Captured on shift, not staged
Live fulfilment space has mixed lighting, multi-person traffic, and real failure cases mis-picks, dropped totes, jams that a mock aisle never reproduces.
Warehouse & logistics
Warehouse Robotics Training Data
Pick, scan, pack and pallet-build inside working fulfilment space, captured on shift rather than in a mock aisle. We collect it egocentrically, against a spec you write.
- Streams
- Ten, synced under 2 ms
- First batch
- Median 48 h
- Billing
- Per validated hour
- License
- Buyer-owned
Requirements
What does warehouse robotics data need?
A named pick-and-pack skill list
Single-item pick, barcode scan, carton break-down, pallet stacking, label application, and two-person carry each need their own coverage, not one generic "warehouse" label.
Depth that holds at range
Pick aisles have long sight-lines. Distance-invalid pixels beyond reliable depth range are flagged rather than interpolated, so you know exactly where the depth stream stops being trustworthy.
Rights you can ship on
A perpetual, buyer-owned commercial license backed by a documented consent chain, so a policy trained on the data can go into a product.
Why it matters
Why is warehouse data hard to capture well?
Incidental warehouse footage in public corpora is unsynchronized across streams and carries no manipulation-grade hand pose, because it was never captured against a named pick-and-pack skill list in the first place.
Live fulfilment space also means mixed sodium and daylight lighting, cabling routed under hi-vis PPE that shifts wrist extrinsics, and genuine multi-person traffic — all of which change how a rig has to be run and recalibrated session to session.
Capture setup
How is warehouse robotics data captured egocentrically?
A head-mounted camera keeps the pick, scan, and place motion in frame from the viewpoint a warehouse policy will actually see from, through pick aisle, pack bench, pallet build, and returns desk work.
Depth, hand pose, and action segments are aligned to video on one hardware clock. Per-session recalibration accounts for cabling routed under PPE, and long sight-lines beyond reliable depth range are flagged rather than interpolated.
The spec
How does a custom warehouse robotics spec work?
You describe the pick-and-pack tasks your policy needs. We turn it into a written spec covering tasks, environments, capture tier, annotation depth, and accept criteria, then quote reach, timeline, and price against that document before any capture begins.
The collection runs through the same seven-stage end-to-end custom collection pipeline as every Firsthand program: contributor recruiting, capture, per-batch QA, anonymization, and licensed delivery. Rejected items are recaptured at our cost.
Fast · on demand
How fast does a first batch ship?
Where coverage is already deep, the first validated batch lands in a median of 48 hours from signed spec. Thin environments typically take three to four weeks, because vetting operators there is the bottleneck, and we quote that timeline up front.
On-demand top-ups run against the same spec and accept criteria after each training run, so new warehouse data stays consistent with what you already hold.
Rights
What do licensing and consent look like?
Every custom collection ships under a perpetual, buyer-owned commercial license, and exclusivity is available so the data is never resold. Each episode links to a signed contributor release through a documented consent chain.
Faces, plates, and screens are detected and irreversibly blurred at ingest, and a reviewer confirms the result before anything ships. The anonymization record is delivered with the data.
Warehouse task vs delivery
What does a warehouse robotics policy get, task by task?
| Warehouse robotics need | What Firsthand captures |
|---|---|
| Pick and place in frame | Head-mounted camera keeping the pick, scan, and place motion visible through the task cycle |
| Named skill coverage | Single-item pick, barcode scan, carton break-down, pallet stacking, label application, two-person carry |
| Hand-level detail | 3D hand pose tracked through reach, grasp, and place |
| Depth at range | Synchronized depth stream, with distance-invalid pixels beyond reliable range flagged rather than interpolated |
| Action structure | Instance segmentation, action segments, grasp and contact events |
| Quality | Measured against signed accept criteria; billed per validated hour; reject log with every batch |
| Formats | RLDS, LeRobot, WebDataset, plus HDF5, zarr, and Rerun .rrd |
Every collection runs through the same seven-stage end-to-end custom collection pipeline.
FAQ
Warehouse robot training data, answered.
- What is warehouse robotics training data?
- It is real-world, first-person demonstration data of pick, scan, pack and pallet-build tasks captured against a written spec: egocentric video with depth, hand pose, and action segments. Firsthand handles contributors, capture, QA, consent, and delivery.
- How is this different from public warehouse footage?
- Public logistics footage is largely incidental rather than spec-driven, unsynchronized across streams, and carries no manipulation-grade hand pose. Custom warehouse robotics data is captured against a named pick-and-pack skill list, hardware-synchronized under 2 ms.
- Is capture done in a live warehouse or a mock aisle?
- Live, working fulfilment space, captured on shift — pick aisle, pack bench, pallet build, and returns desk — rather than a staged mock aisle.
- How is custom warehouse data billed?
- Per validated hour, quoted against the written spec before any capture. You pay only for data that clears the accept criteria, and rejected items are recaptured at our cost.
- Which delivery formats are supported?
- RLDS, LeRobot, and WebDataset, plus HDF5, zarr, and Rerun .rrd. Converter source is readable, so you can retarget the schema to your own loader.
- Can we start with a small pilot?
- Yes. A Spec Pilot is a small, fast first batch that proves the spec and accept criteria before you scale volume on demand.
Sources
Where are the delivery formats documented?
- RLDS (Reinforcement Learning Datasets) (opens in a new tab)Official repository for the RLDS episode format. github.com
- LeRobot (opens in a new tab)Official Hugging Face repository for the LeRobot dataset format and tooling. github.com
- LeRobot documentation (opens in a new tab)Hugging Face documentation for loading and using LeRobot datasets. huggingface.co
- WebDataset (opens in a new tab)Official repository for the WebDataset sharded format. github.com
Request a custom dataset.
Tell us the pick-and-pack tasks your policy needs to perform. We will quote reach, timeline, and price against a written spec, with a first validated batch in a median of 48 hours where coverage is deep.