|Credits:||7.5 ECTS (=5.25 old credit points)|
|Period:||period 4 (week 17 through 27, i.e., 24-4-2006 through 7-7-2006; retake week 35)
|Participants:||up till now 17 subscriptions|
|Schedule:||Dit is een oud rooster!
|Contents:||Games like chess, checkers and go have been a (successful) object of study for Artificial Intelligence for a long time already. In active games like Quake and all the the other Playstation or Xbox variants the use of AI techniques has been mainly limited to simple path planning for virtual characters. In this seminar we will explore the more serious use of AI techniques for these active games.
We will investigate how the central game control can be distributed over several independently operating agents controlling characters and/or parts of the environment. Of course we will also investigate several path-planning techniques which are useful in computer games, and we will describe dynamic replanning algorithms that are useful for dealing with dynamic environments. Furthermore we will discuss machine learning techniques such as evolutionary algorithms with neural networks to let agents learn their own behavior while playing game tournaments. We also discuss the importance of multi-agent cooperation and for this we will discuss some game theory and solutions for cooperative team work. |
|Literature:||Instead of a single book we will use a number of different articles that will be discussed during the class meetings. We will also make slides available on the webpage.|
|Course form:||There will be 1 presentation each week given by Jan Broersen or Marco Wiering. The other time slot each week is used for a meeting where we will discuss articles.
An important part of this course consists of a programming assignment where a group of 4 students has to make and test a particular intelligent agent or algorithm using e.g. the Quake-3 engine. |
|Exam form:||The final mark is determined by:
Final Practical Project (50%)
|Minimum effort to qualify for 2nd chance exam:||Om aan de aanvullende toets te mogen meedoen is ontbreken van ten hoogte 1 toetsactiviteit toegestaan.|
|Description:||Students are required to have the knowledge equivalent to the courses (1) Intelligent Agents, and (2) Reinforcement Learning. Furthermore, practical experience with the computer language C++ is desirable.|