ARTEMISA: Architecture of an eco-driving assistant based on the anticipation

Abs ac This pape p esen s he a chi ec u e o an eco- d i ing assis an . The assis an e alua es he ul il- men o classic eco-d i ing ad ices such as: main- ain a cons an speed, d i ing a high gea , slow down smoo hly and so on. In addi ion, he assis an issues ad ices based on he an icipa ion. An icipa- ion is he key o eco-d i ing. The assis an is ca- pable o de ec ing a ic signs be o ehand and i checks i he speed is sui able o no ha ing o slow down sha ply. In addi ion, he sys em p opos- es an op imal a e age speed acco ding o he con- di ions o he oad. To model he en i onmen whe e he ehicle is mo ing, we use an And oid mobile de ice. These de ices a e ideal due o o hei mul iple ne wo k connec ions (Blue oo h, UTMS and WIFI) and sen- so s (came a, accele a ion senso , GPS and so on). To ob ain he ehicle’s pa ame e s (speed, uel consump ion, RPM, e c.), we use he diagnos ic po (OBD2). The p oposed sys em can imp o e uel consump- ion and sa e y. In addi ion, i is independen o he ype o ehicle. 1 In oduc ion The numbe o ehicles has inc eased in ecen yea s. As a esul , i has inc eased he uel consump ion and he emis- sion o gaseous pollu an s. The emission o gaseous pollu- an s causes mo e dea hs han a ic acciden s. On he o he hand, he ene gy esou ces a e limi ed and he inc ease in demand causes hem e en mo e expensi e. In his con ex , many solu ions ha e appea ed ha seek o educe uel consump ion and g eenhouse gas emissions. The e a e solu ions based on he educ ion o he weigh o he ehicle, imp o emen s in ae odynamics, imp o emen s in he engine, e c. Among hese solu ions, he e is one ha has gained g ea impo ance in ecen yea s: he Eco- D i ing. Eco-d i ing is a d i ing echnique ha sa es uel ega dless o he echnology. This d i ing echnique consis s on applying se o ules such as: Main aining cons an speed, d i ing a a high gea , d i ing a 90 Km/h maximum, a oiding speeding up and slowing down sha ply and so on. Applying he eco-d i ing ules, we may sa e be ween 10 and 25% o uel [Mu aki and Kanoh, 2008][Ba bé and Boy, 2006][Mie lo e al., 2004][Koskinen, 2008]. Howe e , he pe cen age o uel economy will depend on he ype o ehi- cle. Fo example he hyb id ehicles only sa e 10% [Lind- eld e al., 2010] The e a e a lo o esea ch s udies on eco-d i ing. Some au ho s y o ind pa ame e s ha a ec uel consump ion and de e mine how hey in luence uel consump ion. [E ics- son, 2001], sugges ed ha o sa e uel we should a oid hea y accele a ion and high speed d i ing. [Johansson e al, 2003] p oposed main aining low decele a ion le els, mini- mizing he use o 1s and 2nd gea s, inc easing he use o 5 h gea , and block changing gea s o sa e uel. [Kuhle and Ka ens, 1978], iden i ied a se o en a iables ha in lu- ence ene gy consump ion and in he emission o g eenhouse gases. O he au ho s e alua e he e ec o using an eco-d i ing assis an in he uel consump ion and dis ac ions ha may cause. [Bo iboonsomsin e al, 2009] e alua es he sui abili y o eco-d i ing assis an o acqui e knowledge abou eco- d i ing. [Klaue e al., 2006] concluded ha dis ac ion due o seconda y asks like in e ac ing wi h a mobile de ice con ibu ed o o e 22% o all c ashes. [Riene e al., 2010] p oposed a ib o- ac ile no ifica ion sys em in o he ca sea (ei he in he sa e y bel o he sea ing), o wa n abou he cu en CO2 eƥciency. A way o add ess he eco-d i ing is o in luence in he