-
Notifications
You must be signed in to change notification settings - Fork 7
/
index.html
519 lines (451 loc) · 20.6 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
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<meta name="description"
content="VoxPoser: Composable 3D Value Maps for Robotic Manipulation with Language Models">
<meta name="keywords" content="Robotic Manipulation, Language Models">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>VoxPoser</title>
<script async src="https://www.googletagmanager.com/gtag/js?id=G-FV4ZJ9PVSV"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag(){dataLayer.push(arguments);}
gtag('js', new Date());
gtag('config', 'G-FV4ZJ9PVSV');
</script>
<script>
function updateInteractive() {
var task = document.getElementById("interative-menu").value;
console.log("interactive", task)
// if task does not contain the string "towel"
if (task.indexOf("towel") == -1) {
var video = document.getElementById("interactive-video");
video.src = "media/videos/" +
task +
".mp4"
video.play();
var html = document.getElementById("interactive-html-1");
html.src = "media/interactive/" +
task +
".html"
// hide the second iframe container
var iframeContainer2 = document.getElementById("second-iframe-container");
iframeContainer2.style.display = "none";
} else {
var video = document.getElementById("interactive-video");
video.src = "media/videos/hang-towel.mp4"
video.play();
var html1 = document.getElementById("interactive-html-1");
html1.src = "media/interactive/hang-towel-1.html"
// show and set the source for the second iframe
var html2 = document.getElementById("interactive-html-2");
html2.src = "media/interactive/hang-towel-2.html"
// show the second iframe container
var iframeContainer2 = document.getElementById("second-iframe-container");
iframeContainer2.style.display = "block";
}
}
</script>
<link href="https://fonts.googleapis.com/css?family=Google+Sans|Noto+Sans|Castoro"
rel="stylesheet">
<link rel="stylesheet" href="./static/css/bulma.min.css">
<link rel="stylesheet" href="./static/css/bulma-carousel.min.css">
<link rel="stylesheet" href="./static/css/bulma-slider.min.css">
<link rel="stylesheet" href="./static/css/fontawesome.all.min.css">
<link rel="stylesheet"
href="https://cdn.jsdelivr.net/gh/jpswalsh/academicons@1/css/academicons.min.css">
<link rel="stylesheet" href="./static/css/index.css">
<script src="https://ajax.googleapis.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script>
<script defer src="./static/js/fontawesome.all.min.js"></script>
<script src="./static/js/bulma-carousel.min.js"></script>
<script src="./static/js/bulma-slider.min.js"></script>
<script src="./static/js/index.js"></script>
</head>
<body onload="updateInteractive();">
<section class="hero">
<div class="hero-body">
<div class="container is-fullhd">
<div class="columns is-centered">
<div class="column has-text-centered">
<h1 class="title is-1 publication-title">VoxPoser: Composable 3D Value Maps<br>for Robotic Manipulation with Language Models</h1>
<h3 class="title is-4 conference-authors"><a target="_blank" href="https://www.corl2023.org/">CoRL 2023 (Oral)</a></h3>
<div class="is-size-5 publication-authors">
<span class="author-block">
<a target="_blank" href="https://huangwl18.github.io/">Wenlong Huang</a><sup>1</sup>,
</span>
<span class="author-block">
<a target="_blank" href="https://www.chenwangjeremy.net/">Chen Wang</a><sup>1</sup>,
</span>
<span class="author-block">
<a target="_blank" href="https://ai.stanford.edu/~zharu/">Ruohan Zhang</a><sup>1</sup>,
</span>
<span class="author-block">
<a target="_blank" href="https://yunzhuli.github.io/">Yunzhu Li</a><sup>1, 2</sup>,
</span>
<span class="author-block">
<a target="_blank" href="https://jiajunwu.com/">Jiajun Wu</a><sup>1</sup>,
</span>
<span class="author-block">
<a target="_blank" href="https://profiles.stanford.edu/fei-fei-li">Li Fei-Fei</a><sup>1</sup>
</span>
</div>
<div class="is-size-5 publication-authors">
<span class="author-block"><sup>1</sup>Stanford University,</span>
<span class="author-block"><sup>2</sup>University of Illinois Urbana-Champaign</span>
