Skip to content

Commit 23af11b

Browse files
Update index.html
1 parent ff6bae2 commit 23af11b

File tree

1 file changed

+12
-12
lines changed

1 file changed

+12
-12
lines changed

index.html

Lines changed: 12 additions & 12 deletions
Original file line numberDiff line numberDiff line change
@@ -3,8 +3,8 @@
33
<head>
44
<meta charset="UTF-8">
55
<meta name="viewport" content="width=device-width, initial-scale=1.0">
6-
<title>ROBOSE</title>
7-
<meta name="description" content="ROBOSE Project Page">
6+
<title>OpenHelix</title>
7+
<meta name="description" content="OpenHelix Project Page">
88
<!-- Bootstrap 3 -->
99
<link rel="stylesheet" href="https://stackpath.bootstrapcdn.com/bootstrap/3.4.1/css/bootstrap.min.css">
1010
<!-- FontAwesome for icons -->
@@ -38,7 +38,7 @@
3838
<body>
3939
<div class="container">
4040
<div class="row text-center">
41-
<h1><strong>ROBOSE</strong>: A Simple yet Effective Dual System for Robot Learning</h1>
41+
<h1><strong>OpenHelix</strong>: A Short Survey, Empirical Analysis, and Open-Source Dual-System VLA Model for Robotic Manipulation</h1>
4242
<p>Can Cui, Pengxiang Ding*, Wenxuan Song, Hangyu Liu, Yang Liu,<br>
4343
Bofang Jia, Han Zhao, Siteng Huang, Zhaoxin Fan, Donglin Wang†</p>
4444
<p>Westlake University, Zhejiang University, HKUST(GZ), Beijing Advanced Innovation Center for Future Blockchain and Privacy </p>
@@ -48,14 +48,14 @@ <h1><strong>ROBOSE</strong>: A Simple yet Effective Dual System for Robot Learni
4848
<div class="row text-center" style="margin-top: 20px;">
4949
<!-- PDF Link -->
5050
<span class="link-block">
51-
<a href="./static/paper/CARP.pdf" class="btn btn-custom" target="_blank">
51+
<a href="https://openhelix-robot.github.io/" class="btn btn-custom" target="_blank">
5252
<i class="fas fa-file-pdf"></i> Paper
5353
</a>
5454
</span>
5555

5656
<!-- arXiv Link -->
5757
<span class="link-block">
58-
<a href="https://arxiv.org/abs/2412.06782" class="btn btn-custom" target="_blank">
58+
<a href="https://openhelix-robot.github.io/" class="btn btn-custom" target="_blank">
5959
<i class="fas fa-book"></i> arXiv
6060
</a>
6161
</span>
@@ -69,8 +69,8 @@ <h1><strong>ROBOSE</strong>: A Simple yet Effective Dual System for Robot Learni
6969

7070
<!-- Huggingface Link -->
7171
<span class="link-block">
72-
<a href="https://huggingface.co/zhefeigong/carp" class="btn btn-custom" target="_blank">
73-
<i class="fas fa-smile-beam"></i> Huggingface
72+
<a href="https://openhelix-robot.github.io/" class="btn btn-custom" target="_blank">
73+
<i class="fas fa-smile-beam"></i> Huggingface(coming soon)
7474
</a>
7575
</span>
7676
</div>
@@ -86,7 +86,7 @@ <h1><strong>ROBOSE</strong>: A Simple yet Effective Dual System for Robot Learni
8686
<div class="col-md-8 col-md-offset-2">
8787
<h3>Abstract</h3>
8888
<p class="text-justify">
89-
ROBOSE proposes a simple yet effective dual system for robot learning by hierarchically integrating a high-level Multimodal Large Language Model (MLLM) with a low-level policy model. Through pre-alignment, prompt tuning, and multimodal reasoning learning, ROBOSE significantly enhances generalization, reduces training cost, and achieves state-of-the-art results on challenging robot manipulation benchmarks such as CALVIN and CALVIN-D.
89+
Dual-system VLA (Vision-Language-Action) architectures have become a hot topic in embodied intelligence research, but there is a lack of sufficient open-source work for further performance analysis and optimization. To address this problem, this paper will summarize and compare the structural designs of existing dual-system architectures, and conduct systematic empirical evaluations on the core design elements of existing dual-system architectures. Ultimately, it will provide a low-cost open-source model for further exploration. Of course, this project will continue to update with more experimental conclusions and open-source models with improved performance for everyone to choose from.
9090
</p>
9191
<p style="text-align: center;"><img src="./assets/robose_teaser.jpg" class="img-responsive"></p>
9292

@@ -98,7 +98,7 @@ <h3>Revisit</h3>
9898

9999
<h3>Approach</h3>
100100
<p class="text-justify">
101-
ROBOSE bridges the high-level MLLM and the low-level policy using a learned <ACT> token and linear projection. Prompt tuning allows efficient training without altering MLLM parameters. An auxiliary task ensures the MLLM performs multimodal reasoning by predicting actions directly from its latent embeddings. The policy model adopts a diffusion-based learning mechanism conditioned on visual, proprioceptive, and goal features.
101+
OpenHelix bridges the high-level MLLM and the low-level policy using a learned <ACT> token and linear projection. Prompt tuning allows efficient training without altering MLLM parameters. An auxiliary task ensures the MLLM performs multimodal reasoning by predicting actions directly from its latent embeddings. The policy model adopts a diffusion-based learning mechanism conditioned on visual, proprioceptive, and goal features.
102102
</p>
103103
<p style="text-align: center;"><img src="./assets/arch.jpg" class="img-responsive"></p>
104104

@@ -116,13 +116,13 @@ <h3>Dataset</h3>
116116

117117
<h3>Results</h3>
118118
<p class="text-justify">
119-
ROBOSE outperforms previous methods across all metrics in CALVIN, CALVIN-E (language generalization), and CALVIN-D (dynamic vision generalization). It achieves better success rates with fewer parameters and less training data, validating its data efficiency and robustness.
119+
OpenHelix outperforms previous methods across all metrics in CALVIN, CALVIN-E (language generalization), and CALVIN-D (dynamic vision generalization). It achieves better success rates with fewer parameters and less training data, validating its data efficiency and robustness.
120120
</p>
121121
<p style="text-align: center;"><img src="./assets/results.jpg" class="img-responsive"></p>
122122

123123
<h3>Citation</h3>
124-
<pre><code>@article{Cui2024ROBOSE,
125-
title={ROBOSE: A Simple yet Effective Dual System for Robot Learning},
124+
<pre><code>@article{Cui2024OPENHELIX,
125+
title={OpenHelix: A Simple yet Effective Dual System for Robot Learning},
126126
author={Can Cui and Pengxiang Ding and Wenxuan Song and Hangyu Liu and Yang Liu and Bofang Jia and Han Zhao and Siteng Huang and Zhaoxin Fan and Donglin Wang},
127127
journal={arXiv preprint arXiv:2403.13358},
128128
year={2024}

0 commit comments

Comments
 (0)