-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathrl_mpc.html
220 lines (203 loc) · 11.1 KB
/
rl_mpc.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
<!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>RL MPC Project</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" />
<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">RL MPC Project</a>
</header>
<!-- Nav -->
<nav id="nav">
<ul class="links">
<li><a href="index.html">Projects</a></li>
<li class="active"><a href="about.html">RL MPC Project</a></li>
</ul>
<ul class="icons">
<!-- <li><a href="#" class="icon brands fa-twitter"><span class="label">Twitter</span></a></li>
<li><a href="#" class="icon brands fa-facebook-f"><span class="label">Facebook</span></a></li>
<li><a href="#" class="icon brands fa-instagram"><span class="label">Instagram</span></a></li> -->
<li><a href="#" class="icon brands fa-github"><span class="label">GitHub</span></a></li>
</ul>
</nav>
<!-- Main -->
<div id="main">
<!-- Post -->
<section class="post">
<header class="major">
<h1>Reinforcement Learning Aided Model Predictive Controller for
Autonomous Vehicle Lateral Control</h1>
<p>
A nonlinear model predictive controller (NMPC)
coupled with a reinforcement learning (RL) model that
can be applied to lateral control tasks for autonomous vehicles.
For a more in-depth discussion of the project, please find the paper <a href="file/RL_MPC_paper.pdf" target="_blank">here</a>.
</p>
</header>
<header class="image fit">
<img src="images/isuzu_track.png" alt="" />
</header>
<header class="image fit">
<img src="images/lincoln_mkz.png" alt="" />
</header>
<p>
we propose to use a RL model to dynamically select the weights of the NMPC
objective function while performing real-time lateral control of the
autonomous vehicle (we call this RL-NMPC). The RL weight-search
model is trained in a simulator using only one reference path, and is
validated first in a simulation environment and then on a real Lincoln
MKZ vehicle; the RL-NMPC achieved considerably better
performance in lateral tracking during simulation and on-board tests.
</p>
<h3>Training Pipeline</h3>
<header class="image fit">
<img src="images/training_arch.png" alt="" />
</header>
<p>
The training cycle for the RL weight search module. The output of the
RL are a set of matrices (or scalars) that is used in the cost function of the MPC;
the solution of the MPC (steering angle ut) is then executed in the simulator;
and the simulator returns the observations of the environment and calculated
reward r to the RL module which is updated using the information feedback
from the simulator.
</p>
<header class="image fit">
<img src="images/multproc_figure.png" alt="" />
</header>
<p>
To boost the training efficiency and the
utilization of Graphic Processing Unit (GPU), multiple independent
simulations are run in parallel. Multi-threaded
training can perform updates asynchronously, by using one global
network and multiple local networks. During training, each agent is
controlled by one local network to collect experience tuples in its own
independent simulation; and the global network can be updated as
soon as one of the local agent completes experience collection without
the need to wait for other agents to complete.
</p>
<div class="video">
<video width="1280" height="720" autoplay muted loop>
<source src="images/async_demo.mp4" type="video/mp4">
</video>
</div>
<p>
A simple demonstration for the multi-worker RL training pipeline using OpenAI Gym environment and asynchronous version of PPO.
</p>
<h3>Simulated Results</h3>
<p>The tests
are run on four different maps, all from recorded waypoints of test
tracks and public roads. For simulation tests we tested the
performance of all three RL algorithms: DDPG, TD3, and PPO. </p>
<header class="image fit">
<img src="images/isuzu_test.png" alt="" />
<p>For lateral error tracking, all three RL models outperform the baseline parameters.</p>
</header>
<header class="image fit">
<img src="images/wx_plot.png" alt="" />
</header>
<header class="image fit">
<img src="images/wu_plot.png" alt="" />
</header>
<header class="image fit">
<img src="images/wdu_plot.png" alt="" />
</header>
<p>The above three plots show how the MPC weight matrices change during simulation tests at different
road segments.
</p>
<h3>On-board Deployment</h3>
<p>
The on-board testing of the weight selection
scheme is done on a real Lincoln MKZ vehicle. The vehicle is
modified so that its low level controller can communicate with the
host laptop via Robot Operating System (ROS). The RL model is run
on the host laptop along with the MPC, and then the computed
steering angle will be sent to the low level controller on the vehicle in
the form of ROS message; similarly, the states of the vehicle are sent
to the host laptop via ROS messages. The distance from the front
and rear axles to the vehicle’s center of gravity are 1.2m and 1.65m,
respectively; additionally, the vehicle has a full drive-by-wire system
that can communicate with the host computer via ROS, a Polynav
2000P GNSS-inertial system, a Calmcar front view camera, and a
Dspace Autera computing unit. The on-board test is carried
out on the test track in Isuzu Technical Center of America facility
</p>
<header class="image fit">
<img src="images/onboard_laterr.png" alt="" />
</header>
</section>
</div>
<!-- Footer -->
<footer id="footer">
<section>
<form method="post" action="#">
<div class="fields">
<div class="field">
<label for="name">Name</label>
<input type="text" name="name" id="name" />
</div>
<div class="field">
<label for="email">Email</label>
<input type="text" name="email" id="email" />
</div>
<div class="field">
<label for="message">Message</label>
<textarea name="message" id="message" rows="3"></textarea>
</div>
</div>
<ul class="actions">
<li><input type="submit" value="Send Message" /></li>
</ul>
</form>
</section>
<section class="split contact">
<section class="alt">
<h3>Address</h3>
<p>2145 Sheridan Rd<br />
Evanston, IL 60208</p>
</section>
<section>
<h3>Phone</h3>
<p><a href="#">(734) 510-1501</a></p>
</section>
<section>
<h3>Email</h3>
<p><a href="#">muyejia2023@u.northwestern.edu</a></p>
</section>
<section>
<h3>Social</h3>
<ul class="icons alt">
<!-- <li><a href="#" class="icon brands alt fa-twitter"><span class="label">Twitter</span></a></li>
<li><a href="#" class="icon brands alt fa-facebook-f"><span class="label">Facebook</span></a></li>
<li><a href="#" class="icon brands alt fa-instagram"><span class="label">Instagram</span></a></li> -->
<li><a href="https://github.com/muye1202" class="icon brands alt fa-github"><span class="label">GitHub</span></a></li>
</ul>
</section>
</section>
</footer>
<!-- Copyright -->
<div id="copyright">
<ul><li>© Untitled</li><li>Design: <a href="https://html5up.net">HTML5 UP</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>