Rapidly Learn Manipulation Tasks by Demonstration
Hackathon challenge by Franka Emika at ERF2022
This project is our implementation at the Hackathon challenge at the European Robotics Forum 2022, provided by the sponsor Franka Emika. We (Delft University of Technology team) won the final prize.
We solved high-precision robot assembly tasks on a randomly positioned task board. We used a depth camera and the localization algorithm (global point cloud registration + ICP + physical touching motion) to localize the randomly positioned task board precisely. Then, the tasks were done by using the learning-from-demonstration (LfD) method.
The supplementary videos roughly demonstrated our task performance during the competition. We are working on improving our pipeline and giving more interesting results.
Code & Paper
Coming soon! We are working on refining our implementation and more interesting results!