Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

VE2VF: Vision-Enabled to Vision-Free Distillation via Real-world Reinforcement Learning for Robust Contact-Rich Manipulation

About

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.

Victor Kowalski, Chengxi Li, Dongheui Lee• 2026

Related benchmarks

TaskDatasetResultRank
Robotic InsertionNIST Assembly Board (disturbed)
Peg Success Rate (out of 10)100
5
Robotic InsertionNIST Assembly Board Out-of-distribution tasks (test)
M Peg Success Count10
5
Robotic InsertionNIST Assembly Board (train)
Peg Success Rate100
5
Showing 3 of 3 rows

Other info

Follow for update