-
Notifications
You must be signed in to change notification settings - Fork 0
/
3R.html
269 lines (241 loc) · 11.5 KB
/
3R.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
<!DOCTYPE HTML>
<!--
Massively by HTML5 UP
html5up.net | @ajlkn
Free for personal and commercial use under the CCA 3.0 license (html5up.net/license)
-->
<html>
<head>
<title>3R|Purdue Digital Twin Lab</title>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1, user-scalable=no" />
<link rel="stylesheet" href="assets/css/main.css" />
<link rel="icon" href="images/icon_36_36.png"/>
<noscript>
<link rel="stylesheet" href="assets/css/noscript.css" />
</noscript>
</head>
<body class="is-preload">
<!-- Wrapper -->
<div id="wrapper">
<!-- Header -->
<header id="header">
<a href="index.html" class="logo">LLM4AD</a>
</header>
<!-- Nav -->
<nav id="nav">
<ul class="links">
<li><a href="index.html">Overview</a></li>
<li><a href="talk2drive.html">Talk2Drive</a></li>
<!-- <li><a href="generic.html">Demos</a></li> -->
<li class="active"><a href="3R.html">3R</a></li>
<li><a href="survey.html">Survey</a></li>
<li><a href="lampilot.html">Lampilot</a></li>
</ul>
</nav>
<!-- Main -->
<div id="main">
<!-- Post -->
<section class="post">
<center>
<h2>Receive, Reason, and React: Drive as You Say with Large Language Models in Autonomous
Vehicles<br/></h2>
<h4>Can Cui*, Yunsheng Ma*, Xu Cao, Wenqian Ye and Ziran Wang<br/></h4>
</center>
<ul class="actions special">
<li><a href="https://arxiv.org/abs/2310.08034" class="button large">Paper</a></li>
</ul>
<!-- <header class="major">
<h1>3R</h1>
<h2>Receive, Reason, and React: Drive as You Say with Large Language Models in Autonomous Vehicles<br /></h2>
</header>
<p>Talk2Drive Framework is the first a successful end-to-end implementation of an LLM-based autonomous driving system
in a real-world vehicle. This approach is designed to overcome several key challenges faced by traditional autonomous driving systems.<br /></p>
</header> -->
<header>
<p>We present a novel approach where LLMs serve as the decision-making ``brain" within autonomous
vehicles.
Complementing this, various tools within the autonomous vehicle ecosystem, including the
perception module,
localization module, and in-cabin monitor, function as the vehicle's sensory ``eyes." This
configuration enables
LLMs to overcome the inherent limitation of not directly accessing real-time environmental
information. Additionally,
the vehicle's controller function as its ``hands," executing instructions derived from the LLM's
decision-making
process. Through receiving environmental information and drivers' commands, reasoning based on
this information
and human interaction, and finally making decisions, we make the autonomous driving experience
that is not just
technologically superior but also deeply human-centric by LLMs.
<a href="#" class="image main"><img src="images/framework_update.png" alt="" /></a>
<!-- <ul class="actions special">
<li><a href="https://arxiv.org/abs/2310.08034" class="button large">Paper</a></li>
</ul> -->
<hr />
<h2>3R Cases Study</h2>
In our closed-loop driving experiment on HighwayEnv using GPT-4, we try to assess the LLMs'
interpretation,
chain-of-thought, and environmental interaction abilities. Our experimental design includes two
distinct highway
scenarios. In the first one, the environment was safe for overtaking the vehicle ahead; in contrast,
the second
scenario presented conditions where overtaking was considered unsafe and not suitable. The emphasis
was on
observing the LLMs' reactions and decision-making in various conditions. For each scenario, we
employ two
distinct training methods. One method utilized standard prompting for training, and the other used
the
chain-of-thought prompting approach. Through this design, our objective was to discern and highlight
the
comparative advantages of using chain-of-thought prompting over standard prompting techniques. The
whole
working process for safe and unsafe scenarios is shown in below figure respectively.
When prompted using the chain-of-thoughts method, the LLMs first generate comprehensive and reasoned
thoughts before suggesting a driving plan. In contrast, with the standard prompting training method,
the LLMs directly present the plan. The plans derived from these two methods have distinct
differences.
