Adaptive Constraint-Based Agents in Artificial Environments

[MAIN] [Goals]   [Software]   [Documentation]   [Publications]   [FAQ]   [Contact Information]

[ Please note: The project has been discontinued as of May 31, 2005 and is superseded by the projects of the ii Labs. There won't be further updates to these pages. ]


What about artificial intelligence in current games?

Characters in current games are mostly guided in a hard-wired stimulus-response manner. For a specific situation, a specific action or a predefined script is executed. The characters rarely react appropriately in situations that were not foreseen by the programmers (see also our game AI overview article).

What is the contribution of the EXCALIBUR project?

We don't want computer-guided characters to follow predefined behavior scripts, but to plan their behavior on their own. In short, each of our characters has sensors (like: where am I, what do I have in my hand), some primitive actions (like: move forward, eat apple, turn lever, say x to y) and some given goals (like: improve well-being, kill player, be at 9 PM at the bridge). Now the character must automatically generate a behavior plan that most optimally fulfills these goals. If something unexpected happens during plan generation or execution, the plan needs to be revised.

Such a character is able to get along with unforeseen situations as well as changing environments. Cooperation and coordination with other characters may be needed for this, and the character will execute everything that he decides to be necessary to pursue his given goal.

The characters also have to learn about their environment, facts (like: cars are often on streets) as well as dependencies (like: turning lever moves elevator). They can also communicate their knowledge to other characters. This part of research is planned for the later phase of the project.

Does the base technology already exist?

The current state of the art in artificial intelligence technology is the use of mostly static off-line computations. Techniques and models for dynamic real-time environments like computer games are less explored (see also our article on the relevance of AI research). In consequence, EXCALIBUR is concerned with much basic research.

Is EXCALIBUR's technology restricted to computer games?

No. There is an enormous amount of possible applications. Every dynamic real-time domain with the need of autonomous and goal-oriented behavior is a potential application candidate.


Is there a document with the complete documentation?

A complete version is not available as the pages are constantly extended and revised.

Why is the online information sometimes different from the publications' contents?

The publications describe the project's state at the time of their publication and are not changed anymore. Thus, the online documentation is much more up-to-date (but less complete because we are lazy in updating it).


Will there be a demo?

Sure. However, this will take some time because we don't want to produce a good-looking demo "hack" but provide robust cutting-edge software. When a demo will be available, it is downloadable from the EXCALIBUR project's homepage.

Will the source code be available?

Yes, as planned by now, everything will be released under a permissive non-copyleft free software license. This means that you can use, modify, distribute, sell, etc. the software/code free of charge. An early version of a subpart of our software - the DragonBreath search/optimization engine - is already available. See our [Software] page for details.


Who is involved in the project?

The project is led by Dr. Alexander Nareyek and involves persons from diverse institutions - mostly Ph.D. and master students from all over the world. The project is also supported by the game industry (e.g., by way of software support), and we will intensify this cooperation once we get closer to a working prototype.

If you are interested in getting involved, please feel free to contact us.

How long does the project take?

The project started in late 1997 and is open-ended.

[MAIN] [Goals]   [Software]   [Documentation]   [Publications]   [FAQ]   [Contact Information]

For questions, comments or suggestions, please contact us.

Last update:
May 31, 2005 by Alexander Nareyek