=Paper=
{{Paper
|id=Vol-1834/paper1
|storemode=property
|title=Autonomous Driving and Undergraduates: an Affordable Setup for Teaching Robotics
|pdfUrl=https://ceur-ws.org/Vol-1834/paper1.pdf
|volume=Vol-1834
|authors=Nicolò Arnaldi,Chiara Barone,Franco Fusco,Francesco Leofante,Armando Tacchella
|dblpUrl=https://dblp.org/rec/conf/aiia/ArnaldiBFLT16
}}
==Autonomous Driving and Undergraduates: an Affordable Setup for Teaching Robotics==
https://ceur-ws.org/Vol-1834/paper1.pdf
Autonomous Driving and Undergraduates:
an Affordable Setup for Teaching Robotics
N. Arnaldi, C. Barone, F. Fusco, F. Leofante and A. Tacchella
DIBRIS — Università degli Studi di Genova
Via all’Opera Pia, 13 — 16145 Genova — Italia
francesco.leofante@edu.unige.it – armando.tacchella@unige.it
Abstract. This work proposes an affordable setup for the application of
machine learning to autonomous driving. We discuss the hardware details
of the platform, and propose a scenario wherein learning from demon-
stration is leveraged to teach a vehicle how to drive. Preliminary results
are discussed, together with potential targets enabled by the setup.
1 Introduction
Autonomous vehicles have been attracting the attention of researchers in a va-
riety of engineering fields, and the first successful attempts at navigating cars
autonomously can be traced back to more than thirty years ago — see, e.g., [9,
14]. The interest in the topic is well motivated by the anticipated benefits of au-
tomated cars, such as the reduction in deaths from traffic accidents. Researchers
in robotics have taken up this challenge focusing on different applications that
range from low-level control of the vehicle [7, 5], to high level tasks such as safe
navigation through city streets [11, 12, 2]. Prize competitions have been funded
by major research organization, such as the DARPA Grand Challenge1 , created
to promote the development of technologies for fully autonomous ground vehi-
cles. Many companies have also joined the race to self-driving cars: Google X
launched an ambitious project to develop driverless cars2 [8], Tesla is already
commercializing cars with autopilot [10], while Apple, Microsoft and Amazon
allegedly made important investments in the same technology.
Given the current state of affairs, universities ought to introduce students to
Artificial Intelligence (AI) themes that are useful in autonomous driving, as this
addition enhances their curricula and helps them to keep up with the state of the
art (see, e.g., [4, 15, 13, 1] for similar initiatives). However, full-scale autonomous
cars are a daunting challenge when it comes to teaching for many reasons, e.g.,
limited budgets, demanding logistics, lack of qualified technical support. Mobile
robots could be devoted to this purpose, but their kinematic models are often
different from those of real cars. This abstract proposes an affordable setup to en-
gage undergraduate students in AI techniques for autonomous driving, involving
both hardware and software elements.
1
http://archive.darpa.mil/grandchallenge/
2
https://www.google.com/selfdrivingcar/
More specifically, our setup considers the problem of designing a controller for
the NXP-Cup challenge3 , a competition which aims at introducing undergrad-
uate students to embedded programming through the solution of a (simplified)
autonomous driving problem for 1/18th scale battery-powered cars. Instead of
using traditional control approaches, i.e., manual synthesis and tuning of a con-
troller, we propose to synthesize a controller using machine learning techniques.
Our proposal is based on Learning from Demonstration (LfD) — see, e.g., [3]
for a survey. Accordingly, the vehicle learns how to drive using a set of refer-
ence trajectories provided by a human controlling the car along a track. The
elements of the setup are affordable and the techniques involved are relatively
easy to grasp, so that undergraduates can keep up with the main challenge of the
NXP-Cup, which consists in having the model car drive faster than its competi-
tors while staying on track. A companion site containing datasets, preliminary
experimental results and software can be found at www.aimslab.org/teaching.
The remainder of this abstract is organized as follows. Section 2 details the main
features and requirements of the proposed benchmark, while Section 3 presents
a preliminary evaluation. Finally, we discuss challenges and future directions in
Section 4.
2 The setup
Context. The setup we propose is meant to compete in the NXP-Cup, an annual
context organized by NXP where students are required to build and program
a model car, fully equipped with a range of sensors, so that it can run au-
tonomously on a given track. The implementation of the car requires different
skills, including embedded software programming, control theory, and some basic
electronics.
