50 h Annual Con e ence in Sep embe 2022
979
Joaquim Blesa, Pe e Ponsa, An onio Caloma de, Jona han Ga cia and Vic o RepechoIn oducing Au onomous Vehicles in o an Une g adua e Enginee ing Cou se - 10.5821/con e -
ence-9788412322262.1274
doi:10.5821/con e ence-9788412322262.1274
INTRODUCING AUTONOMOUS VEHICLES INTO AN
UNERGRADUATE ENGINEERING COURSE
J. Blesa 1
Ad anced Con ol Sys ems (SAC) Resea ch G oup, Se a Hún e Fellow, Uni e si a
Poli ècnica de Ca alunya (UPC), Au oma ic Con ol Depa men (ESAII), Edua d
Ma is any, 16, 08019 Ba celona, Spain
ORCID 0000-0002-5626-3753
P. Ponsa
In elligen Con ol Sys ems (SIC) Resea ch G oup, Uni e si a Poli ècnica de
Ca alunya (UPC), Au oma ic Con ol Depa men (ESAII), Edua d Ma is any, 16,
08019 Ba celona, Spain
ORCID 0000-0001-6306-7251
A. Caloma de
High Pe o mance In eg a ed Ci cui s and Sys ems Design (HIPICS) G oup,
Uni e si a Poli ècnica de Ca alunya (UPC), Ba celona, SPAIN
ORCID 0000-0002-4459-8505
J. Ga cía
Uni a de se ei labo a o is d'elec ònica i au omà ica, Escola d’Enginye ia de
Ba celona Es (EEBE) Uni e si a Poli ècnica de Ca alunya (UPC), Edua d
Ma is any, 16, 08019 Ba celona, Spain
ORCID 0000-0002-0707-6409
V. Repecho
Ad anced Con ol o Ene gy Sys ems (ACES) Resea ch g oup, Se a Hún e Fellow,
Uni e si a Poli ècnica de Ca alunya (UPC), Au oma ic Con ol Depa men (ESAII),
Edua d Ma is any, 16, 08019 Ba celona, Spain
ORCID 0000-0003-0584-5443
1 Co esponding Au ho
J. Blesa
[email protected]
50 h Annual Con e ence in Sep embe 2022
980
Con e ence Key A eas: Enginee ing Skills, Na iga ing Open Lea ning
En i onmen s
Keywo ds: Au onomous Vehicles, Lea ning by doing, Low-cos AV, Unde g adua e
s uden s
ABSTRACT
Au onomous ehicles (AVs) a e o g ea in e es o he au omo i e indus y and a e
expec ed o e olu ionize mobili y and public anspo a ion. The uni e si y can
con ibu e o he design and de elopmen o au onomous ehicles bo h in he ield o
eaching and in esea ch and echnology ans e . In his pape , i is desc ibed how
his opic is in oduced in an unde g adua e enginee ing cou se, “Implemen a ion o
Au oma ic Con ol Sys ems (IACS)”. The IACS cou se is based on p ojec based
lea ning (PBL) and lea ning by doing me hodologies. Se e al p ac ical examples ha
co espond o eal au oma ic sys ems a e discussed h oughou he cou se and one
o hem, a low-cos AV o which a Raspbe y pi has been adap ed, o ms he basis o
a inal p ojec o he cou se. The con ol algo i hms a e de eloped on
MATLAB/SIMULINK and a e sen o he Raspbe y h ough a wi eless communica ion
ne wo k. The con ol objec i e o he sys em is he au oma ic guidance o he ehicle
h ough a single lane indoo closed ci cui , he de ec ion and iden i ica ion o di e en
a ic signals and he au oma ic esponse o hese signals. S uden s check he
beha io o he ehicle and p oceed o make imp o emen s. Based on he assessmen
o he s uden s and he obus ness o he au onomous ehicles, i is ime o consolida e
his ype o p ojec wi hin he cou se. S uden s ha wan o ge deepe in o he ma e
ha e he oppo uni y o do a inal deg ee p ojec ela ed wi h he AV.