pa- ame e s ha he d i e con ols a any ime such as: speed, accele a ions and gea . In [Saboohi and Fa zaneh, 2009], he au ho s p oposed a con ol s a egy o d i e e icien ly. This con ol s a egy de e mines he adequa e speed and adequa e gea a any gi en ime. [Casa ola e al., 2010] analyzed wo algo i hms o de e mine he mos e icien gea o uel ARTEMISA: A chi ec u e o an eco- d i ing assis an based on he an icipa ion V.Co coba Magaña1, M.Muñoz O gane o1, A. Fe nández-Mon es2, J. A. O ega2 1Dp o. de Ingenie ía Telemá ica. Uni e sidad Ca los III. Leganes, Spain [email p o ec ed],[email p o ec ed], 2Depa amen o de Lenguajes y Sis emas In o má icos Uni e sidad de Se illa Spain [email p o ec ed],[email p o ec ed] $ FWDVGH;,9-RUQDGDVGH$5&$ 6LVWHPDV&XDOLWDWLYRV VXV$SOLFDFLRQHVHQ ' LDJQRVLV5REyWLFDH,QWHOLJHQFLD$PELHQWDO ) -5XL]1$JHOO-$2UWHJD(GV F  sa ing a each ime. [Ba h and Bo iboonsomsin, 2009] p oposed an algo i hm o de i ing he ecommended se ehicle speed, based on eal- ime measu emen s om a a ic measu emen sys em. In [Ke e al., 2010], he au ho s p oposed a Min-Max an algo i hm o ind an op imized ehicle speed and accele a ion wi h espec o uel- e iciency. In his pape , we p opose an eco-d i ing assis an o help he d i e o adop ing an e icien d i ing s yle. The assis- an will wa n he use when i does no comply wi h he basic ules o eco-d i ing. Also, i ecommends an op imal a e age speed. The eco-d i ing assis an uns on an And oid mobile de ice. These de ices a e sui able due o i s mul iple connec ions and senso s. Howe e he p esen ed a chi ec- u e could be un on ano he pla o m. 2 Eco-d i ing Assis an [A emisa P ojec , 2012] aims o sa e uel by modi ying he beha io o he d i e . To do so, we can use s a ic eco- d i ing ad ices o dynamic eco-d i ing ad ices. Some s a ic ad ices a e: no o d i e as , no accele a e sha ply, d i ing a high gea s, main aining speed cons an and so on. Dynamic Eco-d i ing ad ices a e hose which ake in o accoun he cu en condi ions such as: a ic densi y, ehi- cle speed, wea he condi ions, a ic acciden s and so on. The p oposed eco-d i ing assis an akes in o accoun hese wo ypes o ad ices. The e ec i eness o eco-d i ing assis an a endees o modi y he beha io o he d i e has been widely p o en. Figu e 1. A chi ec u e o eco-d i ing assis an In Figu e 1, we can see a schema o he p oposed sys em. Eco-d i ing assis an has he ollowing componen s: xDa a Acquisi ion Sys em: his componen is e- sponsible o ob aining in o ma ion abou he ehi- cle and he en i onmen . The in o ma ion ob ained will be used o de e mine wha eco-d i ing ad ices should be issue. xP ep ocessing Module: his componen is esponsi- ble o il e ing he da a collec ed by he da a ac- quisi ion componen and gene a ing addi ional in- o ma ion by analyzing he da a collec ed. xEco-d i ing ad ices: his module is esponsible o e alua ing he d i e d i ing s yle and based on his e alua ion issuing eco-d i ing ad ices, so ha , he d i e will change he nega i e aspec s o his o he d i ing s yle. Fo example: I he d i e b akes sha ply, he eco-d i ing assis an will issue: You should slow down smoo hly. xP edic i e ad ices: his module p edic s in ad ance si ua ions ha cause an inc ease in uel consump- ion o wa n he d i e . Also, i indica es ha ac- ions should be aken o p e en such si ua ions. xUse in e ace: his module is esponsible o p e- sen ing he eco-d i ing ad ice and wa ning he us- e . 