Roundup
l4d_2_resized

Game AI Roundup Week #7 2009: 6 Stories, 2 Videos, 1 Paper, 1 Book

Andrew Armstrong on February 17, 2009

Another week flies by and we have another Game AI Roundup here at AiGameDev.com. With us catching up, we have less news then our last updates, but still some interesting pieces. 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.

Maximizing Player Satisfaction

Assistant professor, researcher and long-time reader Georgios Yannakakis sent in a link about his recent work at CIG ‘08. Here’s the gossip:

“Below you may find a link of a paper I recently wrote (in proceedings of the IEEE Computational Intelligence and Games Symposium). It is about a rather efficient adaptation mechanism used for increasing the level of fun of children playing augmented reality games. The game adjusts itself in real-time for the entertainment value of the child to be improved.

The entertainment model is a neural network function that maps between player characteristics (e.g. response time) and reported (expressed) fun. The neural network outputs an entertainment value that predicts children’s level of satisfaction with an accuracy of approx. 80%. The same neuro-evolution methodology has been used quite successfully to simple screen-based games like Pac-Man during my PhD in Edinburgh.”

Real-time Adaptation of Augmented-Reality Games for Optimizing Player Satisfaction
Georgios N. Yannakakis and John Hallam
Download PDF

Left 4 Dead Developers Commentary

Left 4 Dead 2

Excerpts from Left 4 Dead’s developers commentary (They say “Developer’s” - who edits GamesRadar anyway?) for those without the game. Some interesting design information based on the AI:

They already had the idea for co-op, Michael Booth was the guy who had done the AI for Counter-Strike bots so he really wanted to have AI at the forefront of the game. So, even when our companies were separate, we started working with them on the game. Gabe Newell was sitting with me and Erik Wolpaw at lunch and we were just babbling about the game to him. An hour later, he sent out an email that said, ‘Hi Michael. Chet and Eric are going to help you with Counter-Strike’. We asked how much we could do, and he just said to work as much as we liked on it.

SOM Neural Networks

Alex pointed out this interesting article on self organizing map neural networks, updated this week with new points. Here’s some information on the technique:

You have a number of neurons usually arranged in a 2D or 3D grid, each neuron has an associated weight vector. The input, which is a vector of the same size as the weight vector is connected to each neuron. The natural association is that each vector is a point in an n-dimensional space. By providing, during the learning phase, an uniformly distributed number of n-points to the network the neurons will arrange themself uniformly in the n-dimensions hypercube from which the n-points come from. Otherwise said, the neurons will classify the (sub)space, each neuron representing a partition of that (sub)space. When provided with a new point the network will act such that a single neuron will fire and that neuron will be the one closer to the provided sample. The network will classify the new point in an appropriate partition.

PathEngine Updated

Path Engine Logo

In the middleware scene, Thomas Young sent in a note that PathEngine version 5.19 has been released. See the changelog here.

New optimisations offer 2x faster preprocessing times. PathEngine has released version 5.19 of its eponymous pathfinding middleware.

Alongside general bug fixes and minor API changes, the 3D content preprocessing has been significantly sped up, with ‘most’ scenes seeing build reductions of up to half.

Firkin AI

A short rant about racing AI in Mario Kart Wii, and Need for Speed, showing some game series still have a long way to go to not frustrate gamers:

So last night I played some Mario Kart Wii and some more Need For Speed: Undercover. All-in-all, tons of fun, especially MKW. That game is pretty fun and I’m looking forward to unlocking most of the stuff so when party-goers come over we can choose from anything in the game to play.

But today is a day for ranting about video game AI, especially these two racing games. Let’s start with MKW…

F.E.A.R. 2 Artificial Intelligence Video

Now the game has been released, this video about F.E.A.R. 2’s AI few weeks ago, which discusses the goal-oriented behaviors, dynamic cover, and fire response, has resurfaced. Watch it below:




Behavioral Mathematics for Game AI

Behavioral Mathematics for Game AI

Dave Mark has been posting more information about his book over on his blog. This might be of interest for those looking for coverage of the area:

“Human behavior is never an exact science. As a result, the design and programming of artificial intelligence that seeks to replicate human behavior is already an uphill battle. Usually, the answers can not be found in sterile algorithms that are often the focus of artificial intelligence programming. However, by analyzing why people humans (and other sentient beings) behave the way we do, we can break the process down into increasingly smaller components. We can model many of those individual components in the language of logic and mathematics and then reassemble them into larger, more involved decision-making processes.

Drawing from classical game theory, this book covers both the psychological underpinnings of human decisions and the mathematical modeling techniques that AI designers and programmers can use to replicate them. With examples from both “real life” and game situations, the author explores topics such as the fallacy of “rational behavior,” utility, and the inconsistencies and contradictions that human behavior often exhibits. Readers are shown various ways of using statistics, formulas, and algorithms to create believable simulations and to model these dynamic, realistic, and interesting behaviors in video games.

Additionally, the book introduces a number of tools that the reader can use in conjunction with standard AI algorithms to make it easier to utilize the mathematical models. Lastly, the programming examples and mathematical models shown in the book are downloadable, allowing the reader to explore the possibilities in their own creations.”

Game/AI: Game AI Is Obsolete

Finally, we’ll end on a humorous video posted by Paul Tozour at Game/AI. Looks like we’re all out of work or fun things to do in AI now…

Relevant Programming Articles

On a more general note, a few good articles came up this week on Social Media sites, in particular:

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!