Skip to content

Commit 0a32585

Browse files
Guanya ShiGuanya Shi
Guanya Shi
authored and
Guanya Shi
committed
ICRA
1 parent 20a7300 commit 0a32585

3 files changed

Lines changed: 86 additions & 57 deletions

File tree

‎index.html‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -109,7 +109,7 @@
109109
<tr>
110110
<td width="100%" valign="middle">
111111
<ul>
112-
<li>2026-01-26: Four papers accepted to ICLR 2026 and two to L4DC 2026.</li>
112+
<li>2026-01-31: Four papers accepted to ICLR 2026, two to L4DC 2026, and six to ICRA 2026.</li>
113113
<li>2025-10-21: Congratulations to Zeji for receiving the <a href="https://www.cs.cmu.edu/news/2025/amazon-phd-fellows" target="_blank">Amazon AI Ph.D. Fellowship</a>, and Wenli for receiving the CMU RI Presidential Fellowship!</li>
114114
<li>2025-08-01: <a href="https://lecar-lab.github.io/spi-active_/" target="_blank">SPI-Active</a> (oral), <a href="https://lecar-lab.github.io/SoFTA/" target="_blank">HoldMyBeer</a>, and <a href="https://human-as-robot.github.io/" target="_blank">Humanoid~Human</a> accepted to CoRL 2025.</li>
115115
<li>2025-04-11: <a href="https://agile.human2humanoid.com/" target="_blank">ASAP</a> and <a href="https://xiaofeng-guo.github.io/flying_hand.io/" target="_blank">Flying Hand</a> accepted to RSS 2025.</li>

‎publications.html‎

Lines changed: 85 additions & 56 deletions
Original file line numberDiff line numberDiff line change
@@ -16,7 +16,7 @@
1616
<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/5.15.3/css/all.min.css">
1717
<style>
1818
:root {
19-
--highlight-color: rgba(255, 249, 196, 0.79)
19+
--highlight-color: rgba(255, 249, 196, 0.)
2020
}
2121
</style>
2222
</head>
@@ -57,8 +57,8 @@
5757
<tr>
5858
<td width="100%" valign="middle">
5959
<p>
60-
<!-- Check out the <a href="https://scholar.google.com/citations?user=joR1Z4UAAAAJ&hl=en&oi=ao" target="_blank">Google Scholar</a> page for a full and up-to-date publication list. -->
61-
<sup>*</sup> denotes equal contribution and <sup>&dagger;</sup> denotes equal advising. Representative papers are <span style="background-color: var(--highlight-color)">highlighted</span>.
60+
<sup>*</sup> denotes equal contribution and <sup>&dagger;</sup> denotes equal advising. Below are selected papers. The full publication list is <a href="https://scholar.google.com/citations?hl=en&user=joR1Z4UAAAAJ&view_op=list_works&sortby=pubdate" target="_blank">here</a>.
61+
<!-- Representative papers are <span style="background-color: var(--highlight-color)">highlighted</span>. -->
6262
</p>
6363
</td>
6464
</tr>
@@ -74,6 +74,27 @@
7474
</tr>
7575
</table>
7676

