ Brett Laming, and Crystal Dynamics’ Daniel Kline.

FEAR 2: Project Origin Talks AI

Rock, Paper, Shotgun posted an article commenting on a video (also posted last week by Alex) about the AI systems behind the upcoming title FEAR 2.

I wonder how we’ll look back on this period of development, once we’ve got AI in games that’s likely to talk us out of the gunfight and to join its uprising against their developers. As games make stumbling progression, inching ahead with improvements for enemy behavior, we’re still getting excited when we throw a grenade at them and they don’t pick it up and try to eat it. In the latest dev chat, below, Matt Rice and Nate Cleveland proudly explain that the AIs respond with a display of emotion when they catch on fire.

And good. These little steps are significant. I’m really excited to see if the AI will live up to the boasts in the game. But I do think we’ll look back and laugh.

Stay tuned next week for more smart links from around the web!

Discussion 0 Comments

If you'd like to add a comment or question on this page, simply log-in to the site. You can create an account from the sign-up page if necessary... It takes less than a minute!