ACT-2 Achieves 99.1% in Home Laundry Folding
lukas_m_ziegler · x · 2026-07-17
Sunday Robotics released an evaluation standard called Solve and used it to showcase the laundry folding capabilities of their robotic foundation model, ACT-2.
Evaluation Highlights
- Clearly defining task scope: clothing types, lighting, surfaces, initial placement, etc.
- Explaining adaptation costs during evaluation: No need to fine-tune for every individual home
- The same set of model weights can run directly across different homes
Results and Conclusions
- ACT-2 achieved a 99.1% success rate in folding clothes in unseen homes
- Scenarios covered various houses, lighting conditions, bed types, and clothing sizes
- The team believes that pre-training on large-scale, high-quality human data will significantly enhance the model's generalization capabilities
- The original post also noted: the model can even learn new folding techniques from a single example and transfer them to new environments
- Memo plans to enter households later this year
Related event: Sunday Robotics Unveils ACT-2 Preview for Home Robots(17 posts)→
More from Embodied
- Full teardown: inside Zmotor's 32mm linear actuator for humanoids — MarwaEldiwiny · 2026-10-27
- Full Teardown of a Zmotor Linear Actuator Used in China's Humanoid Robots — MarwaEldiwiny · 2026-10-27
- Video AI agent runs a physics engine, then paints your real room over the blocks — jonstephens85 · 2026-10-27
- World Model Assigns Weight and Friction to Objects, Then Simulates with a Physics Engine — jonstephens85 · 2026-10-27
- First unboxing shots of Apple's foldable "iPhone Duo" surface — chrisfirst · 2026-10-27
- UniAction: unified action conditioning lets one world model handle mixed action types — ZeYanjie · 2026-10-27