</div>
<div class="column has-text-centered">
<div class="publication-links">
<!-- PDF Link. -->
<span class="link-block">
<a target="_blank" href="voxposer.pdf"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<!-- Arxiv Link. -->
<span class="link-block">
<a target="_blank" href="https://arxiv.org/abs/2307.05973"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file"></i>
</span>
<span>ArXiv</span>
</a>
</span>
<!-- Video Link. -->
<span class="link-block">
<a target="_blank" href="https://youtu.be/Yvn4eR05A3M"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fab fa-youtube"></i>
</span>
<span>Video</span>
</a>
</span>
<!-- Code Link. -->
<span class="link-block">
<a target="_blank" href="https://github.com/huangwl18/VoxPoser"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fab fa-github"></i>
</span>
<span>Code</span>
</a>
</span>
</div>
</div>
</div>
</div>
</div>
</div>
</section>
<section class="hero teaser">
<div class="container is-fullhd">
<div class="hero-body">
<div class="container">
<div class="columns is-vcentered is-centered">
<video id="teaser" autoplay muted loop height="60%" width="60%">
<source src="media/videos/teaser.mp4"
type="video/mp4">
</video>
</br>
</div>
<br>
<h2 class="subtitle has-text-centered">
<span class="dperact">VoxPoser</span> extracts <b>affordances</b> and <b>constraints</b> from large language models and vision-language models<br>to compose 3D value maps, which are used by motion planners to <b>zero-shot synthesize</b> trajectories for everyday manipulation tasks.
</h2>
</div>
</div>
</div>
</section>
<section class="hero is-light is-small">
<div class="hero-body">
<div class="container">
<div id="results-carousel" class="carousel results-carousel">
<div class="item">
<video poster="" autoplay muted loop height="100%">
<source src="media/videos/take-out-the-toaster-and-put-it-on-the-wooden-plate.mp4"
type="video/mp4">
</video>
</div>
<div class="item">
<video poster="" autoplay muted loop height="100%">
<source src="media/videos/close-top-drawer-dist.mp4"
type="video/mp4">
</video>
</div>
<div class="item">
<video poster="" autoplay muted loop height="100%">
<source src="media/videos/sort-trash-to-tray-dist.mp4"
type="video/mp4">
</video>
</div>
<div class="item">
<video poster="" autoplay muted loop height="100%">
<source src="media/videos/open-bottle.mp4"
type="video/mp4">
</video>
</div>
<div class="item">
<video poster="" autoplay muted loop height="100%">
<source src="media/videos/turn-on-the-lamp.mp4"
type="video/mp4">
</video>
</div>
<div class="item">
<video poster="" autoplay muted loop height="100%">
<source src="media/videos/sweep-the-trash-into-the-dustpan.mp4"
type="video/mp4">
</video>
</div>
<div class="item">
<video poster="" autoplay muted loop height="100%">
<source src="media/videos/set-up-the-utensils-for-my-pasta.mp4"
type="video/mp4">
</video>
</div>
<div class="item">
<video poster="" autoplay muted loop height="100%">
<source src="media/videos/unplug-charger.mp4"
type="video/mp4">
</video>
</div>
<div class="item">
<video poster="" autoplay muted loop height="100%">
<source src="media/videos/get-napkin.mp4"
type="video/mp4">
</video>
</div>
<div class="item">
<video poster="" autoplay muted loop height="100%">
<source src="media/videos/hang-towel.mp4"
type="video/mp4">
</video>
</div>
<div class="item">
<video poster="" autoplay muted loop height="100%">
<source src="media/videos/measure-apple.mp4"
type="video/mp4">
</video>
</div>
</div>
</div>
</div>
</div>
</section>
<h2 class="subtitle has-text-centered">
</br>
VoxPoser can <b>zero-shot synthesize</b> trajectories for real-world manipulation tasks with an <b>open-set</b> of free-form language instructions and an <b>open-set</b> of objects.
</h2>
<section class="section">
<div class="container is-max-desktop">
<!-- Abstract. -->
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">Abstract</h2>
<div class="content has-text-justified">
<p>
Large language models (LLMs) are shown to possess a wealth of actionable knowledge that can be extracted for robot manipulation in the form of reasoning and planning.