2 Da a Acquisi ion Sys em Da a Acquisi ion Sys em ob ains he alue o all a iables ha in luence uel consump ion o ha can help o p edic he ac ions o be pe o med by he d i e o sa e uel. I uses as in o ma ion sou ces he ollowing means: xIn e ne : S a e o he oad and wea he condi ions xCame a: T a ic Signs De ec ion xGPS: Vehicle Loca ion xOBD2: Vehicle speed, RPM, uel consump ion and a el dis ance In e ne in o ma ion T a ic Densi y and Wea he Condi ions a e ob ained om he [DGT, 2012] web se ice and [AEMET, 2012] web se ice. The in o ma ion is p o ided in XML o ma . The And oid XMLNull lib a y is used o p ocessing he ile XML. Came a Eco d i ing assis an uses [OpenCV p ojec , 2012] lib a y o ake he pho og aph om he oad. OpenCV is c oss- pla o m, he e a e e sions o GNU / Linux, Mac OS X , Windows and And oid. I con ains o e 500 unc ions co - e ing a wide ange o a eas in he ision p ocess as objec ecogni ion ( ace ecogni ion). This lib a y makes in e nal use o And oid na i e lib a ies o access o he came a. Use In e ace Holo Fon TTS Eco-d i ing ips Eco-d i ing Ad ices Da a Acquisi ion Sys em In o ma ion Gene a o OBD2 Fil e In e ne Fil e Op imal A e age Speed Module T a ic Signs De ec ion P e en i e Eco-d i ing ips P e-P ocessing Module 9&RUFREDHWDO$57(0,6$$UFKLWHFWXUHRIDQHFRGULYLQJDVVLVWDQWEDVHGRQWKHDQWLFLSDWLRQ  Loca ion Geog aphical Coo dina es a e ob ained h ough he GPS om And oid mobile de ice o iangula ion o an ennas. Eco-d i ing assis an uses he geog aphical coo dina es o de e mine he ehicle loca ion. Vehicle loca ion is used o ob ain he oad condi ions. In addi ion, i also used o sa e he loca ion o he de ec ed a ic signs. OBD2 Vehicle speed, RPM, uel consump ion and a el dis ance a e ob ained om ehicle diagnos ic po (OBD2). OBD po [Goda a y e al., 2000][OBD2 Adap e , 2012], was p oposed in 1984 and i is known as he s anda d OBDI. OBDI is s ong mind ocused on he assessmen o he emis- sion o gaseous pollu an s om he ehicle. In 1988, OBD was imp o ed and i was named as OBD2. OBD2 p o ides much mo e in o ma ion han OBDI because i s aim is no only o e alua e he emission o gas pollu an s, bu also o be able o do in-dep h diagnos ic abou he ope a ion o ehicle. This diagnos ic ehicle po (OBDII) is included in mos o oday's ehicles. To ob ain he ehicle diagnos ic alues, we connec ed a Blue oo h adap e o he OBDII po and he adap e sends da a o an And oid Mobile De ice. Figu e 2 shows he da a acquisi ion sys em. Figu e 2. Da a Acquisi ion Sys em 3 P e-P ocessing module This module is esponsible o il e ing and ex ac ing in- o ma ion om he da a collec ed by he da a acquisi ion sys em. OBD Fil e On some occasions, he da a ob ained h ough he OBD2 po a e unusual. Fo example, when he ehicle is s opped he a el dis ance alue supplied by he diagnos ic po is inco ec . This il e is esponsible o emo ing he alues ha exceed a h eshold. Wea he Fil e This il e is esponsible o ca ego izing wea he condi ions in o h ee classes: good, egula and bad. T a ic Fil e This il e is esponsible o ca ego izing he a ic densi y in h ee classes: smoo h, mode a e and hea y. Image In e pola ion We esize he cap u ed image 10x using cubic in e pola ion. Due o he limi ed p ocessing capabili ies o cu en mobile de ices, we esize only he igh hal o he image. We esize his egion o he image because in Spain he a ic signs a e loca ed on his side o he oad. Gene a ion o in o ma ion We can ob ain ele an in o ma ion abou he d i ing i we look a he da a collec ed by he da a