Artificial Intelligence and Multi-Agent Systems

• describe a number of the most prevalent techniques in artificial intelligence and multi-agent systems – both in overall terms and on a detailed technical level
• compare and assess the appropriateness of various AI techniques within automated planning and multi-agent systems for solving a given concrete problem
• combine different AI techniques in a theoretically sound and practically useful way
• apply a given AI technique to a given concrete problem
• clarify the general complications and pitfalls involved in practical uses of AI techniques
• independently explore the literature relevant to a specific AI project within automated planning and multi-agent systems
• implement non-trivial AI techniques in a relatively large software system
• communicate results within the areas of the course in the style of a research conference contribution

The course primarily focuses on topics within automated planning and multi-agent systems, but will also address other areas of AI (e.g. problem-solving by searching, knowledge representation and reasoning with logical agents).

The programming project concerns the design and implementation of advanced AI techniques in a simulated multiagent environment. The programming project is very open-ended and invites for the development of your own algorithms and multiagent architectures. The project is carried out in groups, and should result in a working system and a video in which you present the system and its underlying ideas in the style of research presentations at AI conferences.

In addition to the programming project there will be 2 smaller assignments during the course.

01017/02101/02105/02180, or equivalent courses, including knowledge about graph search algorithms, search heuristics and a bit of predicate logic. Furthermore, the course requires experience with implementing non-trivial algorithms and larger software systems.

Lectures, exercises, assignments and a large programming project.