An engine for learning probabilistic movement primitives from demonstrations in joint space. Implemented and tested on a Panda robot from Franka Emika. Check engine.py for the code. Released in Open Source with an MIT License.
The methods implemented were adapted for two papers by the labs of Jan Peters in Tübingen and Darmstadt.
These papers are:
[1]. Using Probabilistic Movement Primitives in Robotics
By Paraschos, Daniel, Peters and Neumann
https://www.ias.informatik.tu-darmstadt.de/uploads/Team/AlexandrosParaschos/promps_auro.pdf
and
[2]. Adaptation and Robust Learning of Probabilistic Movement Primitives
By Gomez-Gonzalez, Neumann, Schelkopf and Peters
https://arxiv.org/pdf/1808.10648.pdf
Also there's input from
[3]. Probabilistic Movement Primitives
By Paraschos, Daniel, Peters and Neumann
https://www.ias.informatik.tu-darmstadt.de/uploads/Publications/Paraschos_NIPS_2013a.pdf
Autors of the code: Cheesecake Team 🍰 🤖
- Claudia Winklmayr
- Sevim Calikstan
- Samuel Bustamante