-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathindex.html
530 lines (460 loc) · 19.7 KB
/
index.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
<!-- <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd"> -->
<!-- saved from url=(0028)https://mayuelala.github.io/ -->
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en">
<head>
<!-- <link rel="shortcut icon" href="myIcon.ico"> -->
<meta http-equiv="Content-Type" content="text/html;charset=utf-8" />
<meta name="keywords" content="SYSU, PCL, MBZUAI, SUST">
<meta name="description" content="SYSU">
<meta name="google-site-verification" content="yy_3iiS_X6pJdegdwitJMrH0LRLHXwpjrV9RKLXxKjg" />
<meta name="google-site-verification" content="yy_3iiS_X6pJdegdwitJMrH0LRLHXwpjrV9RKLXxKjg" />
<link rel="stylesheet" href="jemdoc.css" type="text/css">
<title>CVPR 2025 Embodied AI Workshop Challenge:
Social Mobile Manipulation Challenge</title>
<link href="./public/index.css" rel="stylesheet" />
<link href="./public/media.css" rel="stylesheet" />
<link href="./public/sidebars.css" rel="stylesheet" />
<link rel="stylesheet" href="./public/workshop.css">
<style>
.speaker {
text-align: center;
width: calc(25%);
box-sizing: border-box;
}
.speaker img {
width: 100%;
height: auto;
border-radius: 50%;
margin-bottom: 10px;
}
.speaker p {
margin: 5px 0;
}
@media (max-width: 768px) {
.speaker {
width: calc(50% - 20px);
}
}
@media (max-width: 480px) {
.speaker {
width: 100%;
}
}
</style>
<style>
table {
border-collapse: collapse;
}
.smaller-image {
width: 20%;
}
.container {
position: relative;
display: inline-block;
}
.image {
display: block;
border-radius: 10px;
max-width: 100%;
}
.video-label {
position: absolute;
top: 5px;
left: 5px;
background-color: #ea3323;
color: white;
padding: 0px 10px;
border-radius: 3px;
font-weight: bold;
box-shadow: 0 4px 8px 0 rgba(0, 0, 0, 0.2), 0 6px 20px 0 rgba(0, 0, 0, 0.19);
width: 40px;
height: 20px;
display: flex;
justify-content: center;
align-items: center;
font-size: small;
}
.research-hightlight {
display: grid;
grid-template-columns: repeat(2, 1fr);
/* 5列 */
/* grid-template-rows: repeat(4, 1fr); */
gap: 10px;
/* 网格项之间的间隙 */
}
.arxiv-label {
position: absolute;
top: 10px;
left: -5px;
background-color: #a624a6;
color: white;
padding: 0px 10px;
border-radius: 3px;
font-weight: bold;
box-shadow: 0 4px 8px 0 rgba(0, 0, 0, 0.2), 0 6px 20px 0 rgba(0, 0, 0, 0.19);
width: 40px;
height: 20px;
display: flex;
justify-content: center;
align-items: center;
font-size: x-small;
}
</style>
<style>
.video-container {
text-align: center;
}
iframe {
display: block;
margin: 0 auto;
}
</style>
<script type="text/javascript" src="./jquery-1.12.4.min.js"></script>
<meta name="google-site-verification" content="RmRCKoAXAT6rtzTsOTkyvuJiKVINXoP9VDPktJ0Hl3Q" />
</head>
<body data-new-gr-c-s-check-loaded="14.1163.0" data-gr-ext-installed="">
<div id="layout-content" style="margin-top:25px">
<div class="background-image" style="position: relative; width: 100%; height: auto;">
<img src="imgs/background.jpg" alt="Background Image" style="width: 100%; height: auto; display: block;">
<div class="mask"
style="position: absolute; top: 0; left: 0; width: 100%; height: 100%; background: rgba(0, 0, 0, 0.6); z-index: 1;">
</div>
<div class="content"
style="position: absolute; top: 50%; left: 50%; transform: translate(-50%, -50%); z-index: 2; text-align: center; color: white;">
<h1 style="color: white;">
<strong>CVPR 2025 Embodied AI Workshop Challenge:
<br> Social Mobile Manipulation Challenge<br></strong>
</h1>
Sat February 1th - Thu May 1th, 2025<br>
SYSU, Guangzhou, China
<section class="links">
<ul>
<a href="#challenge-details" rel="noreferrer">
<li>
<span>Challenge Details</span>
</li>
</a>
<a href="" rel="noreferrer">
<li>
<span>LeaderBoard (Coming Soon)</span>
</li>
</a>
<a href="#organizers" rel="noreferrer">
<li>
<span>Organizers</span>
</li>
</a>
<!-- 这里填问卷链接 -->
<!-- <a href="https://docs.google.com/forms/d/e/1FAIpQLSfbxXDwFV-g7x_RhciBWep3b1Hldt0YRlqr1ELAKBwoPVIJeQ/viewform?usp=sharing"