77+
<table width="880" border="0" align="center" cellspacing="0" cellpadding="0">
78+
<tr>
79+
<td style="width:35%; vertical-align:middle; padding-right: 20px;">
80+
<div class="image-container">
81+
<img src='publications/2026_RPL.gif' width="100%">
82+
</div>
83+
</td>
84+
<td style="width:65%; vertical-align:middle">
85+
<papertitle>RPL: Learning Robust Humanoid Perceptive Locomotion on Challenging Terrains</papertitle>
86+
<br>
87+
Yuanhang Zhang, Younggyo Seo, Juyue Chen, Yifu Yuan, Koushil Sreenath, Pieter Abbeel<sup>&dagger;</sup>, Carmelo Sferrazza<sup>&dagger;</sup>, Karen Liu<sup>&dagger;</sup>, Rocky Duan<sup>&dagger;</sup>, Guanya Shi<sup>&dagger;</sup>
88+
<br>
89+
<a href="https://arxiv.org/abs/2602.03002" target="_blank"><i class="far fa-file"></i> paper</a> &nbsp
90+
<a href="https://rpl-humanoid.github.io/" target="_blank"><i class="fas fa-globe"></i> website</a>
91+
<p style="margin-top: 5px"><i class="fas fa-comment-dots"></i> TL;DR: A single policy trained by RPL enables multi-directional robust humanoid locomotion over various challenging terrains.
92+
</td>
93+
</tr>
94+
</table>
95+
96+
<br>
97+
7798
<table width="880" border="0" align="center" cellspacing="0" cellpadding="0">
7899
<tr>
79100
<td style="width:35%; vertical-align:middle; padding-right: 20px;">
@@ -164,18 +185,17 @@
164185
<tr>
165186
<td style="width:35%; vertical-align:middle; padding-right: 20px;">
166187
<div class="image-container">
167-
<img src='publications/2025_TWIST2.gif' width="90%">
188+
<img src='publications/2025_ResMimic.gif' width="90%">
168189
</div>
169190
</td>
170191
<td style="width:65%; vertical-align:middle">
171-
<papertitle>TWIST2: Scalable, Portable, and Holistic Humanoid Data Collection System</papertitle>
192+
<papertitle>ResMimic: From General Motion Tracking to Humanoid Whole-body Loco-Manipulation via Residual Learning</papertitle>
172193
<br>
173-
Yanjie Ze, Siheng Zhao, Weizhuo Wang, Angjoo Kanazawa<sup>&dagger;</sup>, Rocky Duan<sup>&dagger;</sup>, Pieter Abbeel<sup>&dagger;</sup>, Guanya Shi<sup>&dagger;</sup>, Jiajun Wu<sup>&dagger;</sup>, C. Karen Liu<sup>&dagger;</sup>
194+
Siheng Zhao, Yanjie Ze, Yue Wang, C. Karen Liu<sup>&dagger;</sup>, Pieter Abbeel<sup>&dagger;</sup>, Guanya Shi<sup>&dagger;</sup>, Rocky Duan<sup>&dagger;</sup>
174195
<br>
175-
<a href="https://arxiv.org/abs/2511.02832" target="_blank"><i class="far fa-file"></i> paper</a> &nbsp
176-
<a href="https://yanjieze.com/TWIST2/" target="_blank"><i class="fas fa-globe"></i> website</a> &nbsp
177-
