UW and Ai2 unveil Flex-π, a 6B-parameter robot policy that can repair its own gripper
rbhar90 · x · 2026-09-08
The University of Washington and Allen Institute for AI introduced Flex-π, a 6-billion-parameter robot manipulation policy applying JEPA principles to action prediction—demonstrated by a robot repairing its own gripper.
Key details:
- Unlike world-action models predicting only future RGB latents, Flex-π jointly denoises three streams in a shared latent space: RGB appearance, 3D geometry (pointmaps), and object-centric DINO semantics.
- Architecture: a single 5B trunk plus a 1B action expert on frozen encoders (Wan-2.2 video VAE for RGB and pointmaps, frozen DINOv3).
- Per-stream dropout with cross-modality forcing lets one checkpoint run 56 input/output combinations, from an action-only fast path onward.
More from Embodied
- Musk touts CyberCab ride experience, hopes service comes to Europe soon — elonmusk · 2026-09-08
- Opendoor founder Eric Wu's NavigateAI builds hands-free AI copilots for construction workers — saranormous · 2026-09-08
- GPT-6 'Astra' learns to control a real robot arm and paint the Golden Gate Bridge — alexeyguzey · 2026-09-08
- US police fear Meta smart glasses could be used to secretly record them — bookofjoe · 2026-09-08
- GPT-6 Astra vs MediaPipe for hand pose: SOTA reasoning, 3 min per frame — chris_j_paxton · 2026-09-08
- After 10+ Kindles, this reader says the Xteink x3 is the most portable yet — 6 books in 23 days — Technovangelist · 2026-09-08