Learning dexterous in-hand manipulation
Top Cited Papers
Open Access
- 18 November 2019
- journal article
- research article
- Published by SAGE Publications in The International Journal of Robotics Research
- Vol. 39 (1), 3-20
- https://doi.org/10.1177/0278364919887447
Abstract
We use reinforcement learning (RL) to learn dexterous in-hand manipulation policies that can perform vision-based object reorientation on a physical Shadow Dexterous Hand. The training is performed in a simulated environment in which we randomize many of the physical properties of the system such as friction coefficients and an object’s appearance. Our policies transfer to the physical robot despite being trained entirely in simulation. Our method does not rely on any human demonstrations, but many behaviors found in human manipulation emerge naturally, including finger gaiting, multi-finger coordination, and the controlled use of gravity. Our results were obtained using the same distributed RL system that was used to train OpenAI Five. We also include a video of our results: https://youtu.be/jwSbzNHGflM.Keywords
This publication has 42 references indexed in Scilit:
- Learning contact-rich manipulation skills with guided policy searchPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2015
- Rotary object dexterous manipulation in hand: a feedback-based methodInternational Journal of Mechatronics and Automation, 2013
- Hands for dexterous manipulation and robust grasping: a difficult road toward simplicityIEEE Transactions on Robotics and Automation, 2000
- Mechanics, Planning, and Control for TappingThe International Journal of Robotics Research, 2000
- Planning quasi-static fingertip manipulations for reconfiguring objectsIEEE Transactions on Robotics and Automation, 1999
- In-Hand Dexterous Manipulation of Piecewise-Smooth 3-D ObjectsThe International Journal of Robotics Research, 1999
- An Exploration of Nonprehensile Two-Palm ManipulationThe International Journal of Robotics Research, 1998
- Long Short-Term MemoryNeural Computation, 1997
- Tumbling Objects Using a Multi-fingered Robot.Journal of the Robotics Society of Japan, 1991
- An exploration of sensorless manipulationIEEE Journal on Robotics and Automation, 1988