📡 Embodied AI Radar具身智能雷达

Topics主题 › Sim-to-Real & Simulators仿真与 Sim-to-Real

Paper论文 2017 · OpenAI

Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real World

Randomizing textures, lighting and cameras in sim makes vision models work on real images with no real data.在仿真中随机化纹理、光照和相机,使视觉模型无需真实数据即可在真实图像上工作。

BibTeX
@misc{tobin2017domain,
  title = {{Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real World}},
  author = {Josh Tobin and Rachel Fong and Alex Ray and Jonas Schneider and Wojciech Zaremba and Pieter Abbeel},
  year = {2017},
  eprint = {1703.06907},
  archivePrefix = {arXiv},
  primaryClass = {cs.RO},
  url = {https://arxiv.org/abs/1703.06907},
}

Authors (from arXiv): 作者(来自 arXiv):Josh Tobin, Rachel Fong, Alex Ray, Jonas Schneider, Wojciech Zaremba, Pieter Abbeel

More from: 同机构更多条目:OpenAIOpenAI

More in Sim-to-Real & Simulators更多仿真与 Sim-to-Real条目

All Sim-to-Real & Simulators entries →查看全部仿真与 Sim-to-Real条目 →