1 INTRODUCTION
1.1 Au onomous ehicles
Au onomous ehicles (AVs) a e expec ed o e olu ionize mobili y and public
anspo a ion [1]. The e a e di e en ypes o au onomous ehicles o esea ch and
academic pu poses. A sys em classi ica ion is equi ed. The mos sys em
classi ica ion is he one p oposed by he Socie y o Au omo i e Enginee s (SAE) [2].
In his classi ica ion he e a e i e le els o au oma ion: om low le els ela ed o
human d i e s assis ance o high le els a e ela ed o high au oma ion o ully
au oma ion ( he au onomous ehicle has he unc ionali y o aul ole an ha dwa e).
The con ex o his classi ica ion is ope a e an AV in on- oad a ic and de eloping
dynamic d i ing asks. The d i ing au oma ed sys ems is capable o guiding he AV
h oughou comple e ip, in e ac ing wi h o he AV, sol ing con ingencies, minimizing
isk, di e se a ic and wea he condi ions.
In his pape au onomous ehicles a e ela ed o low cos au onomous ehicles wi h
he lexibili y o inco po a e con ol elec onics, communica ions wi h he en i onmen
and scalable unc ionali y o gene a e a wide ange o expe imen al asks. In a i s
app oach he con ex o use is in labo a o y condi ions. The dynamic d i ing asks o
he SAE mus be adap ed in his scena io. The main objec i e is he design and
50 h Annual Con e ence in Sep embe 2022
981
de elopmen o a obus AV and p oo ha his AV can pe o m dynamic d i ing asks
wi h high e ec i eness.
2 METHODOLOGY
2.1 Academic aspec s
In he uni e si y indus ial bachelo deg ees o he Ba celona Eas School o
Enginee ing (EEBE) on he Diagonal-Besòs Campus o he Uni e si a Poli ècnica de
Ca alunya (UPC), au ho s de elop he cou se “Implemen a ion o Au oma ic Con ol
Sys ems (IACS)”. This cou se is an elec i e cou se in he las semes e o he
indus ial deg ees wi h 6 Eu opean C edi T ans e and Accumula ion Sys em (ECTS).
This cou se applies he con ol heo y in he de elopmen o elec onic p o o ypes and
using ools and de ices as MATLAB and Raspbeb y pi.
I is an in e disciplina y cou se since s uden s come om all a eas o enginee ing ha
a e augh in he eaching cen e (mainly indus ial elec onics and au oma ic con ol,
elec ical and mechanical enginee s). The p e equisi es o elec onics, p og amming
and linea con ol heo y ha a e necessa y o s uden success a e p o ided in
p e ious co e cou ses. Good skills in eam wo k a e equi ed o he design,
de elopmen , p og amming and e alua ion o au onomous ehicles p o o ypes. The
cou se is also a ailable o E asmus / Exchange s uden s ha can ha e di e en
backg ounds.
The appea ance o ailu es, unexpec ed ehicle beha io , ba e y pe o mance,
pe o mance o p og ammed algo i hms, equi e he g oup o show i s abili y o esol e
con ingencies and ob ain good ehicle pe o mance.
Au ho s a e d i ing oo inal deg ee p ojec s wi h he aim o de elop i s au onomous
p o o ypes wi h help o he main enance s a . Wi h he idea o building scalable
sys ems in mind, he s uden 's aining is ocused on ecommending ha hey i s ake
he cou se and secondly ha hey can de elop he inal deg ee p ojec .
Fo he de elopmen o ully AV, mo e capabili ies a e equi ed. In his sense, he
app oach p esen ed in his pape should be ex ended o enginee ing mas e 's cou ses
and esea ch ac i i ies. The au ho s a e de eloping ac i e wo k in he de elopmen o
aul ole an and sa e y AV and can p o ide eedback. The esea ch is de eloped in
a esea ch labo a o y and he e is a ela ionship wi h he academic labo a o y
commen ed in his pape .
