Master Thesis Proposal
“Imitation Learning of Human Grasping Skills using a five-fingered Robot Hand"
We are looking for a Master student who is interested in writing his/her Master's thesis on related topics to “Imitation Learning of Human Grasping Skills using a five-fingered Robot Hand".
Today robots are not any more restricted to industrial settings only, but also enter domestic environments where they are supposed to assist the human in performing everyday manipulation tasks. Handling and manipulation of everyday objects is still considered a very challenging task for state-of-the-art robotic systems. In this thesis a programming-by-demonstration framework should be developed which allows transferring human grasping and manipulation skills to robots. In the first step, different types of grasps should be classified based on the analysis of captured motion and force data acquired
during human manipulation of objects, see e.g. [1, 2]. Then, these grasp types should be transferred to a robotic hand which imitates the human behavior . For this purpose algorithms for grasp classification, mapping of grasps and low-level control algorithms for closed-loop grasp synthesis and
adaptation need to be developed, implemented and tested using a five-fingered robotic hand.
Detailed research issues are
• Review of state of the art in robot programming-by-demonstration for grasping and manipulation
• Recording of motion/force data during manipulation and implementation of grasp classification algorithms
• Implementation of mapping algorithms for robot hand
• Implementation of low-level control algorithms and adaptation of grasping and manipulation skills
• Implementation and evaluation on a robot hand
 Ekvall, S., Kragic, D. Grasp Recognition for Programming by Demonstration. Proceedings of the 2005 IEEE International Conference on Robotics and Automation, 2005, p.748-753.
 Bernadin, K., Ogawara K., Ikeuchi, K. and Dillmann, R. A hidden markov model based sensor fusion approach for recognizing continuous human grasping sequencess, IEEE International Conference on Humanoid Robots, 2003
 A. Schmidts, Dongheui Lee, and A. Peer , Imitation Learning of Human Grasping Skills from Motion and Force Data, IEEE/RSJ International Conference on Intelligent Robots and Systems, 2011, p. 1002- 1007
- good programming skills on C, C++, matlab.
- basic knowledge of robotics, machine learning and control.
Supervised by Prof. Lee and Dr.-Ing Angelika Peer.
If you are interested in this topic, please contact Prof. Dongheui Lee (firstname.lastname@example.org) with additional information (e.g., academic transcript and state your skills).