FlashDexRetarget: one RL policy retargets hand demos at 90% success, ~100x less compute
KyleMorgenstein · x · 2026-10-06
FlashDexRetarget uses a single RL policy to retarget diverse human hand-object demonstrations to dexterous robot hands. On a 50-motion benchmark (TACO, OakInk2, Hot3D) with XHand it achieves 90% success using 100x less compute than CHORD. Retweeters add that FlashSAC works remarkably well for dexterous manipulation. Project page and paper are public.
More from Embodied
- KAIST's RobotUse agent harness hits 45% task success on RoboLab, beating CaP-X — kaist-ai · 2026-10-06
- September robotics funding hits $4.86B across 167 rounds, led by Cornerstone's $700M — lukas_m_ziegler · 2026-10-06
- Prompt-Driven Autonomous Drone Flies Fully Unsupervised After Start — AryHHAry · 2026-10-06
- Survey: 65% of Japanese seniors prefer robot-assisted nursing homes, willing to pay 8% more — HealthcareLdr · 2026-10-06
- Tianjin University unveils 3-gram hair-concealable non-invasive brain-computer interface — Dr_Singularity · 2026-10-06
- PerturBot Breaks Shortcut Priors in Vision-Language-Action Models With Perturbative Training — Mingyu Liu · 2026-10-06