Egocentric Cable Routing and Wiring Video Data

Egocentric cable routing and wiring data is first-person video of a person routing, clipping, dressing, coiling, plugging or terminating cables and wires, recorded from the contributor’s own point of view with synchronized sensor streams alongside. The camera sits on the head, close to what a robot’s onboard camera would see over a bench, a rack or a wall, so the frames show what the hands are doing to the cable, the connector and the board rather than a distant view of someone standing near equipment. On this site it is a custom collection: you name the task, the setting and the cable work involved, and capture is run to a written skill spec.

The data is human capture. A paid, consented contributor wears the rig and performs the task. It is not robot or teleoperation data, and it is not a fixed corpus you download. This page is narrower than assembly task data, which covers fastening, measuring, press-fit and wire work at a bench as one family, and narrower than manufacturing workflow data, which covers whole production procedures. The subject here is the cable and wiring family of tasks: which of them the site states it can capture, how they are specified, what the site does and does not state about recording them, and how the episodes are labeled and checked before delivery.

People reach this topic through close cousins: cable routing robot training data, a wire harness video dataset, egocentric cable management data, or custom wiring task data collection. They describe one request from different angles. Whichever wording you start from, the work is the same: write down the task, collect it on demand from real benches, racks and venues, and receive validated episodes with the labels a training pipeline expects.

NVIDIA’s GR00T N1 paper describes a data pyramid with web data and human video at the base, synthetic data in the middle and real robot trajectories at the top. First-person human video of cable work sits in that base layer. The framing matters here because the demonstrations come from people doing the task in the kinds of settings where a robot would eventually work, which is what a human-video layer is for. Human capture does not replace robot data, and this page does not claim that it does.

Cable work is a revealing test of manipulation skill because the object is long, flexible and has no single shape. The site’s assembly page says that tasks such as wire strip-crimp-terminate live in finger-level detail that RGB video alone cannot resolve, and that each needs 3D hand pose tracked through reach, grasp and place. A clip that shows a hand somewhere near a cable does not carry that. The hands, the cable, the connector or terminal and the moment of contact all have to be present in the episode, and the episode has to say where one step ends and the next begins.

Breadth across tasks matters as much as volume on one of them. Generalist’s GEN-0 write-up makes a point about data in different words: "data quality and diversity matters more than sheer volume." For cable work that points at the spread of tasks and settings. Patching cables at a rack, routing a harness on a form board and running cable along a venue floor are different tasks with different geometry, and a corpus that repeats one bench does not cover them. The catalog lists cable and wiring rows across four sectors, which the task section below sets out.

Public corpora are captured for someone else’s task distribution. A custom collection lets you aim at the routing task, the termination step and the setting your model is missing instead of hoping a public release contains them. This page does not claim that any particular public corpus lacks cable work. It makes the narrower point that a custom spec lets you name what you need and have it captured, labeled and checked to terms you agreed.

Fifteen rows of the site’s commercial task catalog name a cable, conductor, wire or wiring step. Each is listed with a skill level, a capture setup and a status exactly as the catalog states them, and the first table below reproduces them. Two further rows, HVAC equipment installation and bicycle repair, mention wiring or cables as one step among many and are left out of the table because the cable work is not their subject.

On the bench, the catalog lists routing cables to a form board, routing conductors to a form board and clipping them as a harness, coiling and strapping a cable, and crimping a terminal onto a wire. These sit in the cloth-and-cable bench and the precision assembly bench for electronics, use a multiview rig with handheld grippers, and are marked available on request. The routing and harness rows are listed as medium skill, and crimping a connector is listed as hard.

The data center and IT hardware bench is described as a server rack plus hardware bench with modules, cabling and drivers. The catalog lists dressing a cable bundle to uniform pitch with ties, patching Ethernet cabling by inserting and routing patch cables, and seating power cables and locking clips. All three are listed as easy, use a multiview rig with handheld grippers, and are marked available on request.

In event venues the catalog lists AV and cable management, and routing and taping AV cables with cable ramps and gaffer tape. Both use a head-mounted phone or multiview rig and are marked ready to deploy. In construction and trades the catalog lists electrical installation on an active construction site, described as running cable, mounting boxes and wiring outlets, switches and fixtures, which is marked active collection. It also lists outlet and fixture wiring, stripping and wiring outlets, and thermostat and control wiring in trade service calls, all hard and ready to deploy, and wiring an outlet at the trades workshop bench, hard and available on request. Device manufacturing in electronics manufacturing and e-waste, which includes fitting boards and cables, completes the list.

