Skip to content

Commit

Permalink
Update paper pdf
Browse files Browse the repository at this point in the history
  • Loading branch information
Murtaza Dalal committed Sep 8, 2024
1 parent 9a72ec9 commit c4cfbab
Show file tree
Hide file tree
Showing 7 changed files with 77 additions and 8 deletions.
77 changes: 69 additions & 8 deletions index.html
View file
Original file line number Diff line number Diff line change
Expand Up @@ -74,19 +74,19 @@

<section class="hero">
<div class="hero-body">
<div class="container is-max-desktop">
<div class="container">
<div class="columns is-centered">
<div class="column has-text-centered">
<h1 class="title is-1 publication-title">Neural MP: A Policy Learning Framework for Generalist Neural Motion Planners</h1>
<h1 class="title is-1 publication-title">Neural MP: A Generalist Neural Motion Planner</h1>
<div class="is-size-5 publication-authors">
<a target="_blank" href="https://mihdalal.github.io/">Murtaza Dalal*</a>&nbsp;&nbsp;&nbsp;
<a target="_blank" href="https://www.linkedin.com/in/jiahui-jim-yang-0666542a0/">Jiahui Yang*</a>&nbsp;&nbsp;&nbsp;
<a target="_blank" href="https://jim-young6709.github.io/">Jiahui Yang*</a>&nbsp;&nbsp;&nbsp;
<a target="_blank" href="https://russellmendonca.github.io/">Russell Mendonca</a>&nbsp;&nbsp;&nbsp;
<a target="_blank" href="https://www.linkedin.com/in/youssefkhaky/">Youssef Khaky</a>&nbsp;&nbsp;&nbsp;
<a target="_blank" href="https://www.cs.cmu.edu/~rsalakhu/">Ruslan Salakhutdinov</a>&nbsp;&nbsp;&nbsp;
<a target="_blank" href="https://www.cs.cmu.edu/~dpathak/">Deepak Pathak</a>&nbsp;&nbsp;&nbsp;
<br><br>
<img src="./resources/cmu-wordmark-horizontal-r.png" width="40%"></img>
<img src="./resources/cmu-wordmark-horizontal-r.png" width="30%"></img>
</span>
</div>

Expand All @@ -107,8 +107,9 @@ <h1 class="title is-1 publication-title">Neural MP: A Policy Learning Framework
<span class="link-block">
<a target="_blank" href=""
class="external-link button is-normal is-rounded is-dark">

<span class="icon">
<i class="fas fa-file"></i>
<i class="ai ai-arxiv"></i>
</span>
<span>ArXiv</span>
</a>
Expand All @@ -124,6 +125,17 @@ <h1 class="title is-1 publication-title">Neural MP: A Policy Learning Framework
<span>Code</span>
</a>
</span>

<!-- Model Link. -->
<span class="link-block">
<a href="https://huggingface.co/mihdalal/NeuralMP"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<img src="static/images/hf_icon.svg" />
</span>
<span>Models</span>
</a>
</span>

<!-- twitter Link. -->
<span class="link-block">
Expand All @@ -136,6 +148,17 @@ <h1 class="title is-1 publication-title">Neural MP: A Policy Learning Framework
</a>
</span>

<!-- twitter Link. -->
<span class="link-block">
<a target="_blank" href="#BibTeX"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file"></i>
</span>
<span>BibTeX</span>
</a>
</span>

</div>
</div>
</div>
Expand All @@ -150,7 +173,7 @@ <h1 class="title is-1 publication-title">Neural MP: A Policy Learning Framework
<div class="container">
<div class="columns is-vcentered is-centered">
<div class="item item-steve"></div>
<iframe width="1920" height="740" src="https://www.youtube.com/embed/HoVoNd0a4G4?autoplay=1&mute=1&loop controls=1&vq=hd1080" frameborder="0" allowfullscreen="allowfullscreen"></iframe>
<iframe width="1920" height="740" src="https://www.youtube.com/embed/0SHJbNTg3yU?autoplay=1&mute=1&loop controls=1&vq=hd1080" frameborder="0" allowfullscreen="allowfullscreen"></iframe>
</div>
</br>
</div>
Expand Down Expand Up @@ -227,7 +250,7 @@ <h2 class="subtitle has-text-centered">
</section>

