Planning manipulation sequences involves both, symbolic task-level planning and geometric path planning. You will study state-of-the-art approaches to tackle each of those and try to combine them.
More specifically, Marc Toussaint developed with his group a powerful, optimization-based framework for task and manipulation planning [1-3], which is capable of planning complex task sequences including tool use. However, as a (local) gradient-based approach, it is prone to local minima and cannot escape from obstacle traps. To resolve such situations, we will utilize sampling based methods when needed [4].
[1] https://ipvs.informatik.uni-stuttgart.de/mlr/marc/publications/15-toussaint-IJCAI.pdf
[2] https://ipvs.informatik.uni-stuttgart.de/mlr/papers/17-toussaint-ICRA.pdf
[3] http://www.roboticsproceedings.org/rss14/p44.pdf
[4] http://ais.informatik.uni-freiburg.de/publications/papers/schmitt19iros.pdf
The robotics courses "Robot manipulators" and "Autonomous Grasping" will be very helpful to start with this project. We will learn about optimization-based planning methods in the beginning of the project, summarizing essential knowledge in some introductory talks. Experience with C/C++ will be helpful as well.
Frequency | Weekday | Time | Format / Place | Period |
---|
Module | Course | Requirements | |
---|---|---|---|
39-M-Inf-GP Grundlagenprojekt Intelligente Systeme | Gruppenprojekt | Ungraded examination
|
Student information |
39-M-Inf-P_BI Projekt Bioinformatik | Projekt | Ungraded examination
|
Student information |
The binding module descriptions contain further information, including specifications on the "types of assignments" students need to complete. In cases where a module description mentions more than one kind of assignment, the respective member of the teaching staff will decide which task(s) they assign the students.