The catalog is a list of what is captured or available, not a ceiling on what can be briefed. A cable type, routing pattern, termination step or setting outside these rows is a skill-spec question to settle in a Spec Pilot rather than an assumption. The site does not describe a cable length, a conductor gauge, a connector family list or a harness size for any of these rows, so none is stated here.

A cable and wiring collection uses the same capture stack as every Firsthand program. The instrumented tier is a head-mounted stereo rig that records first-person RGB, with a metric depth stream recovered from the calibrated stereo rig rather than from a standalone depth sensor, and 6-DoF IMU data at 200 Hz from head- and wrist-mounted sensors. The wider stack is ten streams: a 4K head camera, two wrist cameras, depth, IMU, 3D hand and body pose, instance segmentation and action labels, hardware-aligned under 2 ms. The two wrist cameras give a second, closer view of the hands than the head camera alone, which matters when a hand closes around a connector or a clip.

Every stream is hardware-triggered onto one PTP clock domain, so depth, motion and hand pose line up with the same video frame rather than being approximated afterwards. The rig is calibrated twice per session, before capture starts and after it ends, and an episode is accepted only if RMS reprojection stays under 0.28 px. The catalog lists the capture setup per task: a multiview rig with handheld grippers for the bench and data center rows, and a head-mounted phone or multiview rig for the venue, construction and trade rows. Which tier applies to your program is set in the skill spec.

On depth, the site states a limit for small parts: small fasteners can fall below a depth sensor’s reliable floor, so metric depth on sub-centimetre parts is flagged rather than interpolated, while hand pose stays valid at that range. That statement is about small fasteners and parts. The site does not state how depth behaves on a thin cable, a conductor or a terminal, so this page does not either. Force and tactile readings are not among the streams, and no robot or teleoperation data is captured.

Each attempt is delivered as one episode. Inside it, action segments divide the episode into start- and end-timestamped spans, each carrying start_ns, end_ns, verb and noun. For this task family a segment can be one step as written in the skill spec, so a harness run becomes a sequence of labeled steps rather than one long clip. The other standard fields are hand_joints and hand_visible for 3D hand pose, contact_events for the moment the hand meets the item, objects, and instance_masks, stored at the depth resolution of 848×480 so they line up with the depth stream.

Hand pose is delivered for both hands with a per-joint visibility flag for occluded fingers. That is the label that carries the two-handed structure of cable work: where the hand that holds the cable sits relative to the hand that works the clip or terminal. Contact events mark when a hand meets the cable, the connector or the board, and action segments mark where one step ends and the next begins.

Masks need a spec decision. The site states that clothing, bedding and bags do not track well with masks propagated frame to frame, so deformable captures ship per-frame instance masks where the spec calls for them. Cables are not named in that statement. Whether a cable task should ask for per-frame instance masks is therefore something to settle when the spec is written, not a promise this page makes.

The site does not list labels for the quality of a result. There is no stated label for whether a cable is routed correctly, whether a connector is fully seated, cable tension, wire color or pin-out, and no stated electrical test or continuity result. Two catalog task descriptions mention a voltage tester, but that describes the task the contributor performs, not a delivered measurement. The verb and noun on each action segment is the vocabulary the skill spec defines with you, and anything beyond it is a scoping conversation, not a delivered field.

Several limits the site already states apply directly to cable and wiring capture, and several it leaves open. Keeping the two apart is the point of this section. The stated limits are about depth on small parts, masks on deformable items, hand occlusion and the conditions that fail an episode. Each is something you can plan around, and each is stated elsewhere on the site rather than introduced here.

Hand occlusion is the first. Fingers disappear behind objects in manipulation, which is why 3D hand pose carries a per-joint visibility flag. A hand closing around a connector or a clip is an everyday case in this family, and the flag marks the joints that were not visible instead of presenting a guess as a measurement. The second is depth on small parts, covered above: metric depth on sub-centimetre parts is flagged, not interpolated. The third is masks, where per-frame instance masks are the stated answer for items that do not track well when propagated, and the spec is where you decide whether a cable needs them.