2.2 Labo a o y P ac ices
The IACS cou se is based on PBL [3] and lea ning by doing [4] me hodologies. Se e al
p ac ical examples ha co espond o eal sys ems a e discussed h oughou he
cou se and one o hem, an AV, o ms he basis o a inal p ojec . The e is a s uden
eaching guide o e e y p ac ical example wi h he desc ip ion o he s eps o ollow
o success in he implemen a ion o an au oma ic con ol in a eal sys em (modeling,
iden i ica ion, design, implemen a ion and assessmen o he pe o mance) and a
desc ip ion o he esul s ha he s uden should ob ain.
50 h Annual Con e ence in Sep embe 2022
982
In pa icula , in he AV example, he implemen a ion o a simple d i ing au oma ed
sys em in an adap ed Van ehicle (shown in Figu e 1) is p oposed. The ehicle is
equipped wi h a Raspbe y Pi 3 B boa d, a se omo o (s ee ing angle), a DC mo o
( o wa d and e e se speed o he ehicle) and a came a. The powe supply o he
module consis o 5 AA echa geable ba e ies and a ol age egula o o 5 Vol s. The
o al cos o he componen s o he adap ed Van ehicle is a ound 200€.
Fig. 1. Van ehicle adap ed wi h Raspbe y Pi 3 B+.
The sys em is p esen ed as a pa icula au oma ic con ol sys em composed by he
di e en elemen s: plan (Van), con olle (Raspbe y boa d), ac ua o s (mo o s) and
senso (Came a). And he con ol objec i e is de ined as he au oma ic guidance o
he ehicle h ough he single lane closed indoo ci cui depic ed in Figu e 2.
Fig. 2. Single lane closed ci cui in he Au oma ion and Indsu ial Robo ics Lab. o he
EEBE
50 h Annual Con e ence in Sep embe 2022
983
The AV guided p ac ice consis in h ee di e en s eps:
a) connec ing he ehicle wi h MATLAB/Simulink h ough a wi eless
communica ion ne wo k WIFI conec ion ( ou e WLAN connec ed in he Cisco
Swi ch o he labo a o y) in such a way i is possible o p ocess he images
cap u ed by he came a and con ol he s ee ing angle and speed o he ehicle.
b) how o ex ac he bounds o he ci cui lane (in g een colo ) om he images o
he came a.
c) how o use he in o ma ion ex ac ed om he ci cui lane bounds o y o align
he ehicle wi h he lane o he ci cui .
Fo his las s ep, he il e ed image ha con ains he bounds o he lanes is
bina ized and he cen oid o he image is compu ed (Figu e 3). When he ehicle
is comple ely aligned wi h he lane, he cen oid will be aligned wi h he e ical axis
o he image. The e o e, he angle o he cen oid wi h he e ical axis is used as
he con olled a iable in a close loop sys em whe e he con olled a iable is he
di ec ion o he ehicle and he con ol objec i e is o main ain as close as possible
he con olled a iable o ze o ( ehicle aligned wi h he lane).
Fig. 3. Raw and bina ized images wi h cen oid and angle de ia ion δ compu a ions.
In his con ol sys em he e is ano he a iable ha has an impo an ole in he
pe o mance o he sys em, he speed o he ehicle. In his i s app oxima ion o he
au oma ic guidance i is conside ed o be cons an . Finally, s uden s a e encou aged
o de ec he limi a ions o his i s app oxima ion o he au oma ic guidance o he
ehicle and he inal p ojec o he cou se is p oposed as he implemen a ion o a new
(and o iginal) solu ion. This solu ion should be obus , eliable, s able and should
include o he capabili ies as he de ec ion and iden i ica ion o di e en a ic signals
and he au oma ic esponse o hese signals.
2.3 P ojec de elopmen .
When de ining he cou se con en , p o esso s discussed abou he so wa e o be used
in he labo a o ies. As all s uden s o EEBE school and mos o E asmus s uden s ha e
used MATLAB p og am in p e ious cou ses and some o hem e en ha e used
Simulink, he elec ion was clea . In addi ion, he in e ace o his so wa e makes
p og amming and lea ning easy and i is ela i ely easy o implemen complex sys ems
as he one depic ed in Figu e 4 ha implemen s a possible solu ion o he AV sys em.

50 h Annual Con e ence in Sep embe 2022
984
Fig. 4. Blocs o Simulink ci cui ha implemen s AV solu ion
Red boxes in Figu e 4 show how is possible o de elop scalable unc ionali y using
Simulink (lane de ec ion phase, ehicle con ol phase, e c.).
