The TriFinger Robots
TriFingerPro
TriFingerEdu
The TriFinger robots have been developed and built at the Max Planck Institute for Intelligent Systems (MPI-IS) in Tübingen, Germany. There are currently eight TriFingerPro robots hosted at MPI-IS which can be accessed remotely and were used for the Real Robot Challenge. The TriFingerEdu is an open-source version of the challenge robot, with identical actuators, almost identical kinematics, and identical software. Compared to the challenge platform, the "Edu"-version is easier to construct, such that researchers are able to build it themselves.
Access the TriFingerPro at MPI-IS
If you are interested in using the TriFingerPro robots for your own research project, you can apply for access by sending us a short proposal describing what you are planning to do and a rough estimate how long you would need access. See Contact.
Build Your Own TriFingerEdu
You can build your own TriFingerEdu robot, instructions are online here: TriFingerEdu hardware documentation. If you have questions about building a robot, please see the ODRI forum.
TriFinger Simulation
To facilitate testing code and sim-to-real transfer, we provide a simulation of the platform with same software interface as the real robot (see documentation).
In addition, we developed a more advanced version of the simulator where parameters of the robot and the environment (such as masses, colors of objects, size of objects etc.) can be changed easily (see paper [ahmed2020] and the corresponding website). This allows for learning the causal structure of the control problem and facilitates transfer learning, in particular transferring a policy to the real world.
Links
Website of the Real Robot Challenge
ODRI forum: For any questions about building your own robot.
Contact
If you have general questions about the robots or want to apply for access on the robots at MPI-IS, please contact Felix Widmaier.
For specific questions about building your own robot, please see the ODRI forum.
References
Manuel Wuthrich, Felix Widmaier, Felix Grimminger, Shruti Joshi, Vaibhav Agrawal, Bilal Hammoud, Majid Khadiv, Miroslav Bogdanovic, Vincent Berenz, Julian Viereck, Maximilien Naveau, Ludovic Righetti, Bernhard Schölkopf, and Stefan Bauer. Trifinger: an open-source robot for learning dexterity. In Jens Kober, Fabio Ramos, and Claire J. Tomlin, editors, 4th Conference on Robot Learning, CoRL 2020, 16-18 November 2020, Virtual Event / Cambridge, MA, USA, volume 155 of Proceedings of Machine Learning Research, 1871–1882. PMLR, 2020. URL: https://proceedings.mlr.press/v155/wuthrich21a.html.
Stefan Bauer, Manuel Wüthrich, Felix Widmaier, Annika Buchholz, Sebastian Stark, Anirudh Goyal, Thomas Steinbrenner, Joel Akpo, Shruti Joshi, Vincent Berenz, Vaibhav Agrawal, Niklas Funk, Julen Urain De Jesus, Jan Peters, Joe Watson, Claire Chen, Krishnan Srinivasan, Junwu Zhang, Jeffrey Zhang, Matthew R. Walter, Rishabh Madan, Takuma Yoneda, Denis Yarats, Arthur Allshire, Ethan K. Gordon, Tapomayukh Bhattacharjee, Siddhartha S. Srinivasa, Animesh Garg, Takahiro Maeda, Harshit Sikchi, Jilong Wang, Qingfeng Yao, Shuyu Yang, Robert McCarthy, Francisco Sanchez, Qiang Wang, David Cordova Bulens, Kevin McGuinness, Noel E. O'Connor, Stephen J. Redmond, and Bernhard Schölkopf. Real robot challenge: A robotics competition in the cloud. In Douwe Kiela, Marco Ciccone, and Barbara Caputo, editors, NeurIPS 2021 Competitions and Demonstrations Track, 6-14 December 2021, Online, volume 176 of Proceedings of Machine Learning Research, 190–204. PMLR, 2021. URL: https://proceedings.mlr.press/v176/bauer22a.html.
Ossama Ahmed, Frederik Träuble, Anirudh Goyal, Alexander Neitz, Manuel Wuthrich, Yoshua Bengio, Bernhard Schölkopf, and Stefan Bauer. Causalworld: A robotic manipulation benchmark for causal structure and transfer learning. In 9th International Conference on Learning Representations, ICLR 2021, Virtual Event, Austria, May 3-7, 2021. OpenReview.net, 2021. URL: https://openreview.net/forum?id=SK7A5pdrgov.
Qiang Wang, Francisco Roldan Sanchez, Robert McCarthy, David Cordova Bulens, Kevin McGuinness, Noel E. O'Connor, Manuel Wüthrich, Felix Widmaier, Stefan Bauer, and Stephen J. Redmond. Dexterous robotic manipulation using deep reinforcement learning and knowledge transfer for complex sparse reward-based tasks. CoRR, 2022. URL: https://doi.org/10.48550/arXiv.2205.09683, arXiv:2205.09683, doi:10.48550/arXiv.2205.09683.
Niklas Funk, Charles Schaff, Rishabh Madan, Takuma Yoneda, Julen Urain De Jesus, Joe Watson, Ethan K. Gordon, Felix Widmaier, Stefan Bauer, Siddhartha S. Srinivasa, Tapomayukh Bhattacharjee, Matthew R. Walter, and Jan Peters. Benchmarking structured policies and policy optimization for real-world dexterous object manipulation. IEEE Robotics and Automation Letters, 7(1):478–485, Jan 2022. URL: http://dx.doi.org/10.1109/LRA.2021.3129139, doi:10.1109/lra.2021.3129139.
Claire Chen, Krishnan Srinivasan, Jeffrey Zhang, Junwu Zhang, Lin Shao, Shenli Yuan, Preston Culbertson, Hongkai Dai, Mac Schwager, and Jeannette Bohg. Dexterous manipulation primitives for the real robot challenge. 2021. arXiv:2101.11597.
Team troubledhare. Real robot challenge phase 2: manipulating objects using high-level coordination of motion primitives. In Submitted to Real Robot Challenge 2021. 2021. under review. URL: https://openreview.net/forum?id=9tYX-lukeq.
Arthur Allshire, Mayank Mittal, Varun Lodaya, Viktor Makoviychuk, Denys Makoviichuk, Felix Widmaier, Manuel Wüthrich, Stefan Bauer, Ankur Handa, and Animesh Garg. Transferring dexterous manipulation from gpu simulation to a remote real-world trifinger. 2021. arXiv:2108.09779.
Robert McCarthy, Francisco Roldan Sanchez, Qiang Wang, David Cordova Bulens, Kevin McGuinness, Noel O'Connor, and Stephen J. Redmond. Solving the real robot challenge using deep reinforcement learning. 2021. arXiv:2109.15233.
Qingfeng Yao, Jilong Wang, and Shuyu Yang. Real-world dexterous object manipulation based deep reinforcement learning. 2021. arXiv:2112.04893.
Diego Agudelo-España, Andrii Zadaianchuk, Philippe Wenk, Aditya Garg, Joel Akpo, Felix Grimminger, Julian Viereck, Maximilien Naveau, Ludovic Righetti, Georg Martius, Andreas Krause, Bernhard Schölkopf, Stefan Bauer, and Manuel Wüthrich. A real-robot dataset for assessing transferability of learned dynamics models. In 2020 IEEE International Conference on Robotics and Automation, ICRA 2020, Paris, France, May 31 - August 31, 2020, 8151–8157. IEEE, 2020. URL: https://doi.org/10.1109/ICRA40945.2020.9197392, doi:10.1109/ICRA40945.2020.9197392.