flowhoi

KO | EN

gitlite — search

#javascript #react #llm #go #golang #android #machine-learning #rust #deep-learning #linux #awesome #vue #awesome-list #openai

(★ 35)

FlowHOI: Flow-based Semantics-Grounded Generation of Hand-Object Interactions for Dexterous Robot Manipulation

flowhoi 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

<h1 align="center">FlowHOI: Flow-based Semantics-Grounded Generation of Hand-Object Interactions for Dexterous Robot Manipulation</h1>
 
 <a href="https://huajian-zeng.github.io/">Huajian Zeng</a>1, <a href="https://scholar.google.com/citations?user=vdo-1aYAAAAJ&hl">Lingyun Chen</a>2, <a href="https://scholar.google.com/citations?hl=zh-CN&user=f7ox7CIAAAAJ">Jiaqi Yang</a>1, <a href="https://scholar.google.com/citations?user=Lh2CthAAAAAJ&hl">Yuantai Zhang</a>1, <a href="https://fanshi14.github.io/me/">Fan Shi</a>3, <a href="https://ethliup.github.io/">Peidong Liu</a>4, <a href="https://xingxingzuo.github.io/">Xingxing Zuo</a>1,*
 
 1Mohamed Bin Zayed University of Artificial Intelligence (MBZUAI), 2Technical University of Munich (TUM), 3National University of Singapore (NUS), 4Westlake University
 
 *Corresponding author

In Submission

[![arXiv](https://img.shields.io/badge/arXiv-2602.13444-red)](https://arxiv.org/abs/2602.13444)
[![PDF](https://img.shields.io/badge/PDF-%F0%9F%93%84-green)](https://arxiv.org/pdf/2602.13444)
[![Homepage](https://img.shields.io/badge/Homepage-%F0%9F%8C%90-blue)](https://huajian-zeng.github.io/projects/flowhoi/)
[![Video](https://img.shields.io/badge/Video-%E2%96%B6-red?logo=youtube&logoColor=red)](https://youtu.be/c_ayPvccyi8)

</div>

## 🔥 Highlight <a name="highlight"></a>

**FlowHOI** is a two-stage flow-matching framework that generates semantically grounded, temporally coherent hand-object interaction (HOI) sequences for dexterous robot manipulation.

By decoupling geometry-centric grasping from semantics-centric manipulation and conditioning on 3D Gaussian splatting scene reconstruction, FlowHOI achieves **1.7× higher physics simulation success rate** and **40× inference speedup** compared to the strongest diffusion-based baseline.

These capabilities enable **real-robot execution on diverse dexterous manipulation tasks**, bridging the gap between HOI generation and practical robotic deployment.

## 📢 The code will be released after the paper is accepted. Stay tuned!

flowhoi

flowhoi 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

flowhoi

flowhoi 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README