rel="noreferrer" target="_blank">
<li style="background-color: green">
<span>Pre-Registration</span>
</li>
</a> -->
</ul>
</section>
</div>
</div>
<div style="height: 20px;"></div>
<!-- Overall -->
<section style="text-align: center; width: 95%; margin: 0 auto;">
<p class="abstract" style="font-family: 'Times New Roman', Arial; text-align: left;">
The 1st Social Mobile Manipulation Challenge focuses on developing embodied AI agents capable of
performing long sequences of complex tasks through social interactions. These tasks involve reasoning
about human intentions and planning within dynamic, multi-agent environments (Figure 1). The challenge
includes two interaction modes: human-robot and robot-robot. The goal is to advance the community's
capabilities in areas like human intention reasoning, long-term task planning, and effective social
interaction between embodied agents.<br>
<img src="imgs/asset.png" alt="Background Image" style="width: 100%; height: auto; display: block;">
</p>
<p class="abstract" style="font-family: 'Times New Roman', Arial; text-align: center;">
Figure 1: The dynamic interaction environment that agents will navigate in both human-robot and
robot-robot tasks.<br>
</p>
<p class="abstract" style="font-family: 'Times New Roman', Arial; text-align: center;">
<h3>BASELINE</h3>
</p>
<p class="abstract" style="font-family: 'Times New Roman', Arial; text-align: left;">
<strong> Hierarchical interaction: </strong> <br>To simulate the agent interaction mode with a
hierarchical knowledge structure in the environment.<br>
<strong> Horizontal interaction: </strong> <br>To simulate the "passer-by interaction scene".<br>
<!-- <strong> VLMZero-Shot: </strong> <br>By inputting the global scene information and current observations
into a vision-language model (VLM),we use prompt engineering to output the actions that the agent should
execute.<br>
<strong> Single Semantic Map:</strong><br>Goal-Oriented Semantic Exploration for 2D semantic
mapping,while
employing the FBE algorithm as the
global planner incombination with the FMM planning algorithm for local planning.<br>
<strong> Random: </strong><br>In the robot's action space, actions are randomly sampled for execution,
or
target points are randomly sampled in the planning space, and planning algorithms are used to solve for
them.<br>
<strong> LLM-Based Planning: </strong><br>Using the Co-NavGPT,we employ a large language model (LLM) as
a
planner for multi-agent systems.The merged observation map of the agents is converted intoa textual
description,which is then processed by the LLM to perform goal planning for multiple agents.<br>
<strong>LLM-Planner </strong> <br>is a few-shot grounded planning model.Different from common planning
models,LLM Planner uses LLMs to generate plans directly instead of ranking acceptable skills, reducing
the need for sufficient prior knowledge of the environment and the number of calls to LLMs.Re-planning
of LLM-Planner allow sit to dynamically adjust the planning based on current observations,resulting in
more informed plans. -->
</p>
<div class="video-container">
<!-- <iframe style='width: 600px;height: 338px' frameborder='no' allowfullscreen mozallowfullscreen
webkitallowfullscreen
src='http://go.plvideo.cn/front/video/preview?vid=t1c133894e726cf17eb9313d4d98346d_t'></iframe> -->
<script src='//player.polyv.net/resp/vod-player/latest/player.js'></script>
<div id='plv_t1c133894e726cf17eb9313d4d98346d_t'></div>
<script>
var player = polyvPlayer({
'wrap': '#plv_t1c133894e726cf17eb9313d4d98346d_t',
'width': '600',
'height': '338',
'vid': 't1c133894e726cf17eb9313d4d98346d_t',
'playsafe': '' // 播放加密视频的凭证, 取值参考文档: https://help.polyv.net/index.html#/vod/api/playsafe/token/create_token
});
</script>
</div>
<p style="text-align:center">
Vedio1: Two Stretch collaborate to build the scene graph<br>
</p>
<p style="text-align: left;">
<strong>
Benchmark1: Scene Graph Collaborative Exploration.
