-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathindex.html
573 lines (522 loc) · 25.8 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
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<meta name="description"
content="CLIPort: What and Where Pathways for Robotic Manipulation.">
<meta name="keywords" content="Vision-Language Grounding, Manipulation, CLIP">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>CLIPort</title>
<!-- Global site tag (gtag.js) - Google Analytics -->
<script async src="https://www.googletagmanager.com/gtag/js?id=G-4Y34PZ3XBE"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag(){dataLayer.push(arguments);}
gtag('js', new Date());
gtag('config', 'G-4Y34PZ3XBE');
</script>
<script>
function updateSingleVideo() {
var demo = document.getElementById("single-menu-demos").value;
var task = document.getElementById("single-menu-tasks").value;
var inst = document.getElementById("single-menu-instances").value;
console.log("single", demo, task, inst)
var video = document.getElementById("single-task-result-video");
video.src = "https://cliport.github.io/media/results_web/" +
task +
"-two_stream_full_clip_lingunet_lat_transporter-n" +
demo +
"-train/videos/" +
task +
"-0000" +
inst +
".mp4";
video.playbackRate = 2.0;
video.play();
}
function updateMultiVideo() {
var demo = document.getElementById("multi-menu-demos").value;
var task = document.getElementById("multi-menu-tasks").value;
var inst = document.getElementById("multi-menu-instances").value;
console.log("multi", demo, task, inst)
var video = document.getElementById("multi-task-result-video");
video.src = "https://cliport.github.io/media/results_web/" +
task +
"-two_stream_full_clip_lingunet_lat_transporter-n" +
demo +
"-train/videos/multi-language-conditioned-" +
task +
"-0000" +
inst +
".mp4";
video.playbackRate = 2.0;
video.play();
}
</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>
<nav class="navbar" role="navigation" aria-label="main navigation">
<div class="navbar-brand">
<a role="button" class="navbar-burger" aria-label="menu" aria-expanded="false">
<span aria-hidden="true"></span>
<span aria-hidden="true"></span>
<span aria-hidden="true"></span>
</a>
</div>
<div class="navbar-menu">
<div class="navbar-start" style="flex-grow: 1; justify-content: center;">
<a class="navbar-item" target="_blank" href="https://mohitshridhar.com">
<span class="icon">
<i class="fas fa-home"></i>
</span>
</a>
<div class="navbar-item has-dropdown is-hoverable">
<a class="navbar-link">
More Research
</a>
<div class="navbar-dropdown">
<a class="navbar-item" target="_blank" href="https://askforalfred.com/">
ALFRED
</a>
<a class="navbar-item" target="_blank" href="http://alfworld.github.io/">
ALFWorld
</a>
<a class="navbar-item" target="_blank" href="https://arxiv.org/pdf/1806.03831.pdf">
INGRESS
</a>
</div>
</div>
</div>
</div>
</nav>
<section class="hero">
<div class="hero-body">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column has-text-centered">
<h1 class="title is-1 publication-title">CLIPort: What and Where Pathways for Robotic Manipulation</h1>
<h3 class="title is-4 conference-authors"><a target="_blank" href="https://www.robot-learning.org/">CoRL 2021</a></h3>
<div class="is-size-5 publication-authors">
<span class="author-block">
<a target="_blank" href="https://mohitshridhar.com/">Mohit Shridhar</a><sup>1</sup>,</span>
<span class="author-block">
<a target="_blank" href="http://lucasmanuelli.com/">Lucas Manuelli</a><sup>2</sup>,</span>
<span class="author-block">
<a target="_blank" href="https://homes.cs.washington.edu/~fox/">Dieter Fox</a><sup>1, 2</sup>
</span>
</div>
<div class="is-size-5 publication-authors">
<span class="author-block"><sup>1</sup>University of Washington,</span>
<span class="author-block"><sup>2</sup>NVIDIA</span>
</div>
<div class="column has-text-centered">
<div class="publication-links">
<!-- PDF Link. -->
<span class="link-block">
<a target="_blank" href="https://arxiv.org/pdf/2109.12098.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>
<!-- <span class="link-block">
<a target="_blank" href="https://drive.google.com/file/d/1xzG5e1XF958HPuD_FZTiKROd9AQyd1fS/view?usp=sharing"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="ai ai-arxiv"></i>
</span>
<span>arXiv</span>
</a>
</span> -->
<!-- Video Link. -->
<span class="link-block">
<a target="_blank" href="https://youtu.be/UdzoagBgWTA"
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/cliport/cliport"
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">
<video id="teaser" autoplay muted loop height="100%">
<source src="https://cliport.github.io/media/videos/10sim_web_teaser.mp4"
type="video/mp4">
</video>
<h2 class="subtitle has-text-centered">
</br>
<span class="dcliport">CLIPort</span> is an end-to-end imitation-learning agent that can learn a single language-conditioned policy for various tabletop tasks.