To de elop he p ojec , he s uden s wo k in g oups o wo o h ee and hey ha e 12
hou s o expe imen al sessions o ca y ou all he wo k. The p o esso discuss wi h
he s uden eam and helps o iden i y he co ec solu ions bu he inal solu ion and
he ea u es o be imp o ed a e comple ely p oposed and de eloped by he s uden s.
In he las lab session, each s uden eam p esen s a summa y o he p ojec as well
as a demons a ion o how he AV sys em wo ks. A e his p esen a ion, p o esso s
ask some ques ions in o de o cla i y possible con using poin s and o assess ha all
he membe s o he eam g oup know in deep he inal design. The e alua ion o he
inal p ojec is done conside ing he o al exam and he epo o he p ojec . This inal
e alua ion o he o e all p ojec coun s o 40% o he inal ma k o he cou se.
2.4 Final deg ee p ojec .
S uden s ha wan o ge deepe in o he ma e ha e he oppo uni y o do a inal
deg ee p ojec ela ed wi h he opic imp o ing he ehicle used in he IACS cou se
and using mo e p o essional p og amming en i omen s as he Robo Ope a ing
Sys em (ROS). ROS equi es lowe -le el concep s and a s eepe lea ning cu e han
MATLAB/Simulink en i omen bu allows he use o powe ul ision lib a ies as
OpenCV [5] and he implemen a ion o ad anced a i icial in elligence algo i hms o
he con ol o he AV as in cu en p o essional solu ions [6][7].
Figu e 5 shows he g aph diag am ha implemen s a possible solu ion o he AV
sys em: F om he ini ial nodes aspicam_node and con olle ha allow he cap u e o
images and he implemen a ion o manual/au oma ic mode selec ion espec i ely. To
he inal node ca _con ol ha implemen s he aun onomous con ol conside ing he
en i onmen in o ma ion ob ained om he image p ocessing s eps and he con olle
s a e. This solu ion was p oposed in a inal deg ee p ojec [8].
50 h Annual Con e ence in Sep embe 2022
985
Fig. 5. ROS g aph diag am showing nodes and opics o con ol he AV.
Taking in o accoun he s uden s’expe ience and eedback, new and mo e complex
solu ions usually equi e modi ica ions no only in he p og amming bu also in he
ha dwa e se up as o example he use o mo e powe ul Boa ds ha can equi e a
enhanced powe supply. Fo example, ROS so wa e equi es a leas Rasbe y Pi 4
B boa d ha inc eases he cu en consump ion. The e o e, he new AV equi es a
mo e complex powe supply uni . In [8] a li hium polyme ba e y (LIPO) was selec ed
as a sui able solu ion in ead o he o iginal AA echa geable ba e ies solu ion as i is
shown in Figu e 6.
Fig. 6. Van ehicle adap ed wi h Raspbe y Pi 4 B.
50 h Annual Con e ence in Sep embe 2022
986
3 RESULTS
A he end o e e y qua e , an anonymous su ey is conduc ed by he UPC o ob ain
eedback om he s uden s o he di e en cou ses. In pa icula , wo ques ions a e
asked: “1. The con en s o he cou se seemed in e es ing o me” and “2. O e all, I am
sa is ied wi h he cou se”. Bo h ques ions a e sco ed om 1 (s ongly disag ee) o 5
(s ongly ag ee) and in he case o IACS cou se he s uden s ha e g aded an a e age
sco e o 4.14 and 4.43 o he i s and second ques ions espec i ely. Tha is a sco e
0.3 and 0.9 poin s abo e he sco e ob ained in a e age o all he cou ses o e ed by
he UPC. In addi ion, mo e han a hal o he s uden s ha ake he IACS cou se a e
in e es ed in doing he inall deg ee p ojec ela ed wi h he opics o he cou se.
Rega ding academic esul s, 13% o he s uden s ha e sa is ac o y esul s [5,7), 40%
e y good esul s [7-9) and 40% e y good esul s. Finally, only 7% o s uden s ha e
no sa is ac o y esul s [0,5) because hey ha e p oblems o a end o he expe imen al
sessions.
