What question did this study set out to answer?

This study aims to explore the effectiveness of domain randomization in transferring deep neural networks from simulated to real-world environments for robotics tasks.

September 1, 2017

Domain randomization for transferring deep neural networks from simulation to the real world

View Full Paper

Ask AI

Bookmark

Authors

JTJosh TobinOpenAI (United States)RFRachel FongOpenAI (United States)ARAlex RayOpenAI (United States)

Discussion

Member takes

Implication

Randomized trial shows effective object localization in robotic tasks using simulation data, suggesting improved real-world application.

Key Points

This study aims to explore the effectiveness of domain randomization in transferring deep neural networks from simulated to real-world environments for robotics tasks.
Trained deep neural networks using simulated images with randomized rendering.
Focused on object localization for robotic manipulation tasks.
Evaluated performance using only simulator data without real-world pre-training.
Achieved object localization accuracy of 1.5 cm under various conditions.
Demonstrated robustness to distractors and occlusions during testing.
Successful grasping of objects in cluttered environments using the trained model.

Cite This Study

Tobin et al. (2017) studied this question.

synapsesocial.com/papers/69d6b5f8a0177bf533ed8b09 https://doi.org/10.1109/iros.2017.8202133

View Full Paper

Ask AI

Bookmark

Domain randomization for transferring deep neural networks from simulation to the real world

Discussion

Member takes

Implication

Key Points

Cite This Study

Domain randomization for transferring deep neural networks from simulation to the real world

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider