ZEST: Zero-shot embodied skill transfer for athletic robot control
Science Robotics · 12.08.2026 · 11 (117)
DOI: 10.1126/scirobotics.aec7695 ↗Kurz gefasst
Fachartikel in der Zeitschrift Science Robotics, erschienen am 12.08.2026. Verfasst von Jean Pierre Sleiman und 27 weiteren Autor:innen. Beteiligte Einrichtung: RAI Institute, Cambridge, MA, USA. Hauptthemen laut MEDLINE: Robotics, Sports. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 68 Literaturangaben.
Abstract (Originalsprache)
Achieving robust, humanlike whole-body control on humanoid robots for agile, contact-rich behaviors remains a central challenge, demanding heavy per-skill engineering and a brittle process of tuning controllers. We introduce ZEST (zero-shot embodied skill transfer), a streamlined motion-imitation framework that trains policies via reinforcement learning from diverse sources—high-fidelity motion capture, noisy monocular video, and non–physics-constrained animation—and deploys them to hardware zero-shot. ZEST generalizes across behaviors and platforms without relying on contact labels, reference or observation windows, state estimators, or extensive reward shaping. Its training pipeline combines adaptive sampling, which focuses training on difficult motion segments, and an automatic curriculum using a model-based assistive wrench, together enabling dynamic, long-horizon maneuvers. We further provide a procedure for selecting joint-level gains from approximate analytical armature values for closed-chain actuators, along with a refined model of actuators. Trained entirely in simulation with moderate domain randomization, ZEST demonstrated broad generality. On Boston Dynamics’ Atlas humanoid, ZEST learned dynamic, multicontact skills (army crawl and breakdancing) from motion capture. It transferred expressive dance and scene-interaction skills, such as box climbing, directly from videos to Atlas and the Unitree G1. Furthermore, it extended across morphologies to the Spot quadruped, enabling acrobatics, such as a continuous backflip, through animation. Together, these results demonstrate robust zero-shot deployment across heterogeneous data sources and embodiments, establishing ZEST as a scalable interface between biological movements and their robotic counterparts.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- RAI Institute, Cambridge, MA, USA.
- Boston Dynamics, Waltham, MA, USA.
ORCID-Profile: Jean Pierre Sleiman, He Li, Alphonsus Adu-Bredu, Robin Deits, Arun Kumar, Matthew A Estrada, Alexander Lambert, Erica Lin, M Eva Mungai, Shane Rozen-Levy, Yuyao Shi, Jiashun Wang, Jakob Welner, Mike Zhang, Jessica Hodgins, Sylvain Bertrand, Yeuhi Abe, Scott Kuindersma, Farbod Farshidian
Zitieren
APA
Sleiman, J. P., Li, H., Adu-Bredu, A., Deits, R., Kumar, A., Bergamin, K., Bhardwaj, M., Biddlestone, S., Burger, N., Estrada, M. A., Iacobelli, F., Koolen, T., Lambert, A., Lin, E., Mungai, M. E., Nobles, Z., Rozen-Levy, S., Shi, Y., Wang, J., …, & Farshidian, F. (2026). ZEST: Zero-shot embodied skill transfer for athletic robot control. Science Robotics, 11(117). https://doi.org/10.1126/scirobotics.aec7695
Vancouver
Sleiman JP, Li H, Adu-Bredu A, Deits R, Kumar A, Bergamin K, et al. ZEST: Zero-shot embodied skill transfer for athletic robot control. Science Robotics. 2026;11(117). doi:10.1126/scirobotics.aec7695
BibTeX
@article{Sleiman2026,
author = {Sleiman, Jean Pierre and Li, He and Adu-Bredu, Alphonsus and Deits, Robin and Kumar, Arun and Bergamin, Kevin and Bhardwaj, Mohak and Biddlestone, Scott and Burger, Nicola and Estrada, Matthew A and Iacobelli, Francesco and Koolen, Twan and Lambert, Alexander and Lin, Erica and Mungai, M Eva and Nobles, Zach and Rozen-Levy, Shane and Shi, Yuyao and Wang, Jiashun and Welner, Jakob and Yu, Fangzhou and Zhang, Mike and Rizzi, Alfred and Hodgins, Jessica and Bertrand, Sylvain and Abe, Yeuhi and Kuindersma, Scott and Farshidian, Farbod},
title = {ZEST: Zero-shot embodied skill transfer for athletic robot control},
journal = {Science Robotics},
year = {2026},
volume = {11},
number = {117},
doi = {10.1126/scirobotics.aec7695}
}
EUVIMED