Guardado en:
Detalles Bibliográficos
Autores principales: Kowalski, Victor, Li, Chengxi, Lee, Dongheui
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:https://arxiv.org/abs/2605.29564
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866916059050672128
author Kowalski, Victor
Li, Chengxi
Lee, Dongheui
author_facet Kowalski, Victor
Li, Chengxi
Lee, Dongheui
contents When using reinforcement learning (RL) for contact-rich robotic manipulation, vision can provide task-relevant information that accelerates learning beyond what proprioception alone can achieve. However, vision-enabled policies tend to overfit to the visual conditions seen during training, limiting their robustness and transferability. We present a human-in-the-loop RL framework that employs teacher-student distillation to achieve robust performance across multiple task variants, trained entirely in the real world without requiring domain randomization or data augmentation. A vision-enabled teacher distills its knowledge into a vision-free student that relies solely on pose, twist, and wrench sensing, combining fast training with strong task generalization. On the real-world NIST assembly benchmark board, our approach achieves 95\% overall success after approximately 50 minutes of training on 3 representative tasks, including robust generalization to 8 unseen task variants. Fine-tuning with distillation achieves full success on the most challenging task. We demonstrate that the resulting policies outperform baselines in both robustness and adaptability.
format Preprint
id arxiv_https___arxiv_org_abs_2605_29564
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle VE2VF: Vision-Enabled to Vision-Free Distillation via Real-world Reinforcement Learning for Robust Contact-Rich Manipulation
Kowalski, Victor
Li, Chengxi
Lee, Dongheui
Robotics
When using reinforcement learning (RL) for contact-rich robotic manipulation, vision can provide task-relevant information that accelerates learning beyond what proprioception alone can achieve. However, vision-enabled policies tend to overfit to the visual conditions seen during training, limiting their robustness and transferability. We present a human-in-the-loop RL framework that employs teacher-student distillation to achieve robust performance across multiple task variants, trained entirely in the real world without requiring domain randomization or data augmentation. A vision-enabled teacher distills its knowledge into a vision-free student that relies solely on pose, twist, and wrench sensing, combining fast training with strong task generalization. On the real-world NIST assembly benchmark board, our approach achieves 95\% overall success after approximately 50 minutes of training on 3 representative tasks, including robust generalization to 8 unseen task variants. Fine-tuning with distillation achieves full success on the most challenging task. We demonstrate that the resulting policies outperform baselines in both robustness and adaptability.
title VE2VF: Vision-Enabled to Vision-Free Distillation via Real-world Reinforcement Learning for Robust Contact-Rich Manipulation
topic Robotics
url https://arxiv.org/abs/2605.29564