</strong><br>
In traditional single-robot systems, the robot explores unknown areas sequentially, gradually building
up the scene graph. However, this approach is often inefficient in large scale or dynamically changing
environments, limiting the speed of scene graph construction and the richness of information obtained.
Introducing multi-agent scene graph construction can significantly improve the efficiency and quality of
this process. Multiple robots work collaboratively, sharing information and merging their views to build
a unified scene graph. While each robot independently perceives and maps parts of the environment, the
agents share map data, update object semantic labels, and synchronize their positions via wireless
communication, effectively boosting mapping efficiency.
</p>
<div class="video-container">
<!-- <iframe style='width: 600px;height: 338px;' frameborder='no' allowfullscreen mozallowfullscreen
webkitallowfullscreen
src='http://go.plvideo.cn/front/video/preview?vid=t1c133894e5a67a12c7d237320cab33f_t'></iframe> -->
<script src='//player.polyv.net/resp/vod-player/latest/player.js'></script>
<div id='plv_t1c133894e5a67a12c7d237320cab33f_t'></div>
<script>
var player = polyvPlayer({
'wrap': '#plv_t1c133894e5a67a12c7d237320cab33f_t',
'width': '600',
'height': '338',
'vid': 't1c133894e5a67a12c7d237320cab33f_t',
'playsafe': '' // 播放加密视频的凭证, 取值参考文档: https://help.polyv.net/index.html#/vod/api/playsafe/token/create_token
});
</script>
</div>
<p style="text-align:center">
Vedio2: Two Go2 collaborate to grab a bottle of beverage and interact with humans during the process<br>
</p>
<p style="text-align: left;">
<strong>
Benchmark2: Open World Social Mobile Manipulation.
</strong><br>
In this benchmark, we designed an open-world social mobile manipulation.It mainly includes two
interaction methods: hierarchical interaction and horizontal interaction. The former simulates embodied
AI interaction with hierarchical knowledge structure, and the latter simulates embodied AI interaction
with equal knowledge acquisition capabilities.<br>
<strong>Hierarchical interaction:</strong> <br>
In hierarchical interaction tasks, it is used to simulate the agent
interaction mode with a hierarchical knowledge structure in the environment. For example, compared to
ordinary agents, administrators (such as salespersons, etc.) clearly have more knowledge about the
environment. Encouraging agents to have conversations with administrators, can help agents
better understand user intentions and improve task execution success rates. <br>
<strong>Horizontal interaction.</strong> <br>
In horizontal interaction tasks, it is used to simulate the “passer-by
interaction scene”. There is no administrator with a “God's perspective” in the scene, and all agents
can obtain scene knowledge equally. Specifically, the scene contains multiple agents with the same
status. They can independently build their own scene graphs and transfer knowledge through social
dialogue to improve the efficiency and success rate of task completion.
