Ordinary people training robots is becoming a concrete workflow.
Robotics data collection used to depend on labs, expensive devices, and specialized teams. CHEK EGO Miner lowers that entry point: with a phone, a computer, and a stable mount, a collector can record real first-person work or daily actions and enter an embodied-AI EGO data task flow.
CHEK EGO Miner is brought by the CHEK Robot（擎控机器人） team. The mobile app is called CHEK, and the desktop app is called Qingkong Miker.
The goal is not to create another video-recording tool. The goal is to organize real human actions, viewpoints, decisions, and operating experience into a data supply network that robots can learn from.
Washing dishes, organizing a desk, sorting objects, carrying items, opening doors, assisting mobility, or operating tools used to be everyday experience. In the embodied-AI era, those actions can become training material for robots that need to understand the physical world.
The internet era turned human text, photos, and short videos into training material for AI that speaks and generates content. Embodied AI now needs human actions, viewpoints, and experience so future robots can learn how real people handle real tasks.
CHEK EGO Miner opens that entry point: more people can use the phone and computer they already have to participate in the next robot-data network.
A normal video ends when it is shot. EGO data collection tries to turn a real human task into a data asset that can be uploaded, reviewed, searched, validated, reused, and settled according to task rules.
The workflow cares about more than “is there footage”:
- which task and scene the collector is working on;
- whether the CHEK mobile app records a stable first-person view;
- whether Qingkong Miker on desktop manages the task, devices, upload, and review flow;
- whether the action is complete, real, and contextual;
- whether privacy, consent, quality, and delivery checks are satisfied;
- whether the resulting data can be searched, reused, and rewarded according to task rules.
Robots do not only need polished demos. They need large amounts of messy, contextual, real-world human-action data.
Labs can reproduce standard actions, but they struggle to cover long-tail reality: cluttered desks, changing light, blocked tools, different door handles, interrupted workflows, and the small corrections people make without thinking.
Those imperfect details are exactly what robots must learn before entering factories, warehouses, care facilities, homes, and service environments. Ordinary people already handle those details every day.
CHEK EGO Miner turns scattered daily action experience into robot-training data that can be organized, governed, and reused.
When you participate in CHEK EGO Miner, you will see three names:
- CHEK EGO Miner: the EGO data collection project started by the CHEK Robot（擎控机器人） team. Its goal is to organize first-person human action into a data supply network that robots can learn from.
- CHEK: the mobile app. You use it to enter mobile capture flows, record first-person footage, and follow task requirements.
- Qingkong Miker: the desktop app. You use it for task coordination, device checks, capture management, upload, and review flow.
In short: CHEK EGO Miner is the project, CHEK is the mobile app, and Qingkong Miker is the desktop app.
The rest of the docs use those names consistently so you can tell which client to download and which step to operate.
- Public collectors who want to use a phone and a computer to participate in EGO data tasks, download CHEK and Qingkong Miker, and complete their first capture, upload, and review flow.
- Community readers and media explainers who want to explain why EGO data is being described as a new kind of data mining, and why it is not just ordinary video recording.
- Device, scene, and task partners who need to understand hardware needs, suitable scenarios, privacy, consent, and delivery boundaries.
| Goal | Link |
|---|---|
| Download CHEK and Qingkong Miker | Download guide |
| Start your first phone-and-computer collection | Quick start |
| Prepare phone, computer, mount, camera, or IMU hardware | Hardware guide |
| Record a real EGO data session | Capture guide |
| Understand what a session will save | Delivery contract |
| Fix download, preview, storage, or device-recognition issues | Troubleshooting |
| Understand consent and public-screenshot boundaries | Privacy |
| Read common questions | FAQ |
| Product | Platform | Official entry | Notes |
|---|---|---|---|
| Qingkong Miker desktop app | macOS / Windows / Linux |
smart-download | Open on a desktop browser to reach the desktop-client branch. |
| CHEK mobile app | iOS |
App Store / TestFlight | Use the App Store for the public iOS build. TestFlight remains available for testing builds when requested. |
| CHEK mobile app | Android |
smart-download | Routes through app-market flows first and falls back to APK download. |
CHEK EGO Miner will keep making the data task flow easier to understand and complete. The product path is simple:
- Record the task: use CHEK on the phone and Qingkong Miker on the desktop to capture real first-person work.
- Check the recording: review whether the media, timing, devices, and task requirements look usable before upload.
- Review for robot learning: for tasks that need deeper review, add a workspace where suggested labels, human approval, pose and trajectory evidence, QA freeze, and LeRobot output can be handled step by step.
- Upload the result: upload only the task-approved result, then track upload, reward review, settlement, and training-data approval as separate statuses.
LeRobot support belongs to the robot-learning review step. It does not replace recording or uploading, and a LeRobot export does not automatically mean that a task has passed settlement or training-data approval.
Public explainers sometimes describe this as data mining because real human actions and experience can become robot-training material. Actual rewards, review rules, and settlement terms depend on the specific task and platform policy. This repository does not promise a fixed hourly income.
This is the public information entry for CHEK EGO Miner. It helps new users start with download, setup, capture, privacy, and troubleshooting.
It mainly provides:
- download and install instructions;
- CHEK mobile app and Qingkong Miker desktop app guidance;
- phone, computer, mount, camera, and IMU guidance;
- EGO capture workflow;
- privacy, consent, safety, and public issue boundaries;
- troubleshooting for ordinary users.
If you want to improve public docs, add hardware feedback, or fix download guidance, start with the contribution guide. App source code, internal runtime code, private deployment notes, and non-public logs are not published here.
Documentation content is open for use. App binaries, services, hardware protocols, trademarks, and official release materials stay outside that open boundary.