-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathresearch.html
171 lines (145 loc) · 12.6 KB
/
research.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
<!DOCTYPE HTML>
<!--
Editorial by HTML5 UP
html5up.net | @ajlkn
Free for personal and commercial use under the CCA 3.0 license (html5up.net/license)
-->
<html>
<head>
<title>Somil Bansal</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" />
<script>
(function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){
(i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o),
m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m)
})(window,document,'script','https://www.google-analytics.com/analytics.js','ga');
ga('create', 'UA-78683315-1', 'auto');
ga('send', 'pageview');
</script>
</head>
<body class="is-preload">
<!-- Wrapper -->
<div id="wrapper">
<!-- Main -->
<div id="main">
<div class="inner">
<!-- Header -->
<header id="header">
<a href="index.html" class="logo"><strong>Somil Bansal</strong></a>
<ul class="icons">
<li><a href="https://github.com/smlbansal" class="icon brands fa-github"><span class="label">Medium</span></a></li>
</ul>
</header>
<!-- Content -->
<section>
<header class="main">
<h1>Research Areas</h1>
</header>
<p>
I am broadly interested in robotics and autonomous systems. The long-term goal of my research is to make robots safe, yet "street-smart". By street-smart, I mean that the robot should be able to make quick and intelligent decisions in novel and complex situations based on what it sees. Yet, the robot should be able to reason about the consequences of its decisions on its own safety and that of other robots and humans around it.
<br>
<br>
In short term, my research focuses on understanding how machine learning can be used to achieve the above goals. Particularly, I focus on two aspects of robot learning: <br />
(a) how can robots learn to make intelligent decisions in a data-efficient manner by combining learning with classical, model-based planning and control methods, especially when the robot relies on perception and vision sensors?<br />
(b) how can robots maintain safety while using learning for decision-making?
</p>
<!-- Content -->
<h2 id="content">1) Data-efficient task-based learning using models and optimal control </h2>
<p> Autonomous systems will inevitably experience external effects in unstructured environments, which are often hard to model using first principles. I have developed learning frameworks that leverage methods from system identification (SysID) and optimal control to efficiently capture these effects during the controller design process.
<br>
<br>
These frameworks include both learning the residual dynamics of the system while using the already known system structure, as well as learning models of external effects that are specific to the control task at hand. The latter does not necessarily learn the most accurate dynamics model; instead, it identifies a “coarse” model that can be learned with a small amount of data, and yet yields the best closed-loop controller performance when provided to the optimal control method used.
<li><a href="https://arxiv.org/abs/1703.09260"> aDOBO: Goal-Driven Dynamics Learning via Bayesian Optimization</a></li>
<li><a href="https://arxiv.org/abs/1909.05699"> Closed-Loop Model Selection for Kernel-based Models Using Bayesian Optimization</a></li>
<li><a href="https://arxiv.org/abs/1610.05863"> Learning Quadrotor Dynamics Using Neural Network for Flight Control</a></li>
<li><a href="https://arxiv.org/abs/1709.03153"> MBMF: Model-Based Priors for Model-Free Reinforcement Learning</a></li>
</p>
<center>
<div class="row">
<!-- <div class="col-4"><span class="image fit"><img src="images/Crazyflie_v2.png" style="height: 200px; width: auto" alt="" /></span></div> -->
<div class="col-4"><img src="images/Crazyflie_v2.png" style="height: 250px; width: auto" alt="" /><p>Learning residual aerodynamic effects for an improved control performance.</p></div>
<!-- <div class="col-4"><span class="image fit"><img src="images/kernel_selection.png" style="height: 200px; width: auto" alt="" /></span></div> -->
<div class="col-4"><img src="images/kernel_selection.png" style="height: 250px; width: auto" alt="" /><p>aDOBO uses learning with optimal control to stir jelly in a given pattern.</p></div>
<div class="col-4"><iframe width="400px" height="250px" src="https://www.youtube.com/embed/Too_GmSzOAA?start=1" frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe><p>Our presentation on aDOBO from the IEEE Conference on Decision and Control (CDC), 2017.</p></div>
</div>
</center>
<h2 id="content">2) Data-efficient architectures for learning-based perception with model-based control </h2>
<p> In many applications of interest, simple and well understood dynamics models are sufficient for control, and it is rather the vision and perception components that require learning. In this line of work, I focus on developing perception-action loops that efficiently combine deep learning-based perception with an underlying dynamics model for control, such as to navigate in a priori unknown environments. Leveraging underlying dynamics and feedback-based control not only accelerate learning, but also leads to trajectories that are robust to variations in physical properties and noise in actuation.