</h2>
</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 item-steve">
<video poster="" id="steve" autoplay controls muted loop height="100%">
<source src="https://cliport.github.io/media/videos/1_folding.mp4"
type="video/mp4">
</video>
</div>
<div class="item item-fullbody">
<video poster="" id="fullbody" autoplay controls muted loop height="100%">
<source src="https://cliport.github.io/media/videos/4_chess.mp4"
type="video/mp4">
</video>
</div>
<div class="item item-shiba">
<video poster="" id="shiba" autoplay controls muted loop height="100%">
<source src="https://cliport.github.io/media/videos/3_packing.mp4"
type="video/mp4">
</video>
</div>
<div class="item item-blueshirt">
<video poster="" id="blueshirt" autoplay controls muted loop height="100%">
<source src="https://cliport.github.io/media/videos/6_sweeping.mp4"
type="video/mp4">
</video>
</div>
<div class="item item-shiba">
<video poster="" id="shiba" autoplay controls muted loop height="100%">
<source src="https://cliport.github.io/media/videos/9_cherry.mp4"
type="video/mp4">
</video>
</div>
<div class="item item-shiba">
<video poster="" id="shiba" autoplay controls muted loop height="100%">
<source src="https://cliport.github.io/media/videos/7_reading.mp4"
type="video/mp4">
</video>
</div>
<div class="item item-chair-tp">
<video poster="" id="chair-tp" autoplay controls muted loop height="100%">
<source src="https://cliport.github.io/media/videos/2_bowl.mp4"
type="video/mp4">
</video>
</div>
<div class="item item-blueshirt">
<video poster="" id="blueshirt" autoplay controls muted loop height="100%">
<source src="https://cliport.github.io/media/videos/8_rope.mp4"
type="video/mp4">
</video>
</div>
<div class="item item-blueshirt">
<video poster="" id="blueshirt" autoplay controls muted loop height="100%">
<source src="https://cliport.github.io/media/videos/5_stacking.mp4"
type="video/mp4">
</video>
</div>
</div>
</div>
</div>
</section>
<h2 class="subtitle has-text-centered">
</br>
We learn <b>one multi-task policy</b> for 9 real-world tasks including folding cloths, sweeping beans etc. with just <b>179</b> image-action training pairs.
</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>
How can we imbue robots with the ability to manipulate objects precisely but also to
reason about them in terms of abstract concepts?
</p>
<p>
Recent works in manipulation have shown that end-to-end networks
can learn dexterous skills that require precise spatial reasoning, but these methods
often fail to generalize to new goals or quickly learn transferable concepts across tasks. In
parallel, there has been great progress in learning generalizable semantic representations
for vision and language by training on large-scale internet data, however
these representations lack the spatial understanding necessary for fine-grained
manipulation. To this end, we propose a framework that combines the best of
both worlds: a two-stream architecture with semantic and spatial
pathways for vision-based manipulation. Specifically, we present <span class="dcliport">CLIPort</span>, a
language-conditioned imitation-learning agent that combines the broad semantic
understanding <i>(what)</i> of <a target=”_blank” href="https://openai.com/blog/clip/">CLIP</a> with the spatial precision
<i>(where)</i> of <a target=”_blank” href="https://transporternets.github.io/">TransporterNets</a>.
</p>
<p>
Our end-to-end framework is capable of solving a variety of language-specified tabletop
tasks from packing unseen objects to folding cloths, all <b>without any explicit representations</b>
of object poses, instance segmentations, memory, symbolic states, or
syntactic structures. Experiments in simulation and hardware show that our approach
is data-efficient and generalizes effectively to seen and unseen semantic
concepts. We even train <b>one multi-task policy</b> for 10 simulated and 9 real-world
tasks that shows better or comparable performance to single-task policies.
</p>
</div>
</div>
</div>
<!--/ Abstract. -->
</div>
<!-- Paper video. -->
</br>
</br>
<div class="columns is-centered has-text-centered">
<div class="column is-two-thirds">
<h2 class="title is-3">Video</h2>
<div class="publication-video">
<iframe src="https://www.youtube.com/embed/UdzoagBgWTA?rel=0&showinfo=0"
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="dcliport">CLIPort</span></h2>
<!-- Interpolating. -->
<h3 class="title is-4">Two-Stream Architecture</h3>
<div class="content has-text-justified">
<p>
Broadly inspired by (or vaguely analogous to) the <a target=”_blank” href="https://en.wikipedia.org/wiki/Two-streams_hypothesis">two-stream hypothesis in cognitive psychology</a>, we present a two-stream architecture
for vision-based manipulation with semantic and spatial pathways. The semantic stream uses a pre-trained CLIP model
to encode RGB and language-goal input. Since CLIP is trained with large amounts of image-caption pairs from the internet,
it acts as a powerful semantic prior for <a target="_blank" href="https://distill.pub/2021/multimodal-neurons/">grounding visual concepts</a> like colors, shapes, parts, texts, and object categories.
