Egocentric & Wearable Capture

Egocentric and wearable capture is the most accessible way to earn money from the robot boom without owning a robot, moving to a lab city, or holding a degree. You wear a camera, record yourself doing ordinary physical tasks (folding laundry, chopping vegetables, wiping a counter, assembling flat-pack furniture) from your own point of view, and upload the clips. Companies building humanoid and home robots buy that first-person footage because it is the closest thing they have to showing a robot what a task looks like through human eyes. If you wonder what that footage is worth to the robot makers who buy it, AIData.Broker's page on physical-world data, which we also publish, lists public prices for first-person video in bulk and captured to a spec, shows how far bulk prices have fallen, and notes that anyone else in frame must consent. This guide explains why the work exists, what it actually pays (honestly, by region), the gear you record with, and the exact steps to start from zero. For live openings and platform-by-platform pay, see the egocentric capture jobs and pay hub.

1. Why first-person video is useful

A first-person camera records the objects a person reaches for, the order of an action and how the scene changes. Robot researchers can use such recordings to study tasks and representations. Human video does not itself contain a robot's joint commands, so it is different from a teleoperated robot demonstration and requires its own processing.

Figure's Index announcement, published August 25, 2026, describes collecting everyday task recordings and filtering, reviewing and annotating them. This establishes the program's physical-data purpose. Its announced cumulative creator payout does not establish what an individual contributor will earn.

A recording project needs useful variation and clean, relevant footage. Follow the actual brief, record only where you have permission, and check privacy and acceptance terms before including a home, workplace or another person in a clip.

2. What the work actually is, day to day

The loop is simple and repeats: put on the device, record a task the way the project brief specifies, stop, review the clip, and upload it for approval. A typical session runs one to a few hours and produces a batch of short takes rather than one long film. You are usually recording common household or workplace actions, because those are the tasks robot makers want most.

Real programs describe it plainly. Micro1's remote contributors mount an iPhone on their head and record themselves folding laundry, washing dishes, and cooking, all self-recorded at home. Perle's contributors wear a headstrap with their phone, press record, and go about their day. The briefs get specific: keep your hands in frame, do the task at a natural speed, film in good light, avoid clutter that confuses the labelers, and follow the exact object list the project cares about (for example, "fold five different shirts, one take each").

Two rules show up on almost every project and you should expect them. First, privacy: you are typically asked not to show your face or reveal personal details like names, addresses, or documents, and reviewers strip anything that slips through. Second, approval: you are paid for footage that passes review, so learning the acceptance bar (section 6) is the difference between a decent hourly rate and wasted evenings.

3. The gear you record with

The single best piece of news for beginners is that most remote programs either provide the hardware or let you use a smartphone you already own. You rarely need to buy anything to start.

Project Aria is worth understanding even though you cannot buy a pair. It is Meta's egocentric research device, given to over 200 academic and corporate partners, and it is the tool a lot of the frontier datasets are captured on. If you are a student or researcher, applying to the Project Aria research kit is a real path. For everyone else, the paid gig path uses a phone or a provided kit, and that is by design: the whole point of the crowd model is that the barrier is a smartphone, not a lab.

4. What it really pays (and why geography decides so much)

Read from today's open roles. 134 open roles on RobotTraining.jobs are in egocentric and wearable capture, and 29 of them state an hourly rate. Those rates typically run $6 to $25 an hour, with a median of $10 an hour. They come from individual postings with different duties and locations, not from a category-wide wage survey; the pay explorer and pay report show the full measured sample and its coverage.

An advertised hourly amount is gross compensation, not verified take-home earnings. Salary, equity, benefits, overtime and contractor expenses must be compared separately. We do not infer what a beginner or an expert earns by splitting a pay distribution into tiers.

For accepted-footage or task payments, include setup, recording, upload, review and any rework in your time calculation. An hour of accepted video is a unit of output, not necessarily an hour of labor. There is no guaranteed volume of accepted tasks or paid hours.

5. How to start with zero experience

You do not need robotics knowledge, coding, or a portfolio. You need a smartphone, a private space to film, and the willingness to follow instructions precisely. Here is the fastest honest path.

1. Pick one remote-first program and apply. Start with Micro1's contributor signup if you want at-home work open to dozens of countries, Perle if your country is on its list (it includes the US, Canada, Brazil and Indonesia), or Figure's Index app if you want the recording device provided (expect a waitlist). Applying is usually a short form plus a sample clip, not an interview.
2. Do the onboarding task exactly as written. Every program starts you with a small qualification batch. Read the brief twice, match the object list, and film in bright, even light with your hands clearly in frame. This first batch decides whether you get real work.
3. Set up a clean capture spot. A tidy kitchen or table, daylight or a lamp, and a plain background. Remove anything with personal information (mail, screens, documents) from the frame before you record.
4. Record short, deliberate takes. One task per clip, natural speed, no talking unless asked. Redo a take rather than uploading a shaky or half-obscured one.
5. Learn each project's acceptance bar, then batch your work. Once you know what passes, your approval rate climbs and your effective hourly rate follows. Ask the project coordinator what the top rejection reasons are. Recorders also compare notes on acceptance waits, rejections and pay in the AITraining.jobs forum, a community we also run.
6. Add a second and third platform. Availability is bursty, so contributors who earn steadily are usually signed up on several platforms (for example the Atlas Capture iPhone app, or DoorDash Tasks in the US states it lists) and take whichever project is live.

6. How to keep your footage from getting rejected

Because you are paid for approved video, quality control is your real skill here, and it is learnable in a week. The failures that get clips rejected are consistent: motion blur from moving too fast, hands leaving the frame during the important moment, poor lighting, a cluttered or reflective background, a face or personal detail caught on camera, or drifting off the exact task the brief asked for. Fix those and you are most of the way to a good rate.

Treat it like a craft. Keep the object you are manipulating and both hands inside the frame through the whole action, move at a steady natural pace, and film the same task a few times so the buyer gets clean variety. Check the specific offer for payment terms; a general rate is not established.

7. Who this work suits, and who should skip it

Egocentric capture fits you well if you want flexible, work-from-home income, you are patient with detailed instructions, and you have a private space and a decent smartphone. It suits students, caregivers at home, people between jobs, and anyone in a lower-cost country who wants remote work that pays in hard currency (while being clear-eyed that offshore rates are low). It is genuinely open: no degree, no experience, no robot required.

It is a poor fit if you need a predictable full-time paycheck, since work comes in waves and per-approval pay is variable. It is also not the category to choose if you want the highest robot-training rates. Check the specific offer for payment terms; a general rate is not established. Many people start with egocentric capture because it is the easiest door to walk through, then move to teleoperation or embodied-AI data annotation once they know the field.

8. Where to apply and what to read next

If you want to begin today, apply to Micro1 for at-home capture in many countries, Perle if your country is on its list, and Figure Index or Atlas Capture to widen your pool of live projects. Scale AI's physical-AI program and Appen also run first-person collection among their broader data work.

When you are ready to compare numbers, the egocentric capture pay and jobs hub tracks live openings and rates, and the pay explorer lets you compare this category against every other kind of robot-training work. Egocentric capture is the on-ramp: the lowest barrier, the most global reach, and a real look at whether hands-on robot data work is for you before you invest in a higher-paying, higher-commitment role.