Docker Containers#

This first version of Isaac Lab Arena is designed to run inside a Docker container.

We provide two docker containers for Isaac Lab Arena:

  • Base: Contains the Isaac Lab Arena code and all its dependencies. Tagged as isaaclab_arena:latest.

  • Base + GR00T: Additionally includes GR00T and its dependencies. Tagged as isaaclab_arena:cuda_gr00t_gn16.

We include the two containers such that the user can choose between container with minimal dependencies (Base) or container with all dependencies (Base + GR00T).

In order to start the containers run:

./docker/run_docker.sh

The run docker will build the container and then enter in interactive mode.

Note

The container with all dependencies (Base + GR00T) is significantly larger than the container with minimal dependencies (Base), so it is recommended to use the Base container for development and the Base + GR00T container for GR00T policy post-training and evaluation. If you are not sure which container to use, we recommend using the Base container. If you want to use the Base + GR00T container for development, currently it is not supported to run on Blackwell GPUs, and DGX Spark.

Mounted Directories#

The run docker script will mount the following directories on the host machine to the container:

  • Datasets: from host: $HOME/datasets to container: /datasets

  • Models: from host: $HOME/models to container: /models

  • Evaluation: from host: $HOME/eval to container: /eval

In our examples, we download input datasets and pre-trained models. It is useful to download these to a folder mapped on the host machine to avoid re-downloading between restarts of the container. These directories are configurable through argument to the run docker script.

For a full list of arguments see the run_docker.sh script at isaac_arena/docker/run_docker.sh.

Server-Client Mode With GR00T (Optional)#

We also support running the simulation and the policy model in separate containers connected via a lightweight RPC protocol. This is useful when the policy environment has different dependencies or needs to run on a different machine.

It requires two containers to run:

  • The Base Isaac Lab Arena container, started via docker/run_docker.sh.

  • A separate GR00T policy server container, started via docker/run_gr00t_server.sh.

A typical workflow is:

  1. Start the Base container for simulation and evaluation:

bash docker/run_docker.sh

  1. In a second terminal, start the GR00T policy server container:

bash docker/run_gr00t_server.sh \
  --host 127.0.0.1  \
  --port 5555 \
  --policy_type isaaclab_arena_gr00t.policy.gr00t_remote_policy.Gr00tRemoteServerSidePolicy \
  --policy_config_yaml_path {policy_config_yaml_path} # e.g. isaaclab_arena_gr00t/policy/config/gr1_manip_gr00t_closedloop_config.yaml
  1. Inside the Base container, run the evaluation script with a

    client-side remote policy (refer to Closed-Loop Policy Inference and Evaluation for full command lines).

This setup cleanly separates the Isaac Lab Arena simulation environment from the GR00T policy server environment.

If you want to host other policy models as remote servers, you can follow the same pattern: create a dedicated server Dockerfile and launcher script (similar to docker/Dockerfile.gr00t_server and docker/run_gr00t_server.sh), and point it to a custom ServerSidePolicy implementation as described in Remote Policies Design.