Server-Client Mode With GR00T#

Arena supports 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. You can see more details in the Remote Policies Design page.

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 \
  # The models directory is the directory that contains the GR00T model checkpoint.
  -m ${MODELS_DIR} \
  --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.

python -m isaaclab_arena/evaluation/policy_runner.py \
  --visualizer kit \
  --policy_type isaaclab_arena.policy.action_chunking_client.ActionChunkingClientSidePolicy \
  --remote_host 127.0.0.1 \
  --remote_port 5555 \
  # for example: 2000
  --num_steps ${NUM_STEPS} \
  # for example: 10
  --num_envs ${NUM_ENVS} \
  --enable_cameras \
  --remote_kill_on_exit \
  # for example: gr1_open_microwave, put_item_in_fridge_and_close_door, etc.
  ${ARENA_ENVIRONMENT_NAME} \
  # for example: gr1_joint, g1_wbc_joint, etc.
  --embodiment ${ROBOT_EMBODIMENT_NAME} \
  # for example: cracker_box, ketchup_bottle_hope_robolab, ranch_dressing_hope_robolab, etc.
  --object ${OBJECT_NAME}

This setup sets the Isaac Lab Arena simulation environment to the specified environment and runs the GR00T policy in a separate container. The simulation environment and the policy model communicate via the remote policy interface.

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.