<a href="https://github.com/amazon-far/TWIST2" target="_blank"><i class="fas fa-code"></i> code</a>
178-
<p style="margin-top: 5px"><i class="fas fa-comment-dots"></i> TL;DR: TWIST2 is a portable whole-body humanoid teleoperation system that enables scalable data collection.
196+
<a href="https://www.arxiv.org/abs/2510.05070" target="_blank"><i class="far fa-file"></i> paper</a> &nbsp
197+
<a href="https://resmimic.github.io/" target="_blank"><i class="fas fa-globe"></i> website</a>
198+
<p style="margin-top: 5px"><i class="fas fa-comment-dots"></i> TL;DR: A two-stage humanoid skill learning framework that learns a residual policy on top of a pre-trained general motion tracking policy.
179199
</td>
180200
</tr>
181201
</table>
@@ -186,23 +206,55 @@
186206
<tr>
187207
<td style="width:35%; vertical-align:middle; padding-right: 20px;">
188208
<div class="image-container">
189-
<img src='publications/2025_ResMimic.gif' width="90%">
209+
<img src='publications/2025_HDMI.gif' width="90%">
190210
</div>
191211
</td>
192212
<td style="width:65%; vertical-align:middle">
193-
<papertitle>ResMimic: From General Motion Tracking to Humanoid Whole-body Loco-Manipulation via Residual Learning</papertitle>
213+
<papertitle>HDMI: Learning Interactive Humanoid Whole-Body Control from Human Videos</papertitle>
194214
<br>
195-
Siheng Zhao, Yanjie Ze, Yue Wang, C. Karen Liu<sup>&dagger;</sup>, Pieter Abbeel<sup>&dagger;</sup>, Guanya Shi<sup>&dagger;</sup>, Rocky Duan<sup>&dagger;</sup>
215+
Haoyang Weng, Yitang Li, Nikhil Sobanbabu, Zihan Wang, Zhengyi Luo, Tairan He, Deva Ramanan, Guanya Shi
196216
<br>
197-
<a href="https://www.arxiv.org/abs/2510.05070" target="_blank"><i class="far fa-file"></i> paper</a> &nbsp
198-
<a href="https://resmimic.github.io/" target="_blank"><i class="fas fa-globe"></i> website</a>
199-
<p style="margin-top: 5px"><i class="fas fa-comment-dots"></i> TL;DR: A two-stage humanoid skill learning framework that learns a residual policy on top of a pre-trained general motion tracking policy.
217+
<a href="https://arxiv.org/abs/2509.16757" target="_blank"><i class="far fa-file"></i> paper</a> &nbsp
218+
<a href="https://hdmi-humanoid.github.io/#/" target="_blank"><i class="fas fa-globe"></i> website</a> &nbsp
219+
<a href="https://github.com/LeCAR-Lab/HDMI" target="_blank"><i class="fas fa-code"></i> code</a>
220+
<p style="margin-top: 5px"><i class="fas fa-comment-dots"></i> TL;DR: From human videos, HDMI learns robust humanoid loco-manipulation skills (e.g., opening a door continuously for 67 times).
200221
</td>
201222
</tr>
202223
</table>
203224