The spatial stream is a tabula rasa fully-convolutional network that encodes RGB-D input.
</p>
</div>
<img src="https://cliport.github.io/media/images/two_stream_architecture.png" class="interpolation-image"
alt="Interpolate start reference image." />
<br/>
<br/>
<b>Paradigm 1:</b> Unlike existing object detectors, CLIP is not limited to a predefined set of object classes. And unlike other vision-language models, it's not restricted by a top-down pipeline that detects objects with bounding boxes or instance segmentations. This allows us to forgo the traditional paradigm of training explicit detectors for cloths, pliers, chessboard squares, cherry stems, and other arbitrary things.
<br/>
<br/>
<br/>
<!--/ Interpolating. -->
<!-- Re-rendering. -->
<h3 class="title is-4">TransporterNets</h3>
<div class="content has-text-justified">
<p>
We use this two-stream architecture in all three networks of <a target=”_blank” href="https://transporternets.github.io/">TransporterNets</a>
to predict pick and place affordances at each timestep. TransporterNets first attends to a local region to decide where to pick,
then computes a placement location by finding the best match for the picked region through
cross-correlation of deep visual features. This structure serves as a powerful inductive bias for learning <a target="_blank" href="https://fabianfuchsml.github.io/equivariance1of2/">roto-translationally equivariant</a> representations in tabletop environments.
</p>
</div>
<div class="content has-text-centered">
<video id="transporter-gif"
controls
muted
autoplay
loop
width="40%">
<source src="https://transporternets.github.io/images/animation.mp4"
type="video/mp4">
</video>
<p>
Credit: <a href="https://transporternets.github.io/">Zeng et. al (Google)</a>
</p>
</div>
<br/>
<b>Paradigm 2:</b> TransporterNets takes an <a target="_blank" href="https://en.wikipedia.org/wiki/Ecological_psychology">action-centric approach</a> to perception where the objective is to <i>detect actions</i> rather than <i>detect objects</i> and then learn a policy. Keeping the action-space grounded in the perceptual input allows us to exploit geometric symmetries for efficient representation learning.
When combined with CLIP's pre-trained representations, this enables the learning of reusable manipulation skills without any "objectness" assumptions.
<br/>
<br/>
<br/>
<!--/ Re-rendering. -->
<h2 class="title is-3">Results</h2>
<div class="columns">
<div class="column has-text-centered">
<h3 class="title is-5">Single-Task Models</h3>
Trained with
<div class="select is-small">
<select id="single-menu-demos" onchange="updateSingleVideo()">
<option value="1">1</option>
<option value="10">10</option>
<option value="100">100</option>
<option value="1000" selected="selected">1000</option>
</select>
</div>
demos, evaluated on
<div class="select is-small">
<select id="single-menu-tasks" onchange="updateSingleVideo()">
<option value="align-rope">align-rope</option>
<option value="assembling-kits-seq-seen-colors">assembling-kits-seq-seen-colors</option>
<option value="assembling-kits-seq-unseen-colors">assembling-kits-seq-unseen-colors</option>
<option value="packing-boxes-pairs-seen-colors">packing-boxes-pairs-seen-colors</option>
<option value="packing-boxes-pairs-unseen-colors">packing-boxes-pairs-unseen-colors</option>
<option value="packing-seen-google-objects-seq" selected="selected">packing-seen-google-objects-seq</option>
<option value="packing-unseen-google-objects-seq">packing-unseen-google-objects-seq</option>
<option value="packing-seen-google-objects-group">packing-seen-google-objects-group</option>
<option value="packing-unseen-google-objects-group">packing-unseen-google-objects-group</option>
<option value="packing-shapes">packing-shapes</option>
<option value="put-block-in-bowl-seen-colors">put-block-in-bowl-seen-colors</option>
<option value="put-block-in-bowl-unseen-colors">put-block-in-bowl-unseen-colors</option>
<option value="separating-piles-seen-colors">separating-piles-seen-colors</option>
<option value="separating-piles-unseen-colors">separating-piles-unseen-colors</option>
<option value="stack-block-pyramid-seq-seen-colors">stack-block-pyramid-seq-seen-colors</option>
<option value="stack-block-pyramid-seq-unseen-colors">stack-block-pyramid-seq-unseen-colors</option>
<option value="towers-of-hanoi-seq-seen-colors">towers-of-hanoi-seq-seen-colors</option>
<option value="towers-of-hanoi-seq-unseen-colors">towers-of-hanoi-seq-unseen-colors</option>