Hardware. The basic hardware required for the competition is provided by NXP,
which also dictates the rules teams must adhere to. The basic kit includes a
battery-powered model car, a line scan camera used for sensing and a micro-
controller to control the car. Additional sensors can be employed, such as wheel
encoders or cameras, up to the limits defined by the NXP rules. In our setup,
the only additional sensor is a second line scan camera which enables a better
reconstruction of the track unraveling in front of the vehicle. Given our proposal
to teach the car how to drive using learning from demonstration, further compo-
nents are required which will be then removed during the competition. In partic-
ular, we operate the car using a 2.4GHz hobby-grade remote controller coupled
with a receiver. The signals from the controller are decoded to provide throttle
and steering inputs, so that samples for our machine learning algorithm can be
gathered under the form of reference trajectories. An Arduino board provides
throttle and steering decoding, and stores demonstration data to be transferred
3
Previously known as Freescale cup https://community.nxp.com/docs/DOC-1284?
tid=vanFREESCALECUP.
speed DC
motors
Ref.steer steer
User’s ON/OFF steer
Arduino KL25Z Steer
command
Line pos
pixels
dataset Cams
SD
Fig. 1. Functional diagram of the setup (left) and a snapshot of the model car with
the remote control for the teaching phase (right).
via Wi-Fi for further processing. In Figure 1 we show both a functional diagram
of the components involved and a snapshot of the current prototype4 .
Control software. While most NXP-Cup competitors design and implement con-
trollers manually, we propose to leverage state-of-the-art machine learning tech-
niques. By doing so, we give students the chance to explore applications of AI
solutions to a simple, yet challenging, autonomous driving problem. Students are
required to formalize the learning problem as a LfD task and gather demonstra-
tion data by teleoperating the model car. After this initial phase, they have to
choose the specific learning technique — e.g., Neural Networks, SVRs, Gaussian
Processes — and assemble the training dataset accordingly. This means they will
have to reason and experiment about which sensory inputs are to be considered
to obtain the best learning performances from a given learning algorithm. Ide-
ally, students should be able to implement a simple state machine that includes
the learned mapping from sensors to actuators in order to successfully drive the
car along the track.
Simulator. A simulation of the model car running on a track has also been im-
plemented in Matlab. Using the simulator, students can focus on dataset and
algorithm design, avoiding hardware-related issues. A Simulink block containing
the kinematic model of the car has been implemented. This allows the user to
drive the car by specifying reference values for steering angle and speed. The
simulator shows the car displacing on a user-defined track and also allows to
change the geometric parameters of the line-scan camera, obtaining different
field of views, and therefore different performances. The code of the simulator
can be found in the companion site of this project. After data has been collected
from the simulator the learning process can take place. In our preliminary im-
plementation, demonstration data is fed to a Python script which implements
Gaussian Process Regression using the GPy library [6] developed by the Uni-
versity of Sheffield. Once a solution for the problem is learned — e.g., under
4
The total cost of the platform as depicted in Figure 1 is approximately 350 euros.
the form of policy or mapping function — it can be tested in the simulator to
evaluate its performances.
3 Preliminary Evaluation
Preliminary experiments on the setup shown in Figure 1 have been carried out
to investigate the feasibility of our proposal when it comes to teaching under-
graduate students. In the following experiments, the car is not equipped with
wheel encoders — which are a common addition in NXP competitions — and
data about the steering angles is read directly from the motors. Other than this,
the configuration of the car is NXP-Cup-legal, so it reflects intended usage sce-
narios. An instance of the LfD problem is considered where the objective is to
learn a function mapping from camera readings at time t to steering angles to be
applied at time t+1. Demonstration data was collected using the on-board archi-
tecture of Figure 1, learning was done off-board. Trials were run on two different
tracks: (i.) a loop with two straights and two 180-degrees turns, and (ii.) a more
complex track, including crossings of different shapes. The first three authors,
who are graduate students at the University of Genoa, tested the framework as
if they were tackling the problem for the first time. As a result, they managed
to have the car learn how to drive on the first track in a reliable way. However,
the car could only move at a limited speed. The second track proved even more
challenging: the car dealt badly with crossings, suggesting that improvements in
sensory data and/or learning setup should be considered. While analyzing the
dataset gathered during this experimental campaign we noticed that large parts
of the input space where never explored during teaching. This was mostly due
to sensors not being able to provide useful feedbacks for learning, e.g., line-scan
cameras giving blank readings most of the time as the borders of the track were
hardly seen during teaching.
