📡 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 2021 · Google NeurIPS 2021 Datasets and Benchmarks

Brax: A Differentiable Physics Engine for Large Scale Rigid Body Simulation

JAX physics engine that runs thousands of environments in parallel on GPU/TPU together with the RL training loop, cutting locomotion training from hours to minutes.基于 JAX 的物理引擎，在 GPU/TPU 上与强化学习训练循环一起并行运行数千个环境，把运动控制训练从数小时缩短到数分钟。

Paper论文 · Code代码 ★ 3.3k
BibTeX
@misc{freeman2021brax,
  title = {{Brax: A Differentiable Physics Engine for Large Scale Rigid Body Simulation}},
  author = {C. Daniel Freeman and Erik Frey and Anton Raichuk and Sertan Girgin and Igor Mordatch and Olivier Bachem},
  year = {2021},
  eprint = {2106.13281},
  archivePrefix = {arXiv},
  primaryClass = {cs.RO},
  url = {https://arxiv.org/abs/2106.13281},
}
Sim-to-Real & Simulators仿真与 Sim-to-RealRobot Learning: RL & Imitation机器人学习：强化学习与模仿学习physicsjaxgputpudifferentiable

Authors (from arXiv): 作者（来自 arXiv）：C. Daniel Freeman, Erik Frey, Anton Raichuk, Sertan Girgin, Igor Mordatch, Olivier Bachem

More from: 同机构更多条目：Google DeepMind / Google谷歌 DeepMind / 谷歌

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 →