Isaac Gym: High Performance GPU-Based Physics Simulation For Robot Learning

Isaac Gym offers a high performance learning platform to train policies for wide variety of robotics tasks directly on GPU. Both physics simulation and the neural network policy training reside on GPU and communicate by directly passing data from physics buffers to PyTorch tensors without ever going...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2021-08
Main Authors: Makoviychuk, Viktor, Wawrzyniak, Lukasz, Guo, Yunrong, Lu, Michelle, Storey, Kier, Macklin, Miles, Hoeller, David, Rudin, Nikita, Allshire, Arthur, Handa, Ankur, State, Gavriel
Format: Article
Language:English
Subjects:
Neural networks
Physics
Robot learning
Robotics
Simulation
Task complexity
Tensors
Training
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Isaac Gym offers a high performance learning platform to train policies for wide variety of robotics tasks directly on GPU. Both physics simulation and the neural network policy training reside on GPU and communicate by directly passing data from physics buffers to PyTorch tensors without ever going through any CPU bottlenecks. This leads to blazing fast training times for complex robotics tasks on a single GPU with 2-3 orders of magnitude improvements compared to conventional RL training that uses a CPU based simulator and GPU for neural networks. We host the results and videos at \url{https://sites.google.com/view/isaacgym-nvidia} and isaac gym can be downloaded at \url{https://developer.nvidia.com/isaac-gym}.
ISSN:2331-8422