</p>
</section>
<!-- Competition Content -->
<section id="challenge-details">
<section class="details" style="text-align: justify; margin-top: 20px;">
<div style="display: flex; flex-direction: column; align-items: center; width: 100%; ">
<h2><strong> Challenge Details</strong></h2>
<!-- <div style="margin-top: -10px"></div> -->
<!-- <strong style="color: red;">See <a href="">Competition Manual (Coming Soon)</a> for more
details</strong> -->
<!-- <div style="margin-top: 10px"></div> -->
</div>
</section>
<!-- <hr> -->
<!-- <div style="margin-top: -10px"></div> -->
<!-- <h3>Track 1: Rigid Object Manipulation</h3> -->
<div
style="display: flex; flex-direction: column; align-items: center; justify-content: center; text-align: center;">
<!-- <video autoplay loop muted style="width: 40%;">
<source src="https://www.bilibili.com/video/BV1KbPdeKE1K?t=3.0" type="video/mp4">
</video> -->
<div style="margin-top: 10px;">
<a href="https://github.com/pzhren/InfiniteWorld"
style="display: block; color: blue; text-decoration: none;">SMM</a><br>
</div>
</div>
<div style="margin-top: -10px"></div>
<h4>Task Overview</h4>
<div style="margin-top: -10px"></div>
The 1st Social Mobile Manipulation Challenge focuses on developing embodied AI agents capable of
performing long sequences of complex tasks through social interactions. These tasks involve reasoning
about human intentions and planning within dynamic, multi-agent environments (Figure 1). The challenge
includes two interaction modes: human-robot and robot-robot. The goal is to advance the community's
capabilities in areas like human intention reasoning, long-term task planning, and effective social
interaction between embodied agents.
<h4>Dataset</h4>
<div style="margin-top: -10px"></div>
We will provide two main datasets for this challenge. The first dataset includes various human-robot
interaction tasks, designed for hierarchical knowledge-based interactions. The second dataset involves
robot-robot interactions where multiple agents collaborate on equal terms to complete tasks.
These datasets are currently under development and are expected to be released by 03/01/2025. In case of
delays, we plan to provide smaller, simplified versions of the datasets to ensure participants can begin
their work on time.
<h4>Ethical considerations</h4>
<div style="margin-top: -10px"></div>
Since the challenge involves the development of autonomous agents, careful attention is required when
transferring policies trained in simulation to real-world environments. Proper validation processes should
be established to avoid unintended behaviors in robots, especially in human-robot interaction scenarios.
<h4>Submission evaluation</h4>
<div style="margin-top: -10px"></div>
Participants will submit their solution files, which will be evaluated using our simulation platform based
on NVIDIA's Isaac Sim. We will use the evalAI platform to manage submissions and rerun code for the top
submissions. Participants are expected to submit both their code and a ReadMe file with clear instructions
on how to execute their solution.
<h4>Timeline</h4>
<div style="margin-top: -10px"></div>
The challenge will commence on 01/02/2025, with submissions due by 01/05/2025. Final decisions will be
shared with participants by 01/06/2025.
<!-- <h4>Challenge organizers</h4>
<div style="margin-top: -10px"></div>
The challenge is organized by Xiaodan Liang (MBZUAI), Rongtao Xu (MBZUAI), Pengzhen Ren (Pengcheng
Laboratory), Bingqian Lin (Shanghai Jiao Tong University), Xiaojun Chang (University of Technology Sydney),
Ivan Laptev (MBZUAI), Ian Reid(MBZUAI), Cewu Lu (Shanghai Jiao Tong University), Mingfei Han (MBZUAI), and
Xiwen Liang (Sun Yat-sen University).