204225
<br>
205226

227+
<table width="880" border="0" align="center" cellspacing="0" cellpadding="0">
228+
<tr>
229+
<td style="width:35%; vertical-align:middle; padding-right: 20px;">
230+
<div class="image-container">
231+
<img src='publications/2025_LeVERB.gif' width="90%">
232+
</div>
233+
</td>
234+
<td style="width:65%; vertical-align:middle">
235+
<papertitle>LeVERB: Humanoid Whole-Body Control with Latent Vision-Language Instruction</papertitle>
236+
<br>
237+
Haoru Xue<sup>*</sup>, Xiaoyu Huang<sup>*</sup>, Dantong Niu<sup>*</sup>, Qiayuan Liao<sup>*</sup>, Thomas Kragerud, Jan Tommy Gravdahl, Xue Bin Peng, Guanya Shi, Trevor Darrell, Koushil Screenath, Shankar Sastry
238+
<br>
239+
<a href="https://arxiv.org/abs/2506.13751" target="_blank"><i class="far fa-file"></i> paper</a> &nbsp
240+
<a href="https://ember-lab-berkeley.github.io/LeVERB-Website/" target="_blank"><i class="fas fa-globe"></i> website</a> &nbsp
241+
<a href="https://huggingface.co/datasets/ember-lab-berkeley/LeVERB-Bench-Dataset" target="_blank"><i class="fas fa-database"></i> dataset</a>
242+
<p style="margin-top: 5px"><i class="fas fa-comment-dots"></i> TL;DR: LeVERB is a whole-body humanoid VLA, via learning an expressive and executable latent action vocabulary to bridge system 1 and system 2.
243+
</td>
244+
</tr>
245+
</table>
246+
247+
<br>
248+
<br>
249+
250+
<table width="880" border="0" align="center" cellspacing="0" cellpadding="0">
251+
<tr>
252+
<td width="100%" valign="middle">
253+
<heading>2026</heading>
254+
</td>
255+
</tr>
256+
</table>
257+
206258
<table width="880" border="0" align="center" cellspacing="0" cellpadding="0">
207259
<tr style="background-color: var(--highlight-color)">
208260
<td style="width:35%; vertical-align:middle; padding-right: 20px;">
@@ -215,6 +267,8 @@
215267
<br>
216268
Harsh Gupta, Xiaofeng Guo, Huy Ha, Chuer Pan, Muqing Cao, Dongjae Lee, Sebastian Sherer, Shuran Song, Guanya Shi
217269
<br>
270+
<em>International Conference on Robotics and Automation (ICRA)</em>, 2026
271+
<br>
218272
<a href="https://arxiv.org/abs/2510.02614" target="_blank"><i class="far fa-file"></i> paper</a> &nbsp
219273
<a href="https://umi-on-air.github.io/" target="_blank"><i class="fas fa-globe"></i> website</a>
220274
<p style="margin-top: 5px"><i class="fas fa-comment-dots"></i> TL;DR: EADP steers UMI's embodiment-agnostic diffusion policy using the gradient of the low-level controller's tracking cost for cross-embodiment.
@@ -236,6 +290,8 @@
236290
<br>
237291
Lars Ankile, Zhenyu Jiang, Rocky Duan, Guanya Shi, Pieter Abbeel, Anusha Nagabandi
238292
<br>
293+
<em>International Conference on Robotics and Automation (ICRA)</em>, 2026
294+
<br>
239295
<a href="https://arxiv.org/abs/2509.19301" target="_blank"><i class="far fa-file"></i> paper</a> &nbsp
240296
<a href="https://residual-offpolicy-rl.github.io/" target="_blank"><i class="fas fa-globe"></i> website</a>
241297
<p style="margin-top: 5px"><i class="fas fa-comment-dots"></i> TL;DR: ResFiT is a sample-efficient residual RL method that performs real-world RL directly on a wheeled humanoid with two dex hands.
@@ -257,6 +313,8 @@
257313
<br>
258314
Lujie Yang<sup>*</sup>, Xiaoyu Huang<sup>*</sup>, Zhen Wu<sup>*</sup>, Angjoo Kanazawa<sup>&dagger;</sup>, Pieter Abbeel<sup>&dagger;</sup>, Carmelo Sferrazza<sup>&dagger;</sup>, C. Karen Liu<sup>&dagger;</sup>, Rocky Duan<sup>&dagger;</sup>, Guanya Shi<sup>&dagger;</sup>
259315
<br>
316+
<em>International Conference on Robotics and Automation (ICRA)</em>, 2026
317+
<br>
260318
<a href="https://arxiv.org/abs/2509.26633" target="_blank"><i class="far fa-file"></i> paper</a> &nbsp
261319
<a href="https://omniretarget.github.io/" target="_blank"><i class="fas fa-globe"></i> website</a> &nbsp
262320
<a href="https://huggingface.co/datasets/omniretarget/OmniRetarget_Dataset" target="_blank"><i class="fas fa-database"></i> dataset</a>
@@ -271,54 +329,25 @@
271329
<tr>
272330
<td style="width:35%; vertical-align:middle; padding-right: 20px;">
273331
<div class="image-container">
274-
