Robotics & Automation · Projected milestone
Robots learn a job in a day
60% by 2032 · including earlier years
The year is an editorial timing estimate. The percentage applies to occurrence by the stated deadline. These estimates are not calibrated odds.
A general-purpose robot is shown a new task a few times and does it reliably within a day, with no programming.
Explore this milestone on the interactive mapWhat the forecast means
Milestone stage: demonstration
partially-demonstrated
Task transfer to unseen homes is demonstrated for selected behaviors; broad one-day learning remains incomplete.
Completion criteria: Across a pre-registered set of twenty new task families and unseen environments, at least 80% reach 90% success after no more than a day and five demonstrations each. Report failures, developer interventions and excluded tasks; the event does not mean literally every physical occupation.
This page describes a forecast, rather than an achieved result. Read the methodology and assumptions.
Built on
Led to
No connections of this kind are recorded.
Sources & evidence
- Helix 2.5: Zero-shot generalization across 30 homes 2026-09-17
Figure reports three trained behaviors tested in thirty unseen homes without home-specific tuning, with aggregate whole-task success increasing from 9% to 56%. This is meaningful transfer and still falls short of dependable general household autonomy.
- Beyond VLAs: How world action models reshape robot manipulation 2026-08-04
NVIDIA describes video/world-model backbones and Cosmos 3 for robot action learning. This supports demonstrated world-model components, not a claim of universally accurate physics or reliable execution.
Record reviewed 2026-10-18