Despite the progress, most still rely on pre-defined motion primitives to carry out the physical interactions with the environment, which remains a major bottleneck.
In this work, we aim to synthesize robot trajectories, i.e., a dense sequence of 6-DoF end-effector waypoints,
for a large variety of manipulation tasks given an <i>open-set</i> of instructions and an <i>open-set</i> of objects.
We achieve this by first observing that LLMs excel at inferring affordances and constraints given a free-form language instruction
More importantly, by leveraging their code-writing capabilities, they can interact with a vision-language model (VLM) to compose 3D value maps to ground the knowledge into the observation space of the agent.
The composed value maps are then used in a model-based planning framework to <i>zero-shot</i> synthesize closed-loop robot trajectories with robustness to dynamic perturbations.
We further demonstrate how the proposed framework can benefit from online experiences by efficiently learning a dynamics model for scenes that involve contact-rich interactions.
We present a large-scale study of the proposed method in both simulated and real-robot environments, showcasing the ability to perform a large variety of everyday manipulation tasks specified in free-form natural language.
</p>
</div>
</div>
</div>
<!--/ Abstract. -->
</div>
<!-- Paper video. -->
<br>
<br>
<div class="container is-max-widescreen">
<div class="rows">
<h2 class="title is-3">Video</h2>
<div class="publication-video">
<iframe src="https://www.youtube.com/embed/Yvn4eR05A3M"
frameborder="0" allow="autoplay; encrypted-media" allowfullscreen></iframe>
</div>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-widescreen">
<div class="rows">
<!-- Animation. -->
<div class="rows is-centered ">
<div class="row is-full-width">
<h2 class="title is-3"><span class="dperact">VoxPoser</span></h2>
<!-- Interpolating. -->
<div class="content has-text-justified">
<!-- <br> -->
</div>
<img src="media/figures/method.jpg" class="interpolation-image" />
</br>
</br>
<p class="content has-text-justified">
Given the RGB-D observation of the environment and a language instruction, LLMs generate code, which interacts with VLMs, to produce a sequence of 3D affordance maps and constraint maps (collectively referred to as value maps) grounded in the observation space of the robot <strong>(a)</strong>. The composed value maps then serve as objective functions for motion planners to synthesize trajectories for robot manipulation <strong>(b)</strong>. The entire process does not involve any additional training.
</p>
</br>
</br>
<!--/ Re-rendering. -->
<h2 class="title is-3">Interactive Visualization</h2>
<div class="columns">
<div class="column has-text-centered">
Visualize value maps for
<div class="select is-small is-rounded">
<select id="interative-menu" onchange="updateInteractive()">
<option value="hang-towel-1" selected="selected">"Hang towel on rack"</option>
<option value="close-drawer">"Close the top drawer."</option>
<option value="get-napkin">"Take out a napkin."</option>
<option value="open-bottle">"Turn open vitamin bottle."</option>
</select>
</div>
</div>
</div>
<div class="columns">
<div class="column is-half has-text-centered">
<p style="text-align:center;">
<video id="interactive-video" width="100%" height="100%" controls autoplay loop muted>
<source src="media/videos/close-drawer.mp4" type="video/mp4">
</video>
</p>
</div>
<div class="column has-text-centered">
<iframe id="interactive-html-1" src="media/interactive/close-drawer.html" width="100%" height="300" frameborder="0"></iframe>
<p style="text-align:center;">
Interactive Value Map 1
</p>
</div>
<div class="column has-text-centered" id="second-iframe-container">
<iframe id="interactive-html-2" src="media/interactive/close-drawer.html" width="100%" height="300" frameborder="0"></iframe>
<p style="text-align:center;">
Interactive Value Map 2
</p>
</div>
</div>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-widescreen">
<div class="rows">
<h2 class="title is-3">Execution under Disturbances</h2>
<p class="content has-text-justified">
Because the language model output stays the same throughout the task, we can cache its output and re-evaluate the generated code using closed-loop visual feedback, which enables fast replanning using MPC. This enables VoxPoser to be robust to online disturbances.
</p>
<div class="columns">
<div class="column has-text-centered">
<video id="dist1"
controls
muted
autoplay
loop
width="99%">
<source src="media/videos/sort-trash-to-tray-dist.mp4"
type="video/mp4">
</video>
<p style="text-align:center">
"Sort the paper trash into the blue tray."
