Abstract
Human hand-object demonstrations provide a scalable source of data for dexterous robot learning, but transferring them across embodiments requires physically feasible retargeting. Existing physics-based methods typically optimize each demonstration independently, leading to either limited success under finite simulation budgets or training costs that grow with dataset size. We introduce FlashDexRetarget, an RL framework for multi-reference dexterous retargeting. We formulate retargeting as multi-reference tracking, jointly learning a single policy across many demonstrations with off-policy RL and geometric supervision of the demonstrated interactions. This shared training formulation amortizes optimization across references while enabling the policy to track diverse hand-object interactions. On a 50-motion benchmark from TACO, OakInk2, and HOT3D using XHand and Sharpa Wave Hand as target embodiments, FlashDexRetarget retargets 90% of demonstrations using about 30 GPU-hours, compared with about 46% at about 3,000 GPU-hours for CHORD. This corresponds to about 100 times lower training compute and a 44-percentage-point improvement in retargeting success. Ablations examine the key design choices, while experiments with up to 1,000 motions and real-world replay further demonstrate the scalability and practical applicability of our method.
Keywords
Subject
Publication details
- Journal
- Not available
- Open access
- Green open access
Cite this article
APA 7
Lee, K., Kim, S., Hwang, D., Oh, Y., Kim, D., Lee, Y., Nahrendra, I. M. A., Choo, J., & Lee, H. (2026). FlashDexRetarget: Accelerating Dexterous Manipulation Data Generation through Multi-Motion Retargeting. https://omanscience.com/en/articles/flashdexretarget-accelerating-dexterous-manipulation-data-generation-through-multi-motion-retargeting
MLA 9
Lee, Kyungmin, et al. "FlashDexRetarget: Accelerating Dexterous Manipulation Data Generation through Multi-Motion Retargeting." https://omanscience.com/en/articles/flashdexretarget-accelerating-dexterous-manipulation-data-generation-through-multi-motion-retargeting.
Chicago (author–date)
Lee, Kyungmin, Sibeen Kim, Dongyoon Hwang, Yoonsang Oh, Donghu Kim, Youngdo Lee, I Made Aswin Nahrendra, Jaegul Choo, and Hojoon Lee. 2026. "FlashDexRetarget: Accelerating Dexterous Manipulation Data Generation through Multi-Motion Retargeting." https://omanscience.com/en/articles/flashdexretarget-accelerating-dexterous-manipulation-data-generation-through-multi-motion-retargeting.
Harvard
Lee, K., Kim, S., Hwang, D., Oh, Y., Kim, D., Lee, Y., Nahrendra, I. M. A., Choo, J. and Lee, H. (2026) 'FlashDexRetarget: Accelerating Dexterous Manipulation Data Generation through Multi-Motion Retargeting', Available at: https://omanscience.com/en/articles/flashdexretarget-accelerating-dexterous-manipulation-data-generation-through-multi-motion-retargeting.
Vancouver
Lee K, Kim S, Hwang D, Oh Y, Kim D, Lee Y, et al. FlashDexRetarget: Accelerating Dexterous Manipulation Data Generation through Multi-Motion Retargeting. https://omanscience.com/en/articles/flashdexretarget-accelerating-dexterous-manipulation-data-generation-through-multi-motion-retargeting
IEEE
K. Lee, S. Kim, D. Hwang, Y. Oh, D. Kim, Y. Lee, I. M. A. Nahrendra, J. Choo, and H. Lee, "FlashDexRetarget: Accelerating Dexterous Manipulation Data Generation through Multi-Motion Retargeting," https://omanscience.com/en/articles/flashdexretarget-accelerating-dexterous-manipulation-data-generation-through-multi-motion-retargeting.