acquisi ion sys em. Fo example, i we obse e ha he dis ance a eled emains cons an o e a pe iod, we can deduce ha he ehicle is s opped. I he ehicle is s opped o mo e han wo minu es, we could issue an Eco-d i ing ad ice as: You mus u n o he ehicle engine du ing p olonged s ops. 4 Eco-d i ing Ad ices This componen assesses compliance wi h he ollowing eco-d i ing ad ices: xVehicle Speed should no exceed 110 km / h xAccele a ions should no exceed 1.5 m / s ^ 2 xSlowdowns should no be less han -1.5 m / s ^ 2 xD i e mus u n o he ehicle engine when ehi- cle is s opped o mo e han wo minu es xThe d i e is d i ing a low gea 5 P edic ion Module The e a e nume ous s udies on eco-d i ing, which s a es ha he key o eco-d i ing is he an icipa ion. Ou sys em ana- lyzes he en i onmen and i p edic s whe he he d i e ac ions will cause an inc ease in uel consump ion o no . To achie e his goal, he sys em has wo componen s: T a ic Signs De ec ion Sha p accele a ions cause a conside able inc ease in he demand o ene gy, and he e o e, an inc ease in uel con- sump ion. On he o he hand, sha p slowdowns cause a g ea was e o ene gy. A la ge p opo ion o ab up decele a ions and unneces- sa y accele a ion a e due o he d i e dis ac ions. These dis ac ions make he d i e ails o comply wi h a ic signals ha equi e o may equi e s opping. The eco-d i ing assis an wa ns he use when he speed a which he o she ci cula es is no app op ia e due o he p oximi y o a a ic signal ha equi es o may equi e s opping. The "adequa e" speed is de ined as he one ha allows he ehicle o s op upon eaching he a ic signal OBD2 Cable Scan ool UTMS/LTE Blue oo h GPS Wea he Se ice T a ic Se ice Wea he XML T a ic XML Road Image 9&RUFREDHWDO$57(0,6$$UFKLWHFWXUHRIDQHFRGULYLQJDVVLVWDQWEDVHGRQWKHDQWLFLSDWLRQ  wi hou exceeding a h eshold in he slowdown. We ha e se he h eshold a 1.5 m/s2. G ea e decele a ions make uel consump ion inc ease exponen ially as we ha e obse ed in se e al es s. Ou eco-d i ing assis an de ec s h ee ypes o a ic signs: s op signals, yield and pedes ian c ossing. To de ec hese a ic signs, we use he me hod p oposed by Viola & Jones [18] ha uses a cascade o s ong classi ie s. This p oposal is he i s objec de ec ion algo i hm o p o ide compe i i e objec de ec ion a es in eal- ime. Viola & Jones p opose o de ec objec s based on he change o in ensi y. The me hod employs ea u es ha in- ol e he sums o image pixels wi hin ec angula a eas. A lea ning algo i hm, based on AdaBoos , selec s a small numbe o c i ical isual ea u es and builds a se o s ong classi ie s (cascade) using hese ea u es. The e iciency o his me hod is due o: •A new image ep esen a ion o e y as ea u e e alua ion. •A any s age i a classi ie ejec s he sub-window unde inspec ion, no u he p ocessing will be pe - o med and i will con inue on sea ching he nex sub-window. So in he ea ly s ages, many sub- windows ( he easies ) a e emo ed wi h e y li le p ocessing. Figu e 3. Schema o a ic signs sys em In Figu e 3, we can see a schema o he T a ic Signs De- ec ion Sys em. Ou p oposal uses as inpu a iables: ehi- cle speed, ehicle loca ion and in e pola ed image. The de ec ion module looks o a ic signs on he in- e pola ed image using he me hod o Viola & Jones. I ´s impo an o highligh ha he se o classi ie s is buil on a PC due o low p ocessing powe o he mobile phone. Clas- si ie s a e sa ed on he And oid mobile de ice as an xml ile. I he de ec ion module de ec s any a ic signal, i