Sunday Robotics ACT-2 achieves zero-shot generalization breakthrough
tonyzzhao · x · 2026-08-22
Sunday Robotics' ACT-2 model demonstrates that reliability gains from in-house post-training can transfer zero-shot to new environments. While robotic models often fail when conditions change, ACT-2 closes this generalization gap by scaling high-quality pre-training with diverse human demonstration data. It achieved over 99.1% zero-shot success on laundry folding across unseen homes and garments.
More from Embodied
- Tevel's robotic system harvests fruit around the clock using computer vision — LexiLove · 2026-10-16
- SHOES: 10 minutes of real-robot RL lets quadrupeds walk in shoes, CoRL 2026 paper — kscottz · 2026-10-16
- Tesla Cybercab robotaxis logged zero crashes in first two weeks of public rides, NHTSA data shows — elonmusk · 2026-10-16
- Robotics LLMs tested on 16 tasks: GPT-6 Astra hits 17.2% success vs Fable's 9.1% — pabbeel · 2026-10-16
- WEDGE paper generates training environments conditioned on navigation agent failures — _akhaliq · 2026-10-16
- Viral meme nails AI's irony: it can write symphonies but can't do the dishes — briannekimmel · 2026-10-16