</p>
</div>
<div class="column has-text-centered">
<video id="dist2"
controls
muted
autoplay
loop
width="99%">
<source src="media/videos/close-top-drawer-dist.mp4"
type="video/mp4">
</video>
<p style="text-align:center">
"Close the top drawer."
</p>
</div>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-widescreen">
<div class="rows">
<h2 class="title is-3">Emergent Behavioral Capabilities</h2>
<p class="content has-text-justified">
<li><strong>Behavioral Commonsense Reasoning</strong>: During a task where robot is setting the table, the user can specify behavioral preferences such as “I am left-handed”, which requires the robot to comprehend its meaning in the context of the task. VoxPoser decides that it should move the fork from the right side of the bowl to the left side.</li>
<li><strong>Fine-grained Language Correction</strong>: For tasks that require high precision such as “covering the teapot with the lid”, the user can give precise instructions to the robot such as “you’re off by 1cm”. VoxPoser similarly adjusts its action based on the feedback.</li>
<li><strong>Multi-step Visual Program</strong>: Given a task “open the drawer precisely by half” where there is insufficient information because object models are not available, VoxPoser can come up with multi-step manipulation strategies based on visual feedback that first opens the drawer fully while recording handle displacement, then close it back to the midpoint to satisfy the requirement.</li>
</p>
<br>
<div style="text-align: center;">
<img src="media/figures/emergent.jpg" class="interpolation-image" style="width: 100%; height: auto;"/>
</div>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-widescreen">
<div class="row">
<h2 class="title is-3">
Prompts
</h2>
<p class="content has-text-justified">
Prompts in Real-World Environments:
<br>
<!-- https://vosposer-anon.github.io/ -->
<a href="prompts/real_planner_prompt.txt">Planner</a> |
<a href="prompts/real_composer_prompt.txt">Composer</a> |
<a href="prompts/real_parse_query_obj_prompt.txt">Parse Query Object</a> |
<a href="prompts/real_get_affordance_map_prompt.txt">Get Affordance Maps</a> |
<a href="prompts/real_get_avoidance_map_prompt.txt">Get Avoidance Maps</a> |
<a href="prompts/real_get_rotation_map_prompt.txt">Get Rotation Maps</a> |
<a href="prompts/real_get_velocity_map_prompt.txt">Get Velocity Maps</a> |
<a href="prompts/real_get_gripper_map_prompt.txt">Get Gripper Maps</a>
</p>
<p class="content has-text-justified">
Prompts in Simulation Environments (planners are not used in simulation):
<br>
<a href="prompts/sim_composer_prompt.txt">Composer</a> |
<a href="prompts/sim_parse_query_obj_prompt.txt">Parse Query Object</a> |
<a href="prompts/sim_get_affordance_map_prompt.txt">Get Affordance Maps</a> |
<a href="prompts/sim_get_avoidance_map_prompt.txt">Get Avoidance Maps</a> |
<a href="prompts/sim_get_rotation_map_prompt.txt">Get Rotation Maps</a> |
<a href="prompts/sim_get_velocity_map_prompt.txt">Get Velocity Maps</a> |
<a href="prompts/sim_get_gripper_map_prompt.txt">Get Gripper Maps</a>
</p>
</div>
</div>
</section>
<section class="section" id="BibTeX">
<div class="container is-max-widescreen content">
<h2 class="title">BibTeX</h2>
<pre><code>@article{huang2023voxposer,
title={VoxPoser: Composable 3D Value Maps for Robotic Manipulation with Language Models},
author={Huang, Wenlong and Wang, Chen and Zhang, Ruohan and Li, Yunzhu and Wu, Jiajun and Fei-Fei, Li},
journal={arXiv preprint arXiv:2307.05973},
year={2023}
}</code></pre>
</div>
</section>
<footer class="footer">
<div class="container">
<div class="columns is-centered">
<div class="column">
<div class="content has-text-centered">
<p>
Website template borrowed from <a href="https://github.com/nerfies/nerfies.github.io">NeRFies</a> and <a href="https://peract.github.io/">PerAct</a>.
</p>
</div>
</div>
</div>
</div>
</footer>
</body>
</html>