-
Notifications
You must be signed in to change notification settings - Fork 0
/
index.html
270 lines (247 loc) · 14.5 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
<html>
<head>
<!-- Google tag (gtag.js) -->
<script async src="https://www.googletagmanager.com/gtag/js?id=G-FYQNRK8LHK"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag() { dataLayer.push(arguments); }
gtag('js', new Date());
gtag('config', 'G-FYQNRK8LHK');
</script>
<meta charset="utf-8" />
<title>MANIP: Integrating Interactive Perception into Long-Horizon Robot Manipulation Systems</title>
<link href="https://fonts.googleapis.com" rel="preconnect" />
<link href="https://fonts.gstatic.com" rel="preconnect" crossorigin="anonymous" />
<script src="https://ajax.googleapis.com/ajax/libs/webfont/1.6.26/webfont.js" type="text/javascript"></script>
<script
type="text/javascript">WebFont.load({ google: { families: ["Lato:100,100italic,300,300italic,400,400italic,700,700italic,900,900italic", "Montserrat:100,100italic,200,200italic,300,300italic,400,400italic,500,500italic,600,600italic,700,700italic,800,800italic,900,900italic", "Ubuntu:300,300italic,400,400italic,500,500italic,700,700italic", "Open Sans:300,300italic,400,400italic,600,600italic,700,700italic,800,800italic", "Changa One:400,400italic", "Varela Round:400", "Bungee Shade:regular", "Roboto:300,regular,500", "Bungee Outline:regular"] } });</script>
<!--[if lt IE 9]><script src="https://cdnjs.cloudflare.com/ajax/libs/html5shiv/3.7.3/html5shiv.min.js" type="text/javascript"></script><![endif]-->
<script src="https://ajax.googleapis.com/ajax/libs/jquery/1/jquery.min.js"></script>
<script src="script.js" type="text/javascript"></script>
<link href="style.css" rel="stylesheet" type="text/css" />
<link href="data/favicon.png" rel="shortcut icon" type="image/x-icon" />
<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/4.7.0/css/font-awesome.min.css">
</head>
<body>
<div class="section">
<div class="container">
<div class="title-row">
<h1 class="title">MANIP</h1>
<h1 class="subheader"> Integrating Interactive Perception into Long-Horizon Robot Manipulation Systems </h1>
</div>
<div class="base-row author-row">
<div class="base-col author-col">
<a href="" target="_blank" class="author-text">
Justin Yu
<span class="text-star">* 1</span>
<!-- <span class="text-star">*</span> -->
</a>
</div>
<div class="base-col author-col">
<a href="https://uynitsuj.github.io/" target="_blank" class="author-text">
Tara Sadjadpour
<span class="text-star">* 1</span>
<!-- <span class="text-star">*</span> -->
</a>
</div>
<div class="base-col author-col">
<a href="" target="_blank" class="author-text">
Abby O’Neill
<span class="text-star">1</span>
</a>
</div>
<div class="base-col author-col">
<a href="" target="_blank" class="author-text">
Mehdi Khfifi
<span class="text-star">1</span>
</a>
</div>
</div>
<div class="base-row author-row">
<div class="base-col author-col">
<a href="" target="_blank" class="author-text">
Lawrence Chen
<span class="text-star">1</span>
</a>
</div>
<div class="base-col author-col">
<a href="" target="_blank" class="author-text">
Richard Cheng
<span class="text-star">2</span>
</a>
</div>
<div class="base-col author-col">
<a href="" target="_blank" class="author-text">
Muhammad Zubair
<span class="text-star">2</span>
</a>
</div>
<div class="base-col author-col">
<a href="" target="_blank" class="author-text">
Ashwin Balakrishna
<span class="text-star">2</span>
</a>
</div>
<div class="base-col author-col">
<a href="" target="_blank" class="author-text">
Thomas Kollar
<span class="text-star">2</span>
</a>
</div>
<div class="base-col author-col">
<a href="https://goldberg.berkeley.edu/" target="_blank" class="author-text">
Ken Goldberg
<span class="text-star">1</span>
</a>
</div>
</div>
<div>
<!-- <h1 id="uc-berkeley">UC Berkeley</h1> -->
<h1 id="uc-berkeley"><span class="text-star">1</span> The AUTOLab at UC Berkeley</h1>
<h1 id="uc-berkeley"><span class="text-star">2</span> The Toyota Research Institute</h1>
<!-- <br /> -->
<div id="equal_contrib" style="margin-top: 7px;">
<span class="text-star">*</span>Denotes Equal Contribution
</div>
</div>
<div class="title-row">
<h2 class="subheader">Accepted to IROS 2024</h1>
</div>
<div class="link-labels base-row">
<!-- TODO: Update arxiv link -->
<div class="base-col icon-col"><a href="https://drive.google.com/file/d/1UWYnjzxLKpspPTidDE5WcTN525MWbl3s/view" target="_blank"
class="link-block"><img
src="https://uploads-ssl.webflow.com/51e0d73d83d06baa7a00000f/5cab99df4998decfbf9e218e_paper-01.png"
alt="paper"
sizes="(max-width: 479px) 12vw, (max-width: 767px) 7vw, (max-width: 991px) 41.8515625px, 56.6953125px"
srcset="https://uploads-ssl.webflow.com/51e0d73d83d06baa7a00000f/5cab99df4998decfbf9e218e_paper-01-p-500.png 500w, https://uploads-ssl.webflow.com/51e0d73d83d06baa7a00000f/5cab99df4998decfbf9e218e_paper-01.png 672w"
class="icon-img" /></a></div>
<!-- TODO: Update code link -->
<div class="base-col icon-col"><a href="https://github.com/BerkeleyAutomation/multicable-decluttering" class="link-block" target= "blank"><img
src="https://uploads-ssl.webflow.com/51e0d73d83d06baa7a00000f/5cae3b53b42ebb3dd4175a82_68747470733a2f2f7777772e69636f6e66696e6465722e636f6d2f646174612f69636f6e732f6f637469636f6e732f313032342f6d61726b2d6769746875622d3235362e706e67.png"