4 Conclusion
We believe that, once some aspects related to reliable sensory acquisition are
solved in the current prototype, the proposed setup will provide an excellent
testbed for teaching undergraduate students basic principles in AI and Robotics.
The hardware setup is affordable and can be extended on a limited budget. Stu-
dents can thus experiment easily how different sensors could affect the outcome
of the learning process. At the same time, the setup is standardized as the car
must comply to the NXP-Cup rules. Furthermore, working on this problem al-
lows to deal with a complete robotics project, from low-level control to high-level
decision-making. On the AI side, the problem of teaching a car how to drive is
not trivial, even in this simplified setup. Moreover, students have to deal with
limited hardware resources, e.g., memory shortage, limited sensor feedbacks. On
top of this, the controller obtained not only has to drive the car along the track,
but it also has to be fast enough to compete with traditional controllers.
References
1. AAAI Spring Symposium on Educational Advances in Artificial Intelligence. www.
aaai.org/Symposia/EAAI/eaai-symposia.php, accessed: 2016-11-02
2. Althoff, M., Stursberg, O., Buss, M.: Model-based probabilistic collision detection
in autonomous driving. IEEE Transactions on Intelligent Transportation Systems
10(2), 299–310 (2009)
3. Argall, B.D., Chernova, S., Veloso, M., Browning, B.: A survey of robot learning
from demonstration. Robotics and autonomous systems 57(5), 469–483 (2009)
4. Beux, S., Briola, D., Corradi, A., Delzanno, G., Ferrando, A., Frassetto, F., Guer-
rini, G., Mascardi, V., Oreggia, M., Pozzi, F., Solimando, A., Tacchella, A.: Com-
putational thinking for beginners: A successful experience using prolog. In: Pro-
ceedings of the 30th Italian Conference on Computational Logic, Genova, Italy,
July 1-3, 2015. pp. 31–45 (2015)
5. Falcone, P., Borrelli, F., Asgari, J., Tseng, H.E., Hrovat, D.: Predictive active
steering control for autonomous vehicle systems. IEEE Transactions on control
systems technology 15(3), 566–580 (2007)
6. GPy: GPy: A gaussian process framework in python. http://github.com/
SheffieldML/GPy (since 2012)
7. Gray, A., Gao, Y., Hedrick, J.K., Borrelli, F.: Robust predictive control for semi-
autonomous vehicles with an uncertain driver model. In: Intelligent Vehicles Sym-
posium (IV), 2013 IEEE. pp. 208–213. IEEE (2013)
8. Guizzo, E.: How googles self-driving car works. IEEE Spectrum Online, October
18 (2011)
9. Kanade, T., Thorpe, C., Whittaker, W.: Autonomous land vehicle project at cmu.
In: Proceedings of the 1986 ACM fourteenth annual conference on Computer sci-
ence. pp. 71–80. ACM (1986)
10. Kessler, A.M.: Elon musk says self-driving tesla cars will be in the us by summer.
The New York Times p. B1 (2015)
11. Kümmerle, R., Ruhnke, M., Steder, B., Stachniss, C., Burgard, W.: A navigation
system for robots operating in crowded urban environments. In: Robotics and
Automation (ICRA), 2013 IEEE International Conference on. pp. 3225–3232. IEEE
(2013)
12. Kümmerle, R., Ruhnke, M., Steder, B., Stachniss, C., Burgard, W.: Autonomous
robot navigation in highly populated pedestrian zones. Journal of Field Robotics
32(4), 565–589 (2015)
13. Sintov, N., Kar, D., Nguyen, T.H., Fang, F., Hoffman, K., Lyet, A., Tambe, M.:
From the lab to the classroom and beyond: Extending a game-based research plat-
form for teaching AI to diverse audiences. In: Proceedings of the Thirtieth AAAI
Conference on Artificial Intelligence, February 12-17, 2016, Phoenix, Arizona, USA.
pp. 4107–4112 (2016)
14. Wallace, R., Stentz, A., Thorpe, C.E., Maravec, H., Whittaker, W., Kanade, T.:
First results in robot road-following. In: IJCAI. pp. 1089–1095 (1985)
15. Wollowski, M., Selkowitz, R., Brown, L.E., Goel, A.K., Luger, G., Marshall, J.,
Neel, A., Neller, T.W., Norvig, P.: A survey of current practice and teaching of
AI. In: Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence,
February 12-17, 2016, Phoenix, Arizona, USA. pp. 4119–4125 (2016)