Collect Training Data with Behaviors

Required Version

This feature will not be released until MoveIt Pro version 10.0.

Use data-recording Behaviors when you want an Objective to collect its own demonstrations. This is useful for a repeatable scripted task or an oracle policy: the Objective starts Trainer, runs the task, keeps a successful episode, and converts the dataset without manual timing.

Behavior	Role in the Objective
RecordEpisode	Starts a Trainer recording with a saved Training Config.
SaveEpisode	Keeps the current demonstration after the task succeeds.
StopRecording	Finishes the session, or safely discards an unsaved failed attempt.
ConvertDataset	Starts a background conversion to LeRobotDataset v3.0.

Prerequisites

• Complete Collect VLA Training Data with Trainer first. You need a saved Training Config with working robot-state, command, and camera topics.
• Choose a runnable task Objective to record. This example wraps Pick April Tag Labeled Object and uses the lab_sim Training Config.

1. Build the wrapper Objective

Create an Objective and add a Fallback with two branches:

1. In the first Sequence, add these nodes in order:
   • RecordEpisode
   • WaitForDuration
   • your task as a Subtree
   • SaveEpisode
   • WaitForDuration
   • StopRecording
   • ConvertDataset
2. In a second Sequence, add StopRecording followed by AlwaysFailure.

The second branch is deliberate cleanup, not a forced success. If the task fails before SaveEpisode, StopRecording discards that unsaved attempt and AlwaysFailure preserves the Objective's failed result. A failed pick is never labeled as a successful demonstration.

Set the main ports as follows:

Node	Port	Example value
RecordEpisode	dataset_name	behavior_bottle_pick
RecordEpisode	task	Pick up an AprilTag-labeled bottle.
RecordEpisode	config_name	lab_sim
RecordEpisode	num_episodes	1
first WaitForDuration	delay_duration	2.0
second WaitForDuration	delay_duration	1.0
ConvertDataset	dataset_name	behavior_bottle_pick

The short waits let recording settle before motion begins and let the saved episode finish updating in Trainer before the session closes. Leave camera_topics empty to use the cameras from the named Training Config.

Collapse the task Subtree after adding it. The editor then emphasizes the recording lifecycle while the complete picking task remains reusable behind one node.

2. Record the task

Open Trainer and leave the Record tab visible, then select Run on the wrapper Objective. Do not select Start recording manually; RecordEpisode starts the same recording session and its status appears live in Trainer.

The task Subtree now performs the demonstration. SaveEpisode runs only after that Subtree succeeds, so the episode contains a completed bottle pick rather than an unfinished attempt.

When the run completes, StopRecording finishes the session and registers the episode in the dataset.

3. Review the episode

Open the Dataset tab and select the play button on the episode. Use play/pause and the timeline to inspect the recorded motion. Select Stop in the Playback mode card to return to the live robot.

Playback only reviews recorded data; it does not command the robot.

4. Confirm the conversion

ConvertDataset uses the same dataset_name as RecordEpisode. It derives the output settings from the recording and creates a sibling dataset named <dataset-name>-lerobot.

Conversion runs in the background. The Behavior succeeds when the conversion job is accepted, so open the Convert tab to monitor it through completion.

The converted dataset contains LeRobot data, meta, and videos folders at:

<TRAINER_DATA_ROOT>/recordings/behavior_bottle_pick-lerobot

<TRAINER_DATA_ROOT> is MOVEIT_PRO_TRAINER_DATA_ROOT when set, or ~/.local/share/moveit_pro/trainer by default.

For a larger oracle-data collector, start RecordEpisode once with num_episodes set to the batch size. Inside a loop, randomize the scene, run the task Subtree, call SaveEpisode only after a successful rollout, and reset the robot for the next attempt. Place StopRecording and ConvertDataset after the loop so the complete batch is closed and converted once. Keep the failure cleanup branch so a failed rollout is discarded and the batch stops for inspection.