PhysCanvas world model predicts vision and touch, lifting sim-real correlation to 0.89
stepjamUK · x · 2026-10-21
PhysCanvas is a world model that jointly predicts what a robot sees and what its fingertips feel, letting you test robot policies without running them on hardware — covering tasks like racking test tubes, unscrewing bottle caps, bagging blocks and turning Rubik's Cube faces.
Key findings:
- Adding task-state supervision raises the correlation between world-model success and real-world success from 0.60 to 0.89
- Running RL inside the world model on flow matching and π0.5 policies improves real-world performance on average, though gains vary by task
- Code and model coming soon; paper and project page are live
More from Embodied
- The Verge editor relieved he skipped the $2,000 iPhone Duo after scathing review — tomwarren · 2026-10-21
- CoreWeave launches Physical AI Launchables: run robotics and simulation workloads with one command — ayushthakur0 · 2026-10-21
- David Patterson: The ChatGPT Moment for Robots Will Come From ChatGPT Itself — davidpattersonx · 2026-10-21
- SCAMPER: a virtual rat with realistic motion for RL-based neuroscience experiments — tyrell_turing · 2026-10-21
- VPM: teaching video world models to forecast 3D point trajectories improves videos and robots — sainingxie · 2026-10-21
- iPhone Duo week-long review: fully redesigned dual-screen apps make or break the experience — BenBajarin · 2026-10-21