</select>
</div>
instance
<div class="select is-small">
<select id="single-menu-instances" onchange="updateSingleVideo()">
<option value="01">01</option>
<option value="02">02</option>
<option value="03">03</option>
<option value="04">04</option>
<option value="05" selected="selected">05</option>
</select>
</div>
<br/>
<br/>
<video id="single-task-result-video"
controls
muted
autoplay
loop
width="100%">
<source src="https://cliport.github.io/media/results_web/packing-seen-google-objects-seq-two_stream_full_clip_lingunet_lat_transporter-n1000-train/videos/packing-seen-google-objects-seq-000005.mp4"
type="video/mp4">
</video>
</div>
<div class="column has-text-centered">
<h3 class="title is-5">One Multi-Task Model</h3>
Trained with
<div class="select is-small">
<select id="multi-menu-demos" onchange="updateMultiVideo()">
<option value="1">1 T</option>
<option value="10">10 T</option>
<option value="100">100 T</option>
<option value="1000" selected="selected">1000 T</option>
</select>
</div>
demos, evaluated on
<div class="select is-small">
<select id="multi-menu-tasks" onchange="updateMultiVideo()">
<option value="align-rope">align-rope</option>
<option value="assembling-kits-seq-seen-colors">assembling-kits-seq-seen-colors</option>
<option value="assembling-kits-seq-unseen-colors">assembling-kits-seq-unseen-colors</option>
<option value="packing-boxes-pairs-seen-colors" selected="selected">packing-boxes-pairs-seen-colors</option>
<option value="packing-boxes-pairs-unseen-colors">packing-boxes-pairs-unseen-colors</option>
<option value="packing-seen-google-objects-seq">packing-seen-google-objects-seq</option>
<option value="packing-unseen-google-objects-seq">packing-unseen-google-objects-seq</option>
<option value="packing-seen-google-objects-group">packing-seen-google-objects-group</option>
<option value="packing-unseen-google-objects-group">packing-unseen-google-objects-group</option>
<option value="packing-shapes">packing-shapes</option>
<option value="put-block-in-bowl-seen-colors">put-block-in-bowl-seen-colors</option>
<option value="put-block-in-bowl-unseen-colors">put-block-in-bowl-unseen-colors</option>
<option value="separating-piles-seen-colors">separating-piles-seen-colors</option>
<option value="separating-piles-unseen-colors">separating-piles-unseen-colors</option>
<option value="stack-block-pyramid-seq-seen-colors">stack-block-pyramid-seq-seen-colors</option>
<option value="stack-block-pyramid-seq-unseen-colors">stack-block-pyramid-seq-unseen-colors</option>
<option value="towers-of-hanoi-seq-seen-colors">towers-of-hanoi-seq-seen-colors</option>
<option value="towers-of-hanoi-seq-unseen-colors">towers-of-hanoi-seq-unseen-colors</option>
</select>
</div>
instance
<div class="select is-small">
<select id="multi-menu-instances" onchange="updateMultiVideo()">
<option value="01">01</option>
<option value="02">02</option>
<option value="03">03</option>
<option value="04" selected="selected">04</option>
<option value="05">05</option>
</select>
</div>
</br>
</br>
<video id="multi-task-result-video"
controls
muted
autoplay
loop
width="100%">
<source src="https://cliport.github.io/media/results_web/packing-boxes-pairs-seen-colors-two_stream_full_clip_lingunet_lat_transporter-n1000-train/videos/multi-language-conditioned-packing-boxes-pairs-seen-colors-000004.mp4"
type="video/mp4">
</video>
</div>
</div>
</br>
<h3 class="title is-4">Affordance Predictions</h3>
<div class="content has-text-justified">
<p>
Examples of pick and place affordance predictions from multi-task <span class="dcliport">CLIPort</span> models:
</p>
</div>
<br/>
<img src="https://cliport.github.io/media/images/affordances.png" class="interpolation-image"
alt="Interpolate start reference image."/>
<br/>
<br/>
<img src="https://cliport.github.io/media/images/affordance2.png" class="interpolation-image"
alt="Interpolate start reference image."/>
</div>
</div>
</div>
</section>
<section class="section" id="BibTeX">
<div class="container is-max-widescreen content">
<h2 class="title">BibTeX</h2>
<pre><code>@inproceedings{shridhar2021cliport,
title = {CLIPort: What and Where Pathways for Robotic Manipulation},
author = {Shridhar, Mohit and Manuelli, Lucas and Fox, Dieter},
booktitle = {Proceedings of the 5th Conference on Robot Learning (CoRL)},
year = {2021},
}</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> made by the amazing <a href="https://keunhong.com/">Keunhong Park</a>.
</p>
</div>
</div>
</div>
</div>
</footer>
</body>
</html>