# Real2Sem2Real Turns One Human Video Into Thousands of Robot Demonstrations That Match Teleoperation Data

The pipeline generates more than 2,000 training episodes in about two hours from a single monocular RGB recording, and policies trained on that data matched or beat teleoperation baselines on two real robots and four tasks.

- Published: 2026-10-23T06:19:29.884Z
- Canonical: https://polylog.news/ai/2026-10-23/real2sem2real-turns-one-human-video-into-thousands-of-robot
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [Polylog editors](https://polylog.news), [arXiv cs.RO](https://arxiv.org/abs/2610.28880), [Real2Sem2Real project page](https://real2sem2real.github.io/)

A paper posted to arXiv on October 20, Real2Sem2Real, addresses the most expensive input in robot learning: demonstration data. The pipeline starts from one monocular RGB video of a person doing a task. It produces thousands of robot demons…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-10-23/real2sem2real-turns-one-human-video-into-thousands-of-robot (subscription information: https://polylog.news/pricing).