跳到正文
HuggingFace Daily Papers(社區熱門論文)·· 2026-10-23AI 評分46

DiscoDemo:為模仿學習探索高效且多樣化的機械人示範

DiscoDemo: Discovering Efficient and Diverse Robot Demonstrations for Imitation Learning

AI 導讀

DiscoDemo以強化學習生成多樣而成功的機械人操作軌跡,並將其整理成靜態數據集,供下游視覺運動模仿學習使用。在四項操作任務中,模擬環境的下游成功率由最強基線的40%升至62%,真實機械人上的成功率則由39%升至78%。研究顯示,成功解法的多樣性有助改善預訓練 flow-matching VLA 的模仿學習表現。

正文

Published on Oct 23

Authors:

,

,

,

,

,

,

,

Abstract

DiscoDemo uses reinforcement learning to generate diverse successful robot manipulation trajectories for training more robust visuomotor imitation policies.

Behavior cloning for robotic manipulation is highly sensitive to the coverage of its training data, yet collecting diverse successful demonstrations remains costly. We study reinforcement learning (RL) not as a deployable controller, but as a data generator for downstream visuomotor imitation learning. Our key premise is that multiple successful solution modes cover a broader range of successful trajectories than repeatedly collecting a single narrow solution. We introduce DiscoDemo, which trains a privileged latent-conditioned RL generator with a parallel reverse curriculum and a success-gated diversity objective to discover diverse solutions while preserving task success. The trained generator produces multiple successful closed-loop behaviors, which are safety-filtered and exported as a static dataset for downstream imitation learning. Across four manipulation tasks, DiscoDemo produces broader successful-state coverage and improves downstream visuomotor policies in both simulation and the real world. Averaged across tasks, downstream success increases to 62% from 40% for the strongest baseline in simulation, and to 78% from 39% on the real robot, respectively. These results suggest that successful solution diversity improves imitation learning for a pretrained flow-matching VLA and remains complementary to corrective data collection. Project page: https://davian-robotics.github.io/DiscoDemo

View arXiv page View PDF Project page GitHub 7 Add to collection

Get this paper in your agent:

hf papers read 2610.32093

Don't have the latest CLI?

curl -LsSf https://hf.co/cli/install.sh | bash

Models citing this paper 16

DAVIAN-Robotics/DiscoDemo-Stage3_SFT-PnP_Banana

Robotics •

4B •

Updated about 19 hours ago

•

60

•

1

DAVIAN-Robotics/DiscoDemo-Stage1_RL-PnP_Banana

Reinforcement Learning •

Updated about 19 hours ago

•

48

DAVIAN-Robotics/DiscoDemo-Stage1_RL-PnP_Banana-alpha0

Reinforcement Learning •

Updated about 19 hours ago

•

32

DAVIAN-Robotics/DiscoDemo-Stage1_RL-StackCube

Reinforcement Learning •

Updated about 19 hours ago

•

29

Browse 16 models citing this paper

Datasets citing this paper 12

DAVIAN-Robotics/DiscoDemo-Stage2_GenData-StackCube-alpha0

Franka Panda •

Updated about 19 hours ago

•

3k episodes

•

306

DAVIAN-Robotics/DiscoDemo-Stage2_GenData-FMB_SqCircle

Franka Panda •

Updated about 19 hours ago

•

3k episodes

•

296

DAVIAN-Robotics/DiscoDemo-Stage2_GenData-PnP_Banana-alpha0

Franka Panda •

Updated about 19 hours ago

•

3k episodes

•

292

DAVIAN-Robotics/DiscoDemo-Stage2_GenData-FMB_Round-alpha0

Franka Panda •

Updated about 19 hours ago

•

3k episodes

•

283

Browse 12 datasets citing this paper

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2610.32093 in a Space README.md to link it from this page.

Collections including this paper 0

No Collection including this paper

Add this paper to a collection to link it from this page.

來源:HuggingFace Daily Papers(社區熱門論文) · huggingface.co