The fourth is the set of machine-checked reject criteria. An episode is rejected for a sync breach beyond the ceiling for more than three consecutive frames, calibration drift beyond tolerance, the manipulated object out of frame for more than 15 percent of the episode, more than 35 percent invalid depth pixels in the manipulation region, an undocumented bystander, or a mismatch against the named skill spec. A human reviewer then checks each episode against the task spec.

What the site does not state is just as useful to know before you scope. It gives no depth behavior for a thin cable, no per-frame mask accuracy figure for cables, no cable-path or three-dimensional cable-state label, and no routing-correctness label. If a policy of yours depends on any of those, say so at the start, because it changes what the spec has to ask for and what you need to supply or derive yourself.

Validation runs against signed accept criteria written into the spec before capture. Each batch goes through per-batch QA and a reject log ships with it. Only footage that clears the criteria is billed as a validated hour, and rejected items are recaptured at our cost. Many teams start with a Spec Pilot, a small, fast first batch that proves the spec and the accept criteria on real footage before volume scales.

Capture happens in working or lived-in settings, from benches and racks to venues and trade sites. Every contributor signs a written release before capture, and each delivered episode links to that record through consent artefact IDs in the batch data card. The consent chain has six checkpoints, from before capture to after delivery, and a break at any one discards the footage rather than shipping it at a lower grade. At ingest, automated detection finds faces, plates and screens and blurs them irreversibly, and speech is scrubbed of names, addresses and card numbers while ambient audio is kept. A human reviewer confirms that no un-consented person is identifiable in any frame, and footage that cannot be de-identified is discarded. The anonymization record is delivered with the data.

On safety and qualification, the site states nothing about wiring. It does not state a contributor safety protocol, an electrical qualification, a licensing requirement or any safety certification for the outlet, fixture, thermostat or installation rows, and this page does not describe one. The catalog descriptions of two rows mention testing with a voltage tester, and one of them says with power off. That is the task as listed, not a safety claim. If your program needs terms about who performs the work and under what conditions, raise them when you scope it. None of this is legal advice, and the executed agreement controls.

From signed spec to first validated batch, the timeline is a median of 48 hours where coverage is deep and three to four weeks where it is thin and contributors need to be recruited first. Speed comes from a warm pipeline, not from cutting consent, review or anonymization. After the first batch, on-demand top-ups run against the same accept criteria, so you can add more tasks, settings or variation without rewriting the spec. The site does not state a turnaround for those later batches, so plan them in your quote.

Delivery is in the format your stack already reads: RLDS, LeRobot, WebDataset, HDF5, zarr or Rerun .rrd, or raw streams with metadata. Readable converter source ships alongside the data so the schema can be retargeted to your own loader. Collections are billed per validated hour.

The license is perpetual, irrevocable and worldwide, with the data buyer-owned and non-exclusive by default. Training, evaluating, fine-tuning and deploying models, including commercial ones, is permitted, and exclusivity is available as a separate term. The executed agreement controls. The site does not publish a price list, a minimum order or a batch size for this task family, so those are set in your quote.

Where this site does not state something, this page says "not stated" rather than filling the gap. The table below separates what is delivered from what is not stated. There is no listed label for routing correctness, connector seating, cable tension, wire color or pin-out, no electrical test or continuity result, no force or tactile stream, and no cable length, gauge or connector list. Nothing here claims a training outcome or a model improvement.

The tradeoff to plan for is the embodiment gap between a human hand and a robot end-effector, which retargeting or pairing with robot data has to close. What you get is human first-person demonstrations in which the hands, the cable and the step structure are explicit. Counts of benches, contributors, hours and tasks are not fixed numbers on this site. They are set in your skill spec.

Describe the routing or wiring tasks, the setting, the termination or connection steps and the failure or variation cases your pipeline is missing. That becomes a written skill spec, quoted for reach, timeline and price before any capture begins, and most programs run a Spec Pilot first. The scoping checklist below maps the parts of a skill spec onto a cable request. A published sample recording is on the sample data outputs page if you want to see the streams before you write a spec.

Three choices matter most for this family. Name the step and the item rather than a broad activity, because that is what lets each step become one action segment. Name the setting and the capture tier, since a bench, a rack and a venue floor present different geometry and the catalog lists a different capture setup for each. And name the lighting and participant mix: daylight coverage fills fastest, while evening, low-light capture is the harder cell to fill.