4 SUMMARY AND ACKNOWLEDGMENTS
This a icle p esen s an example o how Au onomous Vehicles ha e been included in
an unde g adua ed enginee ing cou se in a mo i a ing way, in pa icula , in an elec i e
cou se o he indus ial bachelo deg ees o he Ba celona Eas School o Enginee ing
a UPC. AV ha e been in oduced i s in an in oduc o y guided p ac ical lab and a e
ollowing lea ning by doing me hodology o sol e a eal-wo ld p oblem. Lea ning by
doing is used o inc ease s uden s’ in e es and a he same ime help hem o acqui e
and apply new knowledge in a p oblem-sol ing con ex . S uden su eys and
assessmen esul s con i m he s uden ’s g ea in e es and success in he cou se.
As u u e lines, he au ho s wan o in oduce AV in o he cou ses. In pa icula , in an
ad anced con ol elec i e cou se in he las semes e o he Bachelo 's deg ee in
Indus ial Elec onics and Au oma ic Con ol Enginee ing. P e ious backg ound o
s uden s in au oma ic con ol, p og amming and elec onics will be deepe han he
one o a e age s uden s in IACS cou se. This ci cums ance will allow he use o ROS
en i onmen in he implemen a ion o AV algo i hms. Howe e , a key poin o ob ain
sa is ac o y esul s will be he design o educa ional modules o simpli y lea ning abou
his en i onmen . Because ROS is no in he cu icula o p e ious cou ses o his
Bachelo deg ee.
This wo k has been pa ially unded by SaCoAV p ojec (Re . PID2020-114244RB-
I00) om he Spanish Minis y o Science and Inno a ion.
REFERENCES
[1]
Bösch, P. M., Becke , F., Becke , H., and Axhausen, K. W. (2018), Cos -
based analysis o au onomous mobili y se ices,T anspo Policy, Vol 64,
pp. 76-91.
50 h Annual Con e ence in Sep embe 2022
987
[2]
Zanchin, B.T., Adamshuk, R., San os, M. M., Collazos, K.S. (2017), On he
ins umen a ion and classi ica ion o au onomous ca s, IEEE In e na ional
Con e ence on Sys ems, Man, and Cybe ne ics (SMC) Ban Cen e , Ban ,
Canada, pp. 2631-2636.
[3]
Cos a-Cas ello, R., Puig, V., Blesa,
J. (2016), On Teaching Model-Based Faul
Diagnosis in Enginee ing Cu icula [Lec u e No es].
IEEE Con ol Sys ems
Magazine, ol. 36, no. 1, pp. 53-62.
[4]
Dominguez-Ramos, A., Al a ez-Gue a, M., Diaz-Sainz, G., Ibañez, R.,
I abien, A. (2019), Lea ning-by-Doing: The Chem-E-Ca Compe i ion® in he
Uni e si y o Can ab ia as case s udy, Educa ion o Chemical Enginee s, ol.
26, pp. 14-23.
[5]
B adski, G. (2000). The openCV lib a y. D . Dobb's Jou nal: So wa e Tools o
he P o essional P og amme , ol. 25, no 11, pp. 120-123..
[6]
Vincke B, Rod iguez Flo ez S, Aube P. An Open-Sou ce Scale Model Pla o m
o Teaching Au onomous Vehicle Technologies. Senso s. 2021; 21(11):3850
[7]
Lo i, N, No osad, J. A., Phan-Van, H (2019, June), A Mul idisciplina y Cou se
and he Co esponding Labo a o y Pla o m De elopmen o Teaching he
Fundamen als o Ad anced Au onomous Vehicles Pape p esen ed a 2019
ASEE Annual Con e ence & Exposi ion , Tampa, Flo ida. 10.18260/1-2—31975
[8]
Sole , J. (2022). Design, implemen a ion and con ol o an au onomous ehicle
based on aspbe y pi. Final deg ee p ojec o Bachelo 's deg ee in Indus ial
Elec onics and Au oma ic Con ol Enginee ing (EEBE-UPC).