Using Simulation and Domain Adaptation to Improve Efficiency of Deep Robotic Grasping

Bousmalis, Konstantinos; Irpan, Alex; Wohlhart, Paul; Bai, Ye; Kelcey, Matthew; Kalakrishnan, Mrinal; Downs, Laura Lee; Ibarz, Julian; Pastor, Peter; Konolige, Kurt; Levine, Sergey; Vanhoucke, Vincent

doi:10.48550/arxiv.1709.07857

Cited by 30 publications

(41 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…2-dimensional) to high-accuracy (e.g. 3-dimensional) simulations (Verma et al 2018), or from simulations to related real-world applications (Richter et al 2016;Bousmalis et al 2017).…”

Section: Flow Control Via Reinforcement Learningmentioning

confidence: 99%

Machine Learning for Fluid Mechanics

Brunton

Noack

Koumoutsakos

2020

Annu. Rev. Fluid Mech.

1,961

742

View full text Add to dashboard Cite

The field of fluid mechanics is rapidly advancing, driven by unprecedented volumes of data from experiments, field measurements, and large-scale simulations at multiple spatiotemporal scales. Machine learning presents us with a wealth of techniques to extract information from data that can be translated into knowledge about the underlying fluid mechanics. Moreover, machine learning algorithms can augment domain knowledge and automate tasks related to flow control and optimization. This article presents an overview of past history, current developments, and emerging opportunities of machine learning for fluid mechanics. We outline fundamental machine learning methodologies and discuss their uses for understanding, modeling, optimizing, and controlling fluid flows. The strengths and limitations of these methods are addressed from the perspective of scientific inquiry that links data with modeling, experiments, and simulations. Machine learning provides a powerful information processing framework that can augment, and possibly even transform, current lines of fluid mechanics research and industrial applications.

show abstract

“…2-dimensional) to high-accuracy (e.g. 3-dimensional) simulations (Verma et al 2018), or from simulations to related real-world applications (Richter et al 2016;Bousmalis et al 2017).…”

Section: Flow Control Via Reinforcement Learningmentioning

confidence: 99%

Machine Learning for Fluid Mechanics

Brunton

Noack

Koumoutsakos

2020

Annu. Rev. Fluid Mech.

1,961

742

View full text Add to dashboard Cite

show abstract

“…In another work developed concurrently with this one [3], the authors reach a similar conclusion about the utility of procedurally generated objects for the purpose of robotic grasping. In contrast to this work, theirs focuses on how to combine simulated data with real grasping data to achieve successful transfer to the real world, but does not focus on achieving a high overall success rate.…”

Section: A Domain Randomizationmentioning

confidence: 52%

“…• We explore the effect of training a model for grasping using unrealistic procedurally generated objects and show that such a model can achieve similar success to one trained on a realistic object distribution. (Another paper [3] developed concurrently to this one explored a similar idea and reached similar conclusions.)…”

Section: Introductionmentioning

confidence: 62%

“…To alleviate this problem, some work explores automated large-scale data collection [23], [36]. Others have explored replacing real data with synthetic depth data at training time [14], [24], [25], [49], or combining synthetic RGB images with real images [3]. In many cases, simulated data appears to be effective in replacing or supplementing real-world data in robotic grasping.…”

Section: Deep Learning For Robotic Graspingmentioning

confidence: 99%

See 1 more Smart Citation

Domain Randomization and Generative Models for Robotic Grasping

Tobin

Biewald²,

Duan³

et al. 2018

2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

140

106

View full text Add to dashboard Cite

Fig. 1. An overview of our approach. Since creating large numbers of realistic object models is challenging, we train our deep autoregressive model architecture on millions of unrealistic procedurally generated objects (indicated in blue above) and billions of unique grasp attempts. At test time, our model generalizes to realistic objects from the YCB dataset (indicated in green above) [4] with 92% success rate.Abstract-Deep learning-based robotic grasping has made significant progress thanks to algorithmic improvements and increased data availability. However, state-of-the-art models are often trained on as few as hundreds or thousands of unique object instances, and as a result generalization can be a challenge.In this work, we explore a novel data generation pipeline for training a deep neural network to perform grasp planning that applies the idea of domain randomization to object synthesis. We generate millions of unique, unrealistic procedurally generated objects, and train a deep neural network to perform grasp planning on these objects.Since the distribution of successful grasps for a given object can be highly multimodal, we propose an autoregressive grasp planning model that maps sensor inputs of a scene to a probability distribution over possible grasps. This model allows us to sample grasps efficiently at test time (or avoid sampling entirely). We evaluate our model architecture and data generation pipeline in simulation and the real world. We find we can achieve a >90% success rate on previously unseen realistic objects at test time in simulation despite having only been trained on random objects. We also demonstrate an 80% success rate on real-world grasp attempts despite having only been trained on random simulated objects.

show abstract

“…Domain randomization was successfully used to train dexterous in-hand manipulation [22], aligning an optical interferometer [23] and learning robot locomotion [5]. Examples of domain adaptation include using GANs for image-based robotic grasp [24] and image-based auto-tuning of the simulator parameters for robotic hand manipulation [25]. These approaches require data from the test environment during training, which are not always available.…”

Section: Related Workmentioning

confidence: 99%

Adaptation of Quadruped Robot Locomotion with Meta-Learning

Kuzhamuratov

Sorokin²,

Ulanov³

et al. 2021

Preprint

View full text Add to dashboard Cite

Animals have remarkable abilities to adapt locomotion to different terrains and tasks. However, robots trained by means of reinforcement learning are typically able to solve only a single task and a transferred policy is usually inferior to that trained from scratch. In this work, we demonstrate that meta-reinforcement learning can be used to successfully train a robot capable to solve a wide range of locomotion tasks. The performance of the meta-trained robot is similar to that of a robot that is trained on a single task.

show abstract

Using Simulation and Domain Adaptation to Improve Efficiency of Deep Robotic Grasping

Cited by 30 publications

References 0 publications

Machine Learning for Fluid Mechanics

Machine Learning for Fluid Mechanics

Domain Randomization and Generative Models for Robotic Grasping

Adaptation of Quadruped Robot Locomotion with Meta-Learning

Contact Info

Product

Resources

About