alt="paper" style="height: 60" class="icon-img github-img-icon" /></a></div>
</div>
<div class="link-labels base-row">
<div class="base-col icon-col">
<strong class="link-labels-text">Paper</strong>
</div>
<div class="base-col icon-col">
<strong class="link-labels-text"> Code </strong>
</div>
</div>
<div class="link-labels base-row">
<!-- TODO: Update arxiv link -->
<div class="base-col icon-col"><a href="MANIP_Diagram.pptx" download target="_blank"
class="link-block"><img
src="webimgs/powerpoint.png"
alt="pptx"
width="40px"
class="icon-img" /></a></div>
<!-- TODO: Update code link -->
<div class="base-col icon-col"><a href="MANIP_Diagram.key" download target="_blank"
class="link-block"><img
src="webimgs/key.png"
alt="keynote"
width="40px"
class="icon-img" /></a></div>
<div class="base-col icon-col"><a href="https://docs.google.com/drawings/d/1EhBTvXPAdX1KEG3J2alPVtQfJk1X1HmNQeNY15VCUPQ/edit?usp=sharing" target="_blank"
class="link-block"><img
src="webimgs/googledraw.png"
alt="google draw"
width="30px"
class="icon-img" /></a></div>
</div>
<br>
<div class="link-labels base-row">
<div class="base-col icon-col">
<strong class="link-labels-text">Generic MANIP Diagram (Powerpoint)</strong>
</div>
<div class="base-col icon-col">
<strong class="link-labels-text">Generic MANIP Diagram (Keynote)</strong>
</div>
<div class="base-col icon-col">
<strong class="link-labels-text">Generic MANIP Diagram (Google Draw)</strong>
</div>
</div>
<br>
<video controls width="800px">
<source src="MANIP_video.mp4" type="video/mp4" />
</video>
<br>
<figure>
<img src="MANIP_Diagram.png" style="width: 100%;" alt="Manip Diagram"/>
<figcaption>MANIP Diagram.</figcaption>
</figure>
<div class="base-row add-top-padding">
<h1 id="abstract">Overview</h1>
<p class="paragraph">
<!--   -->
<!-- Interactive perception, where a robot acts to change the environment to facilitate perception,
can be useful for robot manipulation, especially for long-horizon tasks, such as folding clothes,
routing cables, untangling knots, and surgical suturing. However, it can often be unclear how to
integrate interactive perception primitives into manipulation policies. Long-horizon robot manipulation
tasks pose significant challenges due to a variety of factors, including partial observability, compounding errors,
and the increasing complexity of the state-action space as the task length increases. We propose MANIP, a
modular architecture that can help integrate combinations of interactive perception and task completion
policies and primitives. MANIP facilitates switching between minimizing uncertainty with interactive perception
and maximizing reward with task policies. We show that MANIP can be mapped to and potentially
improve 3 prior robot manipulation systems. Experiments with 120 physical robot trials across 4 tiers of
difficulty suggest that the resulting system can improve cable tracing by up to 88%. -->
Interactive perception, where a robot acts to change the
environment to facilitate perception, can be useful for robot
manipulation, especially for long-horizon tasks, such as folding
clothes, routing cables, untangling knots, and surgical
suturing. However, it can often be unclear how to integrate
interactive perception primitives into manipulation
policies. Long-horizon robot manipulation tasks pose significant
challenges due to a variety of factors, including partial
observability, compounding errors, and the increasing complexity of
the state-action space as the task length increases. We propose MANIP,
a modular architecture that can help integrate combinations of
interactive perception and task completion policies and
primitives. MANIP facilitates switching between minimizing uncertainty
with interactive perception and maximizing reward with task
policies. We show that MANIP can be mapped to and potentially improve
3 prior robot manipulation systems. We then apply MANIP to extend our
previously published cable tracing system. We show that the resulting
system improves cable tracing by up to 88% across 4 tiers of
difficulty and 120 physical robot trials for adversarial settings with
cables.
</p>
</div>
<h1 id="abstract">Example Rollout</h1>
<div class="gallery-container">
<img src="interactive_display/step1.png" alt="Gallery Image" id="galleryImage" style="height: 576px;" class="gallery-img">
<div>
<button id="prevBtn" class="disabled">Previous</button>
<button id="nextBtn">Next</button>
</div>
<caption>Green arrows show where IP move was executed by the YuMi robot</caption>
</div>
<script>
let currentImageIndex = 1;
const imageCount = 11; // Update this to reflect the actual number of images
function updateButtons() {
document.getElementById('prevBtn').classList.toggle('disabled', currentImageIndex === 1);
document.getElementById('nextBtn').classList.toggle('disabled', currentImageIndex === imageCount);
}
document.getElementById('prevBtn').addEventListener('click', function() {
if (currentImageIndex > 1) {
currentImageIndex--;
updateImage();
}
});
document.getElementById('nextBtn').addEventListener('click', function() {
if (currentImageIndex < imageCount) {
currentImageIndex++;
updateImage();
}
});
function updateImage() {
const imagePath = `interactive_display/step${currentImageIndex}.png`;
document.getElementById('galleryImage').src = imagePath;
updateButtons();
}
// Initial button state update
updateButtons();
</script>