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
- AI researchers guessed robots wouldn't outrun humans until 2034 — it already happened — ben_j_todd · 2026-10-26
- Shanghai AI Lab-led team releases 100-hour egocentric wearable dataset for humanoid training — ChuckDBrooks · 2026-10-26
- RealSense D585 Pro AI-Powered 3D Stereo Depth Camera Unboxed for Physical AI — chrismatthieu · 2026-10-26
- Physical world agent swarm demo shared on X — pduan · 2026-10-26
- Scoble shares his full list of AI wearable recording mics — best ones 'yet to come' — Scobleizer · 2026-10-26
- 'Lamp Swarm': a desk of lamp-shaped robots shows what physical AI work looks like — dee_hw · 2026-10-26