<section class="section">
<div class="container is-max-desktop">
<div class="container">
<!-- Abstract. -->
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
Expand All @@ -239,7 +262,6 @@ <h2 class="title is-3">Abstract</h2>
Our approach builds a large number of complex scenes in simulation, collects expert data from a motion planner, then <i>distills</i> it into a reactive generalist policy.
We then combine this with simple forward samplers and lightweight optimization to obtain a safe path for real world deployment.
We perform a thorough evaluation of our method on <b>64</b> motion planning tasks across four diverse environments with randomized poses, scenes and obstacles, in the real world, demonstrating an improvement of <b>23%</b>, <b>17%</b> and <b>79%</b> motion planning success rate over state of the art sampling, optimization and learning based planning methods.
<!-- The current paradigm for motion planning generates solutions from scratch for every new problem, which consumes time and compute resources. Inspired by the ability of humans to practice from diverse experiences and gradually increase proficiency from slow, unsure behavior to fast, dynamic motions while acquiring new skills, we seek to use data-driven learning. Collecting a diverse set of real world scenes with various objects, obstacles, and scene configurations to enable generalization is prohibitively expensive, especially for motion planning problems. Our approach builds a large number of complex scenes in simulation, where expert data is collected and then <em>distilled</em> into a reactive generalist policy. We build complex scenes by combining procedural, programmatic generation with models of everyday objects sampled from large 3D datasets. We then combine this with simple forward samplers and lightweight optimization to obtain a safe path for real world deployment. We perform a thorough evaluation of our method on <b>64</b> motion planning tasks across <b>four</b> diverse environments with randomized poses, scenes and obstacles, in the real world, demonstrating an improvement of over <b>20%</b> over sampling-based and <b>80%</b> over state-of-the-art neural motion planning methods. All code, models and datasets will be released on acceptance. -->
</p>
</div>
</div>
Expand Down Expand Up @@ -277,6 +299,16 @@ <h1 class="title is-2" style="text-align: center; padding-bottom: 10px;">Neural

</section>

<section class="section">
<div class="container">
<div class="columns is-centered">
<div class="column is-full-width">
<h2 class="title is-4" style="text-align: center;">All videos play at 2x speed</h2>
</div>
</div>
</div>
</section>

<section class="section">
<div class="container">
<div class="columns is-centered">
Expand Down Expand Up @@ -680,6 +712,35 @@ <h3 class="title is-3" style="text-align: center; padding-bottom: 10px;">
</div>
</section>

<a id="BibTeX"></a>
<section class="section" id="BibTeX">
<div class="container is-max-desktop content">
<h2 class="title">BibTeX</h2>
<pre><code>@article{dalal2024neuralmp,
title={OpenVLA: An Open-Source Vision-Language-Action Model},
author={Murtaza Dalal and Jiahui Yang and Russell Mendonca and Youssef Khaky and Ruslan Salakhutdinov and Deepak Pathak},
journal = {},
year={2024},
} </code></pre>
</div>
</section>

<footer class="footer">
<div class="container">
<div class="columns is-centered">
<div class="column is-8">
<div class="content">
<p> Website borrowed from <a href="https://github.com/nerfies/nerfies.github.io">NeRFies</a> under a <a
href="https://creativecommons.org/licenses/by-sa/4.0/">Creative Commons Attribution-ShareAlike 4.0
International</a>
</p>

</div>
</div>
</div>
</div>
</footer>



</body>
Expand Down
Binary file modified resources/gpu compressed.mp4
View file
Binary file not shown.
Binary file modified resources/gpu.mp4
View file
Binary file not shown.
Binary file modified resources/paper.pdf
View file
Binary file not shown.
Binary file modified resources/sword compressed.mp4
View file
Binary file not shown.
Binary file modified resources/sword.mp4
View file
Binary file not shown.
8 changes: 8 additions & 0 deletions static/images/hf_icon.svg
View file
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.

0 comments on commit c4cfbab

Please sign in to comment.