Robot learns new task from a single demo: Machenta unveils Vision-Demo-Action foundation model
Ed__Johns · x · 2026-10-28
Machenta, led by Imperial College professor Ed Johns, unveiled Instant Robot Learning: a Vision-Demo-Action foundation model trained purely in simulation that lets robots master manipulation tasks instantly from one demonstration, with no further training.
Demos show one-person automation of assembly lines in minutes, compositional generalization to novel objects, and rapid iteration via physical prompts. The system supports reactivity, recovery, steerability, and cross-embodiment across varied robots including humanoids.
More from Embodied
- NVIDIA open-sources Flora, a model suite that auto-labels robot videos for training GR00T — qinzytech · 2026-10-28
- NVIDIA Launches SimReady Central: 3,500+ Free OpenUSD Simulation Assets on Hugging Face — mishig25 · 2026-10-28
- Formic CEO Co-Writes Essay on the Realistic Path to General-Purpose Robots That Don't Suck — AnneliesGamble · 2026-10-28
- YC: As robot skills transfer across machines, the bottleneck shifts to hardware — ycombinator · 2026-10-28
- Xplore launches RWE-Bench: physics-verified RL training tasks for engineering — Jsevillamol · 2026-10-28
- Gemini watches Genie 4 render frame-by-frame to simulate Apollo humanoid playing checkers — chrisfirst · 2026-10-28