Supervisor: Prof. Ville Kyrki (firstname.lastname@example.org).
Advisors: Dr. Fares J. Abu-Dakka (email@example.com).
Keywords: robot learning, dynamic movement primitives, learning from demonstration, symmetric positive definite matrix.
In many robot control problems, factors such as stiffness and damping matrices and manipulability ellipsoids are naturally represented as symmetric positive definite (SPD) matrices, which capture the specific geometric characteristics of those factors. Typical learned skill models such as dynamic movement primitives (DMPs) can not, however, be directly
employed with quantities expressed as SPD matrices as they are limited to data in Euclidean space. In order to overcome such limitation, a novel and mathematically principled framework that uses Riemannian metrics is developed to reformulate DMPs such that the resulting formulation can operate with SPD data in the SPD manifold.
- Review of relevant state-of-the-art literature;
- Implementation of GA-DMP using C++;
- Evaluating the method on a physical robot.
Prerequisites: Basics of robot learning, C++, Mtlab, Linux.
Suggested tools: C++, Matlab; MuJoCo or VRep; ROS.
Platform: Franka Panda robotic arm.
Start: Available immediately
 Abu-Dakka, Fares J., and Ville Kyrki. “Geometry-aware Dynamic Movement Primitives.” arXiv preprint arXiv:2003.06061 (2020).