Course on Motion Planning and Control (6+3 hp)
Organization of the Course
A course in motion planning and control is to be held at the Department of Automatic Control, Lund University during the fall semester 2017. The course is open for all Ph.D. students as well as senior undergraduate students. The course will cover both fundamental algorithms and state-of-the-art methods for motion planning and control. Depending on the interests of the participants, more focus could also be put on fundamental planning in general. A significant part of the course will be dedicated to implementation of a number of selected algorithms and subsequent applications on small examples. The course relates to research within several ongoing projects, such as WASP, ELLIIT, and SARAFun.
In the course, there will be approximately one meeting per week. For each meeting in the course, one participant will be assigned in advance to prepare a short lecture (approximately 30 minutes) based on the reading material that has been studied by all course participants during that week. The assigned participant is also expected to, together with the course responsible, lead a joint discussion in the group on the material after the lecture. During this discussion, both the algorithms themselves and other related aspects that have appeared when implementing them are covered.
In connection with each meeting, two hand-in assignments will be requested. The assignments should be submitted to the course responsible prior to each course occasion, since the exercises will be discussed during the meeting. No written reports are required for the hand-in assignments, but the implementation code and scripts with comments should be submitted. The participant should also be prepared to present and discuss the solutions in class.
The course will be divided into two parts. The first part will cover fundamentals of motion planning and will give 6 hp. The second part will focus on more advanced aspects of motion planning and control, and will also comprise a project part where the participant performs a slightly larger theoretical or experimental evaluation of a selected method, preferably related to the research field of the participant. The project could, for instance, be based on the Open Motion Planning Library (OMPL) or the Robot Operating System (ROS). The second part of the course will give 3 hp.
The first meeting is intended to be held in the first week of September 2017 (w. 35). The aim is to have the
final meeting of the course (combined with a project seminar) in w. 49, but the schedule will be adapted to the course participants and other courses running in parallel during the fall.
Schedule – Part I
|Aug. 30, 13-15||Introduction to Motion Planning and Control||Chapters 1-2 in LaValle||[Link]||Björn Olofsson||M:1172|
|Sep. 6, 13-15||Fundamental Concepts in Motion Planning||Chapters 3-4 in LaValle||[Link]||Martin Morin||M:1172|
|Sep. 13, 13-15||Sampling-Based Methods in Motion Planning||Chapter 5 in LaValle and articles||[Link]||Karl Berntorp and Björn Olofsson||M:1172|
|Sep. 19, 13-15||Combinatorial Motion Planning||Chapter 6 in LaValle||[Link]||Martin Karlsson||M:1172|
|Sep. 27, 13-15||Additional Topics in Motion Planning||Chapter 7 in LaValle||[Link]||Marcus Greiff||M:1172|
|Oct. 3, 13-15||Introduction to Motion Planning and Feedback Control||Chapter 8 in LaValle||[Link]||Gabriel Ingesson||M:1172|
Schedule – Part II
|Oct. 11, 13-15||Sampling-Based and Optimization-Based Motion Planning with Differential Constraints||Chapters 13-14 in LaValle and articles||[Link]||Karl Berntorp and Björn Olofsson||M:1172|
|Oct. 17, 13-15||Additional Topics in Motion Planning and Control||Chapter 15 in LaValle and articles.||[Link]||Christian Rosdahl||M:1172|
|Oct. 25, 13-15||Reinforcement Learning for Motion Planning and Control||[Reading (Levine et al. 2016)] [Suggested video seminar by Levine] |
Intro to RL: [Slides], [Video lecture]
|Dec. 13, 09.30-12.00||Project Presentations and Discussions||-||-||Björn Olofsson||M:2112B|
The first part of the course will be based on Chapters 1–8 in the book:
- LaValle, S. M., Planning Algorithms, Cambridge University Press, Cambridge, UK, 2006.
The book is available for free download at the homepage of the author:
The second part of the course will be based on Chapters 14–15 in the same book. Throughout the course, the material in the book will be complemented with additional articles and book chapters, such that state-of-the-art within this rapidly developing field is covered.
- LaValle, S. M., and Kuffner J. J.: "Randomized kinodynamic planning", Int. J. Robotics Research 20:5 (2001), pp. 378–400. [Link]
- Kuwata, Y., Fiore, G. A., Teo, J., Frazzoli, E., and How, J. P.: "Motion planning for urban driving using RRT", IEEE/RSJ Int. Conf. Intelligent Robots and Systems (2008). [Link]
- Karaman, S. and Frazzoli, E.: ''Sampling-based algorithms for optimal motion planning''. Int. J. Robotics Research 30:7 (2011), pp. 846–894. [Link]
- Berntorp, K. and Di Cairano, S., ''Particle filtering for online motion planning with task specifications''. American Control Conf. (ACC), Boston, MA, 2016, pp. 2123–2128. [Link]
- Danielson, C., Weiss, A., Berntorp, K., and Di Cairano, S., ''Path planning using positive invariant sets''. Conf. Decision and Control (CDC), Las Vegas, NV, 2016, pp. 5986–5991. [Link]
- Berntorp, K., Weiss, A., Danielson, C., Kolmanovsky, I. V., and Di Cairano, S.: ''Automated Driving: Safe Motion Planning Using Positively Invariant Sets'', IEEE Int. Conf. Intelligent Transportation Systems (ITSC), Yokohama, Japan, 2017. [Link]
- Murray, R. M., and Sastry, S. S.: ”Nonholonomic motion planning: Steering using sinusoids”. IEEE Trans. Automatic Control 38:5 (1993), pp. 700–716. [Link]
- Lamiraux, F., and Laumond, J.-P.: ”Flatness and small-time controllability of multibody mobile robots: Application to motion planning”. IEEE Trans. Automatic Control 45:10 (2000), pp. 1878–1881. [Link]
- Paden, B., Cap, M., Yong, S. Z., Yershov, D., and Frazzoli, E.: "A Survey of Motion Planning and Control Techniques for Self-driving Urban Vehicles". arXiv preprint arXiv:1604.07446 (2016). [Link]
- Levine, S., Finn, C., Darrell, T., and Abbeel, P.: ”End-to-end training of deep visuomotor policies”. J. Machine Learning Research 17:39 (2016), pp. 1–40. [Link]
- Silver, D., et al.: ”Mastering the game of Go with deep neural networks and tree search”. Nature 529:7587 (2016), pp. 484–489. [Link]
The projects are to be presented at a seminar on December 13, 9.30–12.00 in the seminar room M:2112 of Dept. Automatic Control. Each project is presented for 20 minutes (including questions from the audience). After the seminar, a written report is also to be submitted to the course responsible.
Responsible for the course is Björn Olofsson (firstname.lastname@example.org) at the Department of Automatic Control, Lund University. Guest lectures will be given by Karl Berntorp about his research on motion planning and control at Mitsubishi Electric Research Laboratories (MERL). Possibly also other guest lecturers will be invited during the course.
In order to receive course credits, the participant is required to:
- Attend the weekly meetings and actively take part in the discussions.
- Submit the hand-in assignments prior to each meeting (primarily implementation code or scripts with comments, no extensive written reports required).
- Prepare one, or at most two, short lectures during the course.
- For the second part of the course, a completed mini project and a written report are required.