Skip to main content
Data collection runs across two machines on the same network:
  • Main host (the upper computer) — owns the dataset, runs teleop and the recording loop.
  • ZED box (the Jetson sender) — streams the three ZED-X One cameras.
Both machines must agree on the time (via PTP) so each recorded camera frame is aligned with the joint sample taken at the same instant. Each machine therefore runs two long-lived processes, each in its own terminal, that stay up for the entire session.
Start the clock-sync daemon on both machines before launching collect-data. If PTP isn’t running, the receiver’s capture/decode skew check in ZedCamera.connect() warns and recorded camera/joint pairs will not be time-aligned. See zed.sync-clocks for the PTP canary details.

​
Network setup

The two machines must be able to reach each other over the network. Configure each NIC with an IP however you like — DHCP on a shared switch, or static IPs (e.g. 192.168.10.1/24 on the ZED box and 192.168.10.2/24 on the main host) over a direct ethernet cable. Find the ZED box’s IP (ip addr) and pass it to the main host as --robot_config.zed_host <zed-box-ip>. Interface names vary by machine — the link might be eth0, enp3s0, eno1, and so on. Run ip link (or ip addr) on each machine to find the interface carrying the link, and substitute it for <iface> in the commands below. PTP runs over whichever interface carries the link (zed.sync-clocks --iface). If the two machines are not on the same L2 segment (e.g. separated by a router rather than a switch), add --transport udpv4 to the clock-sync commands.

​
Prerequisites

  • The lerobot and video extras installed on the main host. The quick install includes both; on a development install run uv sync --extra lerobot --extra video (see Installation).
  • CAN set up on the main host (can.setup) and motors verified (motor.info).
  • pyzed installed on the ZED box (zed.install) and the three camera serial numbers on hand.
  • A network link between the two machines — either a direct ethernet cable or a shared router/switch (see Network setup).

​
ZED box (Jetson)

Open two terminals on the ZED box.
1

Start clock sync (slave)

axol zed.sync-clocks --role slave --iface <iface>
Replace <iface> with the ZED box’s link interface (find it with ip link). Leave this running for the whole session. See zed.sync-clocks for transport and timestamping options.
2

Stream the cameras

axol zed.stream \
    --overhead 12345678 \
    --left-arm 23456789 \
    --right-arm 34567890
Streams from the NIC’s current IP until Ctrl+C. Note that IP — you’ll pass it to the main host as --robot_config.zed_host. If the overhead is a stereo ZED X, add --overhead-stereo (and set --robot_config.cameras.overhead.stereo true on the main host — the dataset then records overhead_left / overhead_right). See zed.stream for resolution, fps, and bitrate flags.

​
Main host (upper computer)

Open two terminals on the main host.
1

Start clock sync (master)

axol zed.sync-clocks --role master --iface <iface>
Replace <iface> with the main host’s link interface (find it with ip link). The master role runs on the long-lived upper computer that owns the dataset. Leave it running for the whole session.
2

Run data collection

axol collect-data \
    --repo_id myorg/pick-place \
    --task "Pick the red cube and place it in the bin" \
    --robot_config.zed_host 192.168.1.42
Replace 192.168.1.42 with the ZED box’s IP.The robot is driven over VR (authorize the TLS certificate first — see the Teleoperation quickstart). See collect-data for the full field list, including --fps, --teleop_hz, stiffness, and --push_to_hub, and Command configuration for the draccus override syntax.

​
Controller layout

Episodes are driven entirely from the Quest controllers. The movement controls are the same as in teleoperation; data collection adds the A (record) and B (mode) buttons. Quest controller diagram
#ButtonAction
1Left gripPress both grips (1 + 2) together to enable arm tracking; press either alone to disable (toggle, not hold)
2Right gripSee above
3Left triggerActuate left gripper
4Right triggerActuate right gripper
5XReset to rest pose. During a recording, stops and discards the take (re-record); during the start countdown, cancels it
6ARecord: start a take (3-second countdown), or stop and save the current take
7YExit the VR session
8BToggle between Teleop and Data Collection mode (disabled while recording or counting down)
—Right thumbstickFlick to switch camera view (overhead / wrists / split); click to reset the screens — see Camera views
The camera feeds are relayed to the headset during collection (requires the video extra on the main host), so you can line up takes from the robot’s point of view.

​
Recording an episode

1

Enter Data Collection mode

Press B to switch from Teleop into Data Collection. You can drive the arms in either mode, but a take can only be started from Data Collection.
2

Start a take

Press A to begin recording — a 3-second countdown runs (the headset stays in Data Collection), then frame capture begins. Press A again during the countdown, or X, to cancel.
3

Finish the take

  • Press A to stop and save the episode.
  • Press X to stop and discard it (re-record).
While the episode is written to disk the headset enters the Saving state and every control except Y (exit) is blocked. When the write completes the headset returns to Data Collection and the arms move back to the rest pose, ready for the next take.
Headset stateMeaning
TeleopDriving the arms; recording not armed
Data CollectionDriving the arms; ready to record
RecordingCapturing frames to the dataset
SavingEpisode being written to disk; controls blocked
Collection resumes from an existing dataset at --root; an aborted session that saved no episodes is cleaned up on shutdown. Press Ctrl+C in the collect-data terminal to finish.
The full controller bindings and headset state machine are documented in the VR Interface guide. The interface source lives in this repo under web/.
Prefer a browser to juggling terminals? The web control panel runs both clock-sync/streaming and collect-data from one page — connect the ZED box once and press Start.

​
Next steps

Policy Inference

Run a trained policy autonomously across the same two machines.

collect-data reference

Every flag and the dataset capture internals.