<br> -->
</section>
<div style="margin-top: 20px"></div>
<hr>
<!-- Organizer -->
<section id="organizers">
<a class="anchor" id="organizer"></a>
<section class="details" style="text-align: justify; margin-top: 10px;">
<div style="display: flex; flex-direction: column; align-items: center; width: 100%; height: 100%;">
<h2>Organizers</h2>
</div>
<div class="grid" style="max-width: 90%;">
<div class="speaker" style="margin-top:-800px;">
<figure>
<img src="imgs\organizers\liangxiaodan.png" />
<figcaption><b><a href="https://lemondan.github.io">Xiaodan
Liang</a></b><br />MBZUAI
</figcaption>
</figure>
</div>
<div class="speaker" style="margin-top:-800px;">
<figure>
<img src="imgs\organizers\xurongtao.jpg" />
<figcaption><b><a
href="https://scholar.google.com.hk/citations?user=_IUq7ooAAAAJ&hl=zh-CN&authuser=1">Rongtao
Xu</a></b><br />MBZUAI
</figcaption>
</figure>
</div>
<div class="speaker" style="margin-top:-800px;">
<figure>
<img src="imgs\organizers\rpz.png" />
<figcaption><b><a
href="https://scholar.google.com.hk/citations?user=yVxSn70AAAAJ&hl=zh-CN">Pengzhen
Ren</a></b><br />Pengcheng
Laboratory</figcaption>
</figure>
</div>
<div class="speaker" style="margin-top:-800px;">
<figure>
<img src="imgs\organizers\linbingqian.png" />
<figcaption><b><a href="https://expectorlin.github.io/">Bingqian
Lin</a></b><br />SJTU</figcaption>
</figure>
</div>
<div class="speaker" style="margin-top:-600px;">
<figure>
<img src="imgs\organizers\changxiaojun.jpg" />
<figcaption><b><a href="https://www.xiaojun.ai/">Xiaojun Chang</a></b><br />UTS
</figcaption>
</figure>
</div>
<div class="speaker" style="margin-top:-600px;">
<figure>
<img src="imgs\organizers\IvanLaptev.jpg" />
<figcaption><b><a href="https://www.di.ens.fr/~laptev/">Ivan Laptev</a></b><br />MBZUAI
</figcaption>
</figure>
</div>
<div class="speaker" style="margin-top:-600px;">
<figure>
<img src="imgs\organizers\IanReid.jpg" />
<figcaption><b><a href="">Ian Reid</a></b><br />MBZUAI</figcaption>
</figure>
</div>
<div class="speaker" style="margin-top:-600px;">
<figure>
<img src="imgs\organizers\lucewu.png" />
<figcaption><b><a href="">Cewu Lu</a></b><br />SJTU
</figcaption>
</figure>
</div>
<div class="speaker" style="margin-top:-400px;">
<figure>
<img src="imgs\organizers\hanmingfei.jpg" />
<figcaption><b><a href="https://mingfei.info/">Mingfei Han</a></b><br />MBZUAI
</figcaption>
</figure>
</div>
<div class="speaker" style="margin-top:-400px;">
<figure>
<img src="imgs\organizers\liangxiwen.jpg" />
<figcaption><b><a href="https://scholar.google.com/citations?user=Iwj59kkAAAAJ">Xiwen
Liang</a></b><br />SYSU</figcaption>
</figure>
</div>
</div>
</section>
</section>
<div style="margin-top: -100px"></div>
<hr>
<!-- <div style="margin-top: 30px"></div> -->
<!-- <div style="margin-top: 60px;">
<div style="height: 30px;"></div>
</div> -->
</div>
<div class="jvectormap-tip"></div>
</body>
<grammarly-desktop-integration data-grammarly-shadow-root="true"><template shadowrootmode="open">
<style>
div.grammarly-desktop-integration {
position: absolute;
width: 1px;
height: 1px;
padding: 0;
margin: -1px;
overflow: hidden;
clip: rect(0, 0, 0, 0);
white-space: nowrap;
border: 0;
-moz-user-select: none;
-webkit-user-select: none;
-ms-user-select: none;
user-select: none;
}
div.grammarly-desktop-integration:before {
content: attr(data-content);
}
</style>
<div aria-label="grammarly-integration" role="group" tabindex="-1" class="grammarly-desktop-integration"
data-content="{"mode":"full","isActive":true,"isUserDisabled":false}">
</div>
</template></grammarly-desktop-integration>
</html>