<li><a href="https://smlbansal.github.io/LB-WayPtNav/"> Combining Optimal Control and Learning for Visual Navigation in Novel Environments</a></li>
</p>
<center>
<div class="row">
<div class="col-4"><img src="images/visual_nav.png" style="height: 250px; width: auto" alt="" /><p>Learning-based perception with model-based control for data-efficient navigation in unseen environments.</p></div>
<div class="col-4"><iframe width="400px" height="250px" src="https://www.youtube.com/embed/7nC19Ai9u5E" frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe><p>A brief video explaining the key idea behind our framework for autonomous navigation in completely novel environments.</p></div>
</div>
</center>
<h2 id="content">3) Advancing the theory of optimal control for scalable safety analysis of learning-enabled systems </h2>
<p> In addition to improving data-efficiency, the models can also be used to design learning-based systems that are analyzable. In my work, I use model-based Hamilton-Jacobi-Isaacs (HJI) reachability analysis for safe learning and exploration. HJI analysis provides both the set of safe states and the corresponding safe controller for general nonlinear system dynamics. Since all system constraints are satisfied within this set, learning can be performed safely inside it.
<br>
<br>
The main challenge is to scale HJI analysis to real-world autonomous systems because of its exponential computational complexity with respect to the number of state variables. In my work, I address this challenge on multiple fronts by leveraging (a) the structure in dynamics and control strategy, (b) offline computations, and (c) modern computational tools to perform this analysis tractably.
<li><a href="https://arxiv.org/abs/1603.05208">Safe Sequential Path Planning of Multi-Vehicle Systems Under Disturbances and Imperfect Information</a></li>
<li><a href="Papers/Sequential_Trajectory_Planning.pdf">BEACLS: a C++- based reachability toolbox that leverages modern computational tools such as GPUs to improve computation speed of HJI reachability by nearly 100 times.</a></li>
<li><a href="https://arxiv.org/abs/1903.07715">Reachability-Based Safety Guarantees using Efficient Initializations</a></li>
</p>
<center>
<div class="row">
<div class="col-4"><img src="images/comp_time.pdf" style="height: 250px; width: auto" alt="" /><p>A 50-UAV reachability analysis takes less than 2 minutes with the CUDA implementation in BEACLS compared to 2.8 hours with the existing implementations.</p></div>
<div class="col-4"><iframe width="400px" height="250px" src="https://www.youtube.com/embed/1ocaBGZqSAE" frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe><p>Provably safe trajectory planning for large-scale multiple UAV systems using Hamilton-Jacobi reachability for various vehicle densities and wind speeds. </p></div>
</div>
</center>
<h2 id="content">4) Safety assurance for learning-enabled systems in unknown and human-centric environments </h2>
<p> As the autonomous system is operating in its environment, it may experience changes in system dynamics or external disturbances, or it may evolve via learning-in-the-loop. Consequently, safety assurances need to be evaluated and updated at operation-time. In this line of work, I build upon the developed tools in 3) to construct safety envelopes around learning-based perception and human motion prediction components, allowing robust motion plans to avoid collisions with obstacles and humans.
<li><a href="https://smlbansal.github.io/website-safe-navigation/">An Efficient Reachability-Based Framework for Provably Safe Autonomous Navigation in Unknown Environments</a></li>
<li><a href="https://arxiv.org/abs/1902.10320">A New Simulation Metric to Determine Safe Environments and Controllers for Systems with Unknown Dynamics</a></li>
<li><a href="https://abajcsy.github.io/hallucinate/">A Hamilton-Jacobi Reachability-Based Framework for Predicting and Analyzing Human Motion for Safe Planning</a></li>
</p>
<center>
<div class="row">
<div class="col-4"><iframe width="300px" height="200px" src="https://www.youtube.com/embed/BAm3DJFwHis" frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe><p>Hamilton-Jacobi reachability-based framework for computing a safety envelope around the learning-based perception component.</p></div>
<div class="col-4"><img src="images/spec.png" style="height: 200px; width: auto" alt="" /><p>Combining model-based and sampling-based methods for the safety analysis of a lane-keeping perception module.</p></div>
<div class="col-4"><<iframe width="300px0" height="200px" src="https://www.youtube.com/embed/uZi-zIi1S6A" frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe><p> Hamilton-Jacobi reachability analysis for finding all possible changes in the data-driven human prediction model for safe robot planning.</p></div>
</div>
</center>
</section>
</div>
</div>
<!-- Sidebar -->
<div id="sidebar">
<div class="inner">
<!-- Menu -->
<nav id="menu">
<header class="major">
<h2>Somil Bansal</h2>
</header>
<ul>
<li><a href="index.html">Homepage</a></li>
<li><a href="about.html">Bio</a></li>
<li><a href="research.html">Research</a></li>
<li>
<span class="opener">Teaching and Outreach</span>
<ul>
<li><a href="teaching.html">Teaching</a></li>
<li><a href="tutorial.html">Tutorials and Workshops</a></li>
<li><a href="outreach.html">Outreach</a></li>
<li><a href="mentoring.html">Mentoring</a></li>
</ul>
</li>
<li><a href="publications.html">Publications</a></li>
<li><a href="talks.html">Talks</a></li>
<li><a href="joinus.html">Join Us!</a></li>
</ul>
</nav>
</div>
</div>
</div>
<!-- Scripts -->
<script src="assets/js/jquery.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>