Principles of AI Planning - Overview
Lecturers: Prof. Dr. Bernhard Nebel and Dr. Robert Mattmüller
Exercises: David Speck
Time
Lecture: Wednesday 14:15-16:00 and Friday 14:15-15:00.
Exercises: Friday 15:15-16:00.
Exam: February 24th, 2017. Details to be announced.
Location
Lecture: Building 101, seminar room 01-016.
Exercises: Building 101, seminar room 01-016.
Language
The lecture will be given in English, and the lecture slides will be in English as well. Exercises may be answered in German or English.
Topics
The lecture provides a detailed introduction to the theoretical and algorithmic foundations of modern AI planning systems. In detail, we will cover the following topics:
- Formalization of planning
- Planning as search; progression and regression
- Satisficing heuristic-search planning using relaxation heuristics
- Optimal heuristic-search planning using abstraction heuristics
- Optimal heuristic-search planning using landmark heuristics
- State-space pruning techniques for planning
- Planning in nondeterministic domains
- Theoretical complexity of planning
Prerequisites
The course is primarily aimed at Masters students majoring in Computer Science, but advanced Bachelors students in their final year and CS minors with the necessary background are also welcome.
The essential concepts from complexity theory (NP completeness, polynomial reductions) should be known. We also expect basic knowledge of the basic search algorithms covered in the lecture on Foundations of Artificial Intelligence, such as depth-first search, breadth-first search, heuristic search with the A* algorithm, or greedy best-first search. Basic knowledge of (propositional) logic is expected.
Exercises and Exam
Bachelors and Masters students in Computer Science can take this course as part of their specialization in the area of cognitive technical systems. There will be a final exam that needs to be passed, which will be oral for Bachelors students and which will be either oral or written, at our discretion, depending on the number of students registered for the exam, for all other students excepts Bachelors. The exam will take place in the semester break after the course.
During the semester there will be weekly exercises (theoretical assignments and occasional implementation projects). For the admission to the exam it is necessary to reach 50% of all points.
Exercises and projects may be worked on in groups of two. Larger groups or copied solutions will not be accepted and result in nonadmission to the exam.
In this course 6 ECTS credits can be earned.
Course Materials
The lecture slides will be uploaded to the course website during the semester. Additional material can be found on the bibliography webpage.
The lecture will not be recorded.