VR-DAgger: Immersive VR for Dexterous Data Collection and Uncertainty-Guided On-Policy Correction

René Zurbrügg; Tifanny Portela; Arjun Bhardwaj; Aravind Elanjimattathil Vijayan; Maximum Wilder-Smith; Marco Hutter

VR-DAgger: Immersive VR for Dexterous Data Collection and Uncertainty-Guided On-Policy Correction

Robotics 2026-05-27 v1

Authors: René Zurbrügg , Tifanny Portela , Arjun Bhardwaj , Aravind Elanjimattathil Vijayan , Maximum Wilder-Smith , Marco Hutter

View on arXiv ↗ PDF ↗

Abstract

Learning from demonstrations is effective for robotic manipulation, but collecting sufficient task-specific data remains a major bottleneck. Under distribution shift, small errors compound, performance degrades, and expert time is often spent on redundant, low-value corrections instead of the few critical failure cases.

Keywords

imitation learning sim-to-real transfer robot control

Cite

@article{arxiv.2605.27114,
  title  = {VR-DAgger: Immersive VR for Dexterous Data Collection and Uncertainty-Guided On-Policy Correction},
  author = {René Zurbrügg and Tifanny Portela and Arjun Bhardwaj and Aravind Elanjimattathil Vijayan and Maximum Wilder-Smith and Marco Hutter},
  journal= {arXiv preprint arXiv:2605.27114},
  year   = {2026}
}

Related papers

View all related →

Machine Learning · Computer Science

Data-efficient Deep Reinforcement Learning for Dexterous Manipulation

Ivaylo Popov, Nicolas Heess, Timothy Lillicrap, Roland Hafner +6