<div class="image main"><img src="images/highway.png" alt="" /></div>
<hr />
<!-- -->
<section class="post">
<h2>Highway Demos</h2>
<header class="major">
<!-- <span class="date">April 25, 2017</span> -->
</header>
<!-- <div class="image main"><img src="images/pic01.jpg" alt="" /></div> -->
<div class="container">
<div class="row gtr-200">
<div class="col-9 col-12-medium">
<h3>Highway Safe Passing Scenario Through Standard Prompting</h3>
<iframe width="900" height="500" src="https://www.youtube.com/embed/wPVXUAyhvAM"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen></iframe>
</div>
</section>
<section class="post">
<header class="major">
</header>
<!-- <div class="image main"><img src="images/pic01.jpg" alt="" /></div> -->
<div class="container">
<div class="row gtr-200">
<div class="col-9 col-12-medium">
<h3>Highway Safe Passing Scenario Through Chain-of-thought Prompting</h3>
<iframe width="900" height="500" src="https://www.youtube.com/embed/cPFk1nfRcvM"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen></iframe>
</div>
</section>
<section class="post">
<header class="major">
<!-- <span class="date">April 25, 2017</span> -->
</header>
<!-- <div class="image main"><img src="images/pic01.jpg" alt="" /></div> -->
<div class="container">
<div class="row gtr-200">
<div class="col-9 col-12-medium">
<h3>Highway Unsafe Passing Scenario Through Standard Prompting</h3>
<iframe width="900" height="500" src="https://www.youtube.com/embed/SPenkKMk1ik"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen></iframe>
</div>
</section>
<section class="post">
<header class="major">
</header>
<!-- <div class="image main"><img src="images/pic01.jpg" alt="" /></div> -->
<div class="container">
<div class="row gtr-200">
<div class="col-9 col-12-medium">
<h3>Highway Unsafe Passing Scenario Through Chain-of-thought Prompting</h3>
<iframe width="900" height="500" src="https://www.youtube.com/embed/hSy9ePeAqXs"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen></iframe>
</div>
</section>
</section>
<section class="post">
<h2>Merging Demos</h2>
<header class="major">
<!-- <span class="date">April 25, 2017</span> -->
</header>
<!-- <div class="image main"><img src="images/pic01.jpg" alt="" /></div> -->
<div class="container">
<div class="row gtr-200">
<div class="col-9 col-12-medium">
<h3>Merging Scenario with No Extra Command</h3>
<iframe width="900" height="500" src="https://www.youtube.com/embed/-MIP1jHS__8"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen></iframe>
</div>
<section class="post">
<header class="major">
</header>
<!-- <div class="image main"><img src="images/pic01.jpg" alt="" /></div> -->
<div class="container">
<div class="row gtr-200">
<div class="col-9 col-12-medium">
<h3>Merging Scenario with Command "Drive Aggressively"</h3>
<iframe width="900" height="500" src="https://www.youtube.com/embed/26e_GgLDCKs"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen></iframe>
</div>
<section class="post">
<header class="major">
<!-- <span class="date">April 25, 2017</span> -->
</header>
<!-- <div class="image main"><img src="images/pic01.jpg" alt="" /></div> -->
<div class="container">
<div class="row gtr-200">
<div class="col-9 col-12-medium">
<h3>Merging Scenario with the Command "Drive Conservatively"</h3>
<iframe width="900" height="500"
src="https://www.youtube.com/embed/eO5uiCXPdxs"
allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share"
allowfullscreen></iframe>
</div>
</section>
</section>
</div>
</div>
<hr>
<div style="overflow-x: auto;">
<table style="width:100%">
<h2> Contributors</h2>
<tr>
<td style="text-align:center"><img src="https://purduedigitaltwin.github.io/assets/images/people/can.jpg" height="160"></td>
<td style="text-align:center"><img src="https://purduedigitaltwin.github.io/assets/images/people/yunsheng.jpg" height="160"></td>
<td style="text-align:center"><img src="https://raw.githubusercontent.com/LLVM-AD/llvm-ad.github.io/main/assets/img/Xu_Cao.jpg" height="160"></td>
<td style="text-align:center"><img src="https://wenqian-ye.github.io/images/selfie.jpeg" height="160"></td>
<td style="text-align:center"><img src="https://purduedigitaltwin.github.io/assets/images/people/prof_wang.jpg" height="160"></td>
</tr>
<tr>
<td style="text-align:center"><a href="https://cancui19.github.io/">Can Cui</a> <br> Purdue University</td>
<td style="text-align:center"><a href="https://maysonma.github.io/">Yunsheng Ma</a> <br>Purdue University</td>
<td style="text-align:center"><a href="https://www.linkedin.com/in/irohxu/">Xu Cao</a> <br>PediaMed AI & UIUC</td>
<td style="text-align:center"><a href="https://wenqian-ye.github.io/">Wenqian Ye</a> <br> PediaMed AI & UVA </td>
<td style="text-align:center"><a href="https://ziranw.github.io">Ziran Wang</a> <br>Purdue University</td>
</tr>
<!-- <td style="text-align:center"><a href="">Tong Zhou</a> <br> Tencent</td> -->
</tr>
</table>
</div>
</div>
<!-- Copyright -->
<div id="copyright">
<ul>
<li>© Purdue Digital Twin Lab</li>
<li>Lab Website: <a href="https://purduedigitaltwin.github.io">purduedigitaltwin.github.io</a></li>
</ul>
</div>
</div>
<!-- Scripts -->
<script src="assets/js/jquery.min.js"></script>
<script src="assets/js/jquery.scrollex.min.js"></script>
<script src="assets/js/jquery.scrolly.min.js"></script>
<script src="assets/js/browser.min.js"></script>
<script src="assets/js/breakpoints.min.js"></script>
<script src="assets/js/util.js"></script>
<script src="assets/js/main.js"></script>
</body>
</html>