Workers in India are earning approximately $2.40 per hour to record “egocentric” video footage—first-person clips of everyday tasks like slicing produce or tying shoes—to train humanoid robots for global tech firms. According to data from industry observers and reports on firms like Objectways, this labor-intensive process is becoming the backbone of a robotics market that Goldman Sachs projects could reach $38 billion by 2035.
How Egocentric Video Powers Robot Learning
Robots learn by watching humans perform physical chores, a method known as imitation learning. By strapping cameras to their heads, workers at companies like Objectways create data streams that capture depth, hand motion, and environmental intent. This “egocentric” video is more valuable than standard third-person footage because it maps exactly how a human moves through a kitchen or factory floor. According to industry reports, this data allows developers to bypass manual, step-by-step coding, teaching machines to fold laundry or prep food through visual observation.
The term “egocentric video” refers to footage captured from the wearer’s point of view. In AI training, this is considered the “gold standard” for teaching robots spatial awareness and fine motor skills.
What Is the Economic Outlook for Humanoid Robotics?
Wall Street views the scaling of humanoid robots as a major growth engine for AI hardware. Goldman Sachs analysts estimate the market for these machines could hit $38 billion within the next decade, provided that hardware manufacturing costs decrease and general-purpose AI models become more reliable. The current business model relies on a global supply chain where US-based tech giants outsource the “heavy lifting” of data annotation to firms in hubs like Tamil Nadu, India. While the cost of labor remains low, the potential valuation of the finished robotics products suggests a significant gap between the wages of the data annotators and the market value of the AI they help build.
Why Does Worker Privacy Remain a Point of Contention?
The constant recording of human behavior in staged home and factory environments has triggered concerns regarding data privacy and consent. According to reports from the field, some workers have begun to set boundaries, such as refusing to film in private spaces like bedrooms. The central question for labor advocates is who owns the rights to these gestures once they are converted into commercial datasets. As AI models become more sophisticated, there is ongoing debate about whether workers should receive ongoing compensation or royalties for the specific physical movements that enable a robot to perform a task for a paying customer.
Comparison of AI Data Labor Models
| Feature | Traditional Coding | Egocentric AI Training |
|---|---|---|
| Method | Manual scripting | Imitation learning |
| Data Source | Engineers | Human annotators |
Frequently Asked Questions
Why are robots trained on human video instead of just code?
Coding every possible variable for a robot to move safely in a home is nearly impossible. By watching humans, robots learn intent and fluid motion, which is far more efficient than writing millions of lines of manual code.
Is the data from these workers sold to multiple companies?
Data-labeling firms like Objectways typically work under contract with specific clients, such as large tech platforms or robotics companies, to build proprietary datasets.
What happens if a worker wants to opt out of the data collection?
Current labor agreements in these regions are often project-based. Workers generally have the choice to stop participating, though the economic necessity of the role remains a primary driver for many in the region.
If you are tracking the AI industry, watch for emerging regulations in the EU and US regarding “data provenance.” These laws may soon require companies to disclose exactly how their training data was sourced and whether workers were fairly compensated.
What are your thoughts on the ethics of AI training labor? Join the conversation in the comments section below or subscribe to our weekly tech briefing for more updates on the robotics revolution.
