Domain Randomization for Transferring Deep Neural Networks from
Simulation to the Real World
J. Tobin, R. Fong, A. Ray, J. Schneider, W. Zaremba, und P. Abbeel. (2017)cite arxiv:1703.06907Comment: 8 pages, 7 figures. Submitted to 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2017).
Zusammenfassung
Bridging the 'reality gap' that separates simulated robotics from experiments
on hardware could accelerate robotic research through improved data
availability. This paper explores domain randomization, a simple technique for
training models on simulated images that transfer to real images by randomizing
rendering in the simulator. With enough variability in the simulator, the real
world may appear to the model as just another variation. We focus on the task
of object localization, which is a stepping stone to general robotic
manipulation skills. We find that it is possible to train a real-world object
detector that is accurate to $1.5$cm and robust to distractors and partial
occlusions using only data from a simulator with non-realistic random textures.
To demonstrate the capabilities of our detectors, we show they can be used to
perform grasping in a cluttered environment. To our knowledge, this is the
first successful transfer of a deep neural network trained only on simulated
RGB images (without pre-training on real images) to the real world for the
purpose of robotic control.
Beschreibung
[1703.06907] Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real World
%0 Generic
%1 tobin2017domain
%A Tobin, Josh
%A Fong, Rachel
%A Ray, Alex
%A Schneider, Jonas
%A Zaremba, Wojciech
%A Abbeel, Pieter
%D 2017
%K domainrandomization machinelearning
%T Domain Randomization for Transferring Deep Neural Networks from
Simulation to the Real World
%U http://arxiv.org/abs/1703.06907
%X Bridging the 'reality gap' that separates simulated robotics from experiments
on hardware could accelerate robotic research through improved data
availability. This paper explores domain randomization, a simple technique for
training models on simulated images that transfer to real images by randomizing
rendering in the simulator. With enough variability in the simulator, the real
world may appear to the model as just another variation. We focus on the task
of object localization, which is a stepping stone to general robotic
manipulation skills. We find that it is possible to train a real-world object
detector that is accurate to $1.5$cm and robust to distractors and partial
occlusions using only data from a simulator with non-realistic random textures.
To demonstrate the capabilities of our detectors, we show they can be used to
perform grasping in a cluttered environment. To our knowledge, this is the
first successful transfer of a deep neural network trained only on simulated
RGB images (without pre-training on real images) to the real world for the
purpose of robotic control.
@misc{tobin2017domain,
abstract = {Bridging the 'reality gap' that separates simulated robotics from experiments
on hardware could accelerate robotic research through improved data
availability. This paper explores domain randomization, a simple technique for
training models on simulated images that transfer to real images by randomizing
rendering in the simulator. With enough variability in the simulator, the real
world may appear to the model as just another variation. We focus on the task
of object localization, which is a stepping stone to general robotic
manipulation skills. We find that it is possible to train a real-world object
detector that is accurate to $1.5$cm and robust to distractors and partial
occlusions using only data from a simulator with non-realistic random textures.
To demonstrate the capabilities of our detectors, we show they can be used to
perform grasping in a cluttered environment. To our knowledge, this is the
first successful transfer of a deep neural network trained only on simulated
RGB images (without pre-training on real images) to the real world for the
purpose of robotic control.},
added-at = {2019-10-11T18:55:42.000+0200},
author = {Tobin, Josh and Fong, Rachel and Ray, Alex and Schneider, Jonas and Zaremba, Wojciech and Abbeel, Pieter},
biburl = {https://www.bibsonomy.org/bibtex/2944d326e0423f2bf0090c32f55693273/cpankow},
description = {[1703.06907] Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real World},
interhash = {7f8c9229cc57c1356dbf25f1fb3c8bfc},
intrahash = {944d326e0423f2bf0090c32f55693273},
keywords = {domainrandomization machinelearning},
note = {cite arxiv:1703.06907Comment: 8 pages, 7 figures. Submitted to 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2017)},
timestamp = {2019-10-11T18:55:42.000+0200},
title = {Domain Randomization for Transferring Deep Neural Networks from
Simulation to the Real World},
url = {http://arxiv.org/abs/1703.06907},
year = 2017
}