Video-pretrained model learns tactile prediction from little data, researcher calls it non-obvious
KyleMorgenstein · x · 2026-10-16
Kyle Morgenstein discusses a robotics result on X: a video-pretrained backbone (DVA) can generate tactile maps from only a fraction of tactile data. He argues this is non-obvious—tactile input is nominally out-of-distribution for a video-pretrained model, so a vision model is doing a task with no vision, and accurate tactile prediction with very little data is hard. He notes actions enter only via an IDM, and that DVA's deep alignment for video prediction likely matters; the same result on a small unaligned model would be less surprising.
More from Embodied
- Transforming quadruped robot runs on real hardware: per-pose RL policies in Isaac switched on morph — _Stocko_ · 2026-10-16
- Foldable iPhone Duo reportedly sells out launch-day inventory, some deliveries slip to 2027 — Polymarket · 2026-10-16
- KinoGen synthesizes robot loco-manipulation reference motions directly via kinematically grounded video generation — ChongZzZhang · 2026-10-16
- Science Robotics paper introduces F³, predictive event-based vision beyond optical flow — CSProfKGD · 2026-10-16
- iPhone Duo foldable appears on South African retailer iStore at R47,999 in apparent leak — shauntrennery · 2026-10-16
- New paper: game-theoretic motion planning shows it can be legal to break the rules — GioeleZardini · 2026-10-16