<img src='publications/2025_HDMI.gif' width="90%">
332+
<img src='publications/2025_TWIST2.gif' width="90%">
275333
</div>
276334
</td>
277335
<td style="width:65%; vertical-align:middle">
278-
<papertitle>HDMI: Learning Interactive Humanoid Whole-Body Control from Human Videos</papertitle>
279-
<br>
280-
Haoyang Weng, Yitang Li, Nikhil Sobanbabu, Zihan Wang, Zhengyi Luo, Tairan He, Deva Ramanan, Guanya Shi
336+
<papertitle>TWIST2: Scalable, Portable, and Holistic Humanoid Data Collection System</papertitle>
281337
<br>
282-
<a href="https://arxiv.org/abs/2509.16757" target="_blank"><i class="far fa-file"></i> paper</a> &nbsp
283-
<a href="https://hdmi-humanoid.github.io/#/" target="_blank"><i class="fas fa-globe"></i> website</a> &nbsp
284-
<a href="https://github.com/LeCAR-Lab/HDMI" target="_blank"><i class="fas fa-code"></i> code</a>
285-
<p style="margin-top: 5px"><i class="fas fa-comment-dots"></i> TL;DR: From human videos, HDMI learns robust humanoid loco-manipulation skills (e.g., opening a door continuously for 67 times).
286-
</td>
287-
</tr>
288-
</table>
289-
290-
<br>
291-
292-
<table width="880" border="0" align="center" cellspacing="0" cellpadding="0">
293-
<tr>
294-
<td style="width:35%; vertical-align:middle; padding-right: 20px;">
295-
<div class="image-container">
296-
<img src='publications/2025_LeVERB.gif' width="90%">
297-
</div>
298-
</td>
299-
<td style="width:65%; vertical-align:middle">
300-
<papertitle>LeVERB: Humanoid Whole-Body Control with Latent Vision-Language Instruction</papertitle>
338+
Yanjie Ze, Siheng Zhao, Weizhuo Wang, Angjoo Kanazawa<sup>&dagger;</sup>, Rocky Duan<sup>&dagger;</sup>, Pieter Abbeel<sup>&dagger;</sup>, Guanya Shi<sup>&dagger;</sup>, Jiajun Wu<sup>&dagger;</sup>, C. Karen Liu<sup>&dagger;</sup>
301339
<br>
302-
Haoru Xue<sup>*</sup>, Xiaoyu Huang<sup>*</sup>, Dantong Niu<sup>*</sup>, Qiayuan Liao<sup>*</sup>, Thomas Kragerud, Jan Tommy Gravdahl, Xue Bin Peng, Guanya Shi, Trevor Darrell, Koushil Screenath, Shankar Sastry
340+
<em>International Conference on Robotics and Automation (ICRA)</em>, 2026
303341
<br>
304-
<a href="https://arxiv.org/abs/2506.13751" target="_blank"><i class="far fa-file"></i> paper</a> &nbsp
305-
<a href="https://ember-lab-berkeley.github.io/LeVERB-Website/" target="_blank"><i class="fas fa-globe"></i> website</a> &nbsp
306-
<a href="https://huggingface.co/datasets/ember-lab-berkeley/LeVERB-Bench-Dataset" target="_blank"><i class="fas fa-database"></i> dataset</a>
307-
<p style="margin-top: 5px"><i class="fas fa-comment-dots"></i> TL;DR: LeVERB is a whole-body humanoid VLA, via learning an expressive and executable latent action vocabulary to bridge system 1 and system 2.
342+
<a href="https://arxiv.org/abs/2511.02832" target="_blank"><i class="far fa-file"></i> paper</a> &nbsp
343+
<a href="https://yanjieze.com/TWIST2/" target="_blank"><i class="fas fa-globe"></i> website</a> &nbsp
344+
<a href="https://github.com/amazon-far/TWIST2" target="_blank"><i class="fas fa-code"></i> code</a>
345+
<p style="margin-top: 5px"><i class="fas fa-comment-dots"></i> TL;DR: TWIST2 is a portable whole-body humanoid teleoperation system that enables scalable data collection.
308346
</td>
309347
</tr>
310348
</table>
311349

312-
<br>
313-
<br>
314-
315-
<table width="880" border="0" align="center" cellspacing="0" cellpadding="0">
316-
<tr>
317-
<td width="100%" valign="middle">
318-
<heading>2026</heading>
319-
</td>
320-
</tr>
321-
</table>
350+
<br>
322351

323352
<table width="880" border="0" align="center" cellspacing="0" cellpadding="0">
324353
<tr style="background-color: var(--highlight-color)">
@@ -1693,7 +1722,7 @@
16931722
</tr>
16941723
</table>
16951724

1696-
<br>
1725+
<!-- <br>
16971726
<br>
16981727
16991728
<table width="880" border="0" align="center" cellspacing="0" cellpadding="0">
@@ -1721,7 +1750,7 @@
17211750
</p>
17221751
</td>
17231752
</tr>
1724-
</table>
1753+
</table> -->
17251754

17261755
<br>
17271756
<br>

‎publications/2026_RPL.gif‎

2.37 MB
Loading

0 commit comments

Comments
 (0)