AI Search Methods for Mobile Robots, 3 credits

Aims and objectives

General aims for second cycle education

Second-cycle courses and study programmes shall involve the acquisition of specialist knowledge, competence and skills in relation to first-cycle courses and study programmes, and in addition to the requirements for first-cycle courses and study programmes shall

• further develop the ability of students to integrate and make autonomous use of their knowledge
• develop the students' ability to deal with complex phenomena, issues and situations, and
• develop the students' potential for professional activities that demand considerable autonomy, or for research and development work.

(Higher Education Act, Chapter 1, Section 9)

Course objectives

Knowledge and understanding
After completion of this course, the student will have knowledge about three advanced applications of AI reasoning, namely: resource scheduling, robot motion planning, and multi-robot coordination. The student will understand the computational bottlenecks of different algorithms, and will have gained a deeper understanding of the limits of current state of the art methods.

Applied knowledge and skills
Completing this course, the student will be able to formulate real-world problems as search problems, and sketch methods to solve them based on heuristic, sampling-based and constraint-based search algorithms. The student will be able to develop solutions for particular applications that are relevant in industry, namely, scheduling, robot motion planning, and coordination of fleets of autonomous robots.

Making judgments and attitudes
Completing this course, the student will be able to judge the suitability of a particular approach to automated reasoning for a given problem, have an understanding of the capabilities and limitations of the considered algorithms. Furthermore, the student will understand how problem structure relates to computational overhead.

Main content of the course

• Overview of systematic and local search methods.
• Introduction to constraint reasoning, backtracking search, k-consistency.
• Temporal constraint reasoning, constraint-based resource scheduling.
• Lattice- and sampling-based robot motion planning algorithms.
• State of the art methods for multi-robot motion planning, coordination and control.

Teaching methods

The course will be given in the form of lectures and practical project work. Three lectures will require physical presence at Örebro University, the others will be given remotely. Practical project work will involve the use of a state of the art tool for scheduling, motion planning or multi-robot coordination. The tools are explained by the instructors during the lectures, and one particular tool for carrying out the project is to be chosen by the student.

Examination methods

Exercises, 1.5 credits (Code: A001)
Examination is done based on written reports on obligatory task assignments.

Presentation at seminar, 1.5 credits (Code: A002)
Examination is based on oral presentation at seminars.

For further information, see the university's local examination regulations (in Swedish).

Grades

According to regulations on grading systems for first- and second-cycle education (vice-chancellor's decision 2019-01-15, ORU 2019/00107), one of the following grades is to be used: fail, pass, or pass with distinction. The vice-chancellor or a person appointed by the vice-chancellor may decide on exceptions from this provision for a specific course, if there are special reasons.

Grades used on course are Fail (U) or Pass (G).

Exercises
Grades used are Fail (U) or Pass (G).

Presentation at seminar
Grades used are Fail (U) or Pass (G).

For further information, see the university's local examination regulations (in Swedish).

Comments on grades

Deviations from the U-VG grading scale
Under the Vice-Chancellor's decision RB CF 55-135/2009, deviations from the three-step grading scale (Fail, Pass, Pass with Distinction) are permitted for contract education courses.

Other provisions

The course is given in English.

Reading list and other teaching materials

Additional Reading

Dechter, Rina (2003)
Constraint Processing The Morgan Kaufmann Series in Artificial Intelligence
Elsevier Science, ISBN: 0080502954, 9780080502953, 480 pages

LaValle, Steven (2006)
Planning algorithms
Cambridge university press, ISBN 978-0-521-86205-9, 831 pages

Russell, Stuart, Norvig, Peter (2010)
Artificial Intelligence, A modern Approach Prentice Hall
Prentice Hall, ISBN: 0136042597, 9780136042594, 1132 pages

Additional material (research articles) will be distributed during the course.