Command Line Reference

Version notes (important, please read first)

Starting from LeRobot 0.6.0, a trained model must be deployed with lerobot-rollout. The old lerobot-record --policy.path=... syntax was already removed in version 0.5.2.

In step 1 of this tutorial, LeRobot was installed with git clone, which gives you the current latest version, so please use the lerobot-rollout command line below. If you insist on using lerobot-record, the program will directly raise an error and prompt you to switch to lerobot-rollout.

The division of labor between the two commands is as follows:

• lerobot-record: only responsible for collecting demonstration data (it is what step 7 uses), and it now rejects dataset names starting with eval_
• lerobot-rollout: responsible for deploying a trained model, using --strategy.type to choose the working mode

rollout command line parameters

Parameter	Description
--strategy.type	The working mode. base only runs the model without recording data, for checking the results on site; episodic records by episode with a reset phase, behaving close to the old lerobot-record
--policy.path	The model path, pointing to checkpoints/last/pretrained_model in the training output
--task	The task description, used together with --strategy.type=base
--duration	The number of seconds to run; 0 means no time limit
--interactive	Add this when you need to take over midway; you can use commands such as /stop and /reset in the terminal
--display_data	Whether to start the rerun.io visualization interface
--policy.device	The compute device, such as cuda, cpu

Command Line Reference

With real-time visualization: --display_data=true

Without real-time visualization: --display_data=false

When --display_data=true, the cool rerun.io visualization interface starts, but every frame is saved as an image under the /Users/tommy/.cache/huggingface/lerobot/rollout_lerobot_my_dataset_a/images/observation.images.front/episode-000000 directory, which takes up a lot of space. Later you can set it to --display_data=false

Inference with a model on a HuggingFace model Repo: --policy.path=Tommymy/lerobot_my_model_a

Taking the grab-oranges task as an example

• Inference with a local model (with real-time visualization)

lerobot-rollout  \
  --strategy.type=episodic \
  --robot.type=so101_follower \
  --robot.port=/dev/tty.usbmodem5AAF2193061 \
  --robot.cameras="{ front: {type: opencv, index_or_path: 0, width: 1920, height: 1080, fps: 60, fourcc: "MJPG"}}" \
  --robot.id=my_follower_arm \
  --display_data=true \
  --dataset.repo_id=Tommymy/rollout_lerobot_my_dataset_a \
  --dataset.single_task="Grab Oranges" \
  --dataset.episode_time_s=1000 \
  --policy.path=/Users/tommy/Downloads/7-lerobot/checkpoints/last/pretrained_model

• Inference with a local model (without real-time visualization)

lerobot-rollout  \
  --strategy.type=episodic \
  --robot.type=so101_follower \
  --robot.port=/dev/tty.usbmodem5AAF2193061 \
  --robot.cameras="{ front: {type: opencv, index_or_path: 0, width: 1920, height: 1080, fps: 60, fourcc: "MJPG"}}" \
  --robot.id=my_follower_arm \
  --display_data=false \
  --dataset.repo_id=Tommymy/rollout_lerobot_my_dataset_a \
  --dataset.single_task="Grab Oranges" \
  --dataset.episode_time_s=1000 \
  --policy.path=/Users/tommy/Downloads/7-lerobot/checkpoints/last/pretrained_model

• Inference with a model on a HuggingFace model Repo

lerobot-rollout  \
  --strategy.type=episodic \
  --robot.type=so101_follower \
  --robot.port=/dev/tty.usbmodem5AAF2193061 \
  --robot.cameras="{ front: {type: opencv, index_or_path: 0, width: 1920, height: 1080, fps: 60, fourcc: "MJPG"}}" \
  --robot.id=my_follower_arm \
  --display_data=true \
  --dataset.repo_id=Tommymy/rollout_lerobot_my_dataset_a \
  --dataset.single_task="Grab Oranges" \
  --policy.path=Tommymy/lerobot_my_model_a

The model will be downloaded after running