will sa e i s geog aphical coo dina es in a nAnd oid SQL da a- base. The objec i e is o de ec he a ic signs in ad ance because he sys em is only able o de ec signals up o 20 me e s dis an due o he limi a ions o he came as on mo- bile de ices. Then, he Nea es T a ic Sign Module ge s he dis ance o he closes a ic signal. Finally, he Speed Check Mod- ule is esponsible o checking i he cu en speed is app o- p ia e using he cu en ehicle speed and he dis ance o he nea es a ic signal. Op imal A e age Speed Module This module ob ains he op imal a e age speed o minimize uel consump ion and inc ease sa e y. Al hough, ehicle speed is no he only pa ame e in luencing in uel consump- ion, i is one o he mos decisi e because i in luences o he ac o s such as accele a ion, decele a ions and secu i y (con ol o e he ehicle). To ge he op imal speed, his module uses an algo i hm based on gene ic algo i hms (ASGA). ASGA de ines he p oblem as a combina o ial op imiza ion p oblem whe e he indi iduals a e ep esen ed as ec o s. Each posi ion o he ec o ep esen s a sec ion o he ip. In addi ion, he posi- ion o he ec o con ains an a e age speed alue and a el ime. Fo example, in Table 1, he 2-posi ion o he ec o indica es ha he ehicle mus un a 90 km / h and a el ime is 200 seconds. Table 1 Encoding o an indi idual o he ASGA Algo i hm Va iables S age 0 1 2 Speed 20 90 45 T a el ime 500 200 300 The algo i hm has as inpu pa ame e s: speed, a ic low, wea he condi ions, R.P.M and he numbe o es ima ed s ops. The i ness unc ion is de ined as: ܨ௡=ቌ൭൬ேೄ೟೚೛ೞכଵ଴଴଴ ௌ೏ೄ೟೚೛ೞ ൰כሺ௏ ଵ଴ሻ൱൅ሺሺܨ௪൅ܨ்ሻכܸሻቍ൅ܨ஼ Da a Acquisi ion Sys em T a ic Signs Recogni ion Module Nea es T a ic Signs Module Speed Check Module T a ic signs da abase P e-P ocessing Module Geog aphic Coo dina es, Image and Vehicle Speed Geog aphic Coo dina es, In e pola ed Image and Vehicle Speed T a ic Signal De ec ed, Geog aphic Coo dena es T a ic Signs Dis ance o he nea es a ic signal 9&RUFREDHWDO$57(0,6$$UFKLWHFWXUHRIDQHFRGULYLQJDVVLVWDQWEDVHGRQWKHDQWLFLSDWLRQ  whe e ܨ௖ is he es ima ion uel consump ion(L/100Km) gi en by Eqs.1, ܵௗ is he sec ion dis ance(me e s), ܰௌ௧௢௣௦ is he es ima ed numbe o s ops, ܨ௪ is a wea he ac o gi en by Eqs.2, ܨ் is a a ic densi y ac o gi en by Eqs. 3 and ݒ is he ehicle speed (Km/h). ܨ஼ൌܨ௠כሺሺܴܲܯாכܲ௠ሻȀ͹ͲʹͶሻݒ(1) whe e ܨ௠ is he uel consump ion measu ed using he es ECE-15 cycle. This cycle was in oduced by he EEC Di- ec i e 90/C81/01 in 1999,ݒ is he ehicle speed,ܲ௠ is he maximum o que alue (Nm) and ܴܲܯா is he es ima ed e olu ion pe minu es a ݒ ehicle speed. ܨ௪ቐͲ݂݅ݓ݁ܽݐ݄݁ݎൌ݃݋݋݀ ͲǤͲͷ݂݅ݓ݁ܽݐ݄݁ݎൌݎ݁݃ݑ݈ܽݎ ͲǤͳ݂݅ݓ݁ܽݐ݄݁ݎൌܾܽ݀ (2) ܨ்ቐͲ݂݅ݓ݁ܽݐ݄݁ݎൌ݈݋ݓ ͲǤͲͷ݂݅ݓ݁ܽݐ݄݁ݎൌ݉݋݀݁ݎܽݐ݁ ͲǤͳ݂݅ݓ݁ܽݐ݄݁ݎൌ݄݁ܽݒݕ (3) Figu e 4 shows a schema o he ASGA algo i hm. Ini ial Popula ion P1,P2,...PN Selec ion 1 (Bes Fi ness) P2,P12, P20, P35 Selec ion 2 (Bes T a el Time) P2,P12, P20, P35 C ossO e h1 = (P2+x35)/2 h2=(P12+P20)/2 Mu a ion h1 = (60+56/2) h2=h2 Figu e 4. ASGA Algo i hm 6 Use In e ace Dis ac ions o manipula e de ices such as GPS o mobile a e he cause o a la ge numbe o acciden s (Map e, 2006). The use in e ace module is in ended ha he eco-d i ing ad ices a e as leas in usi e as possible. Use in e ace module is esponsible o showing eco- d i ing ad ices wi h clea ypog aphic, and also, i con e s ex o oice. D i e does no ha e o look a he sc een. We use he ROBOTO on o show he Eco-d i ing ad- ices. This ypog aphy was in oduced on And oid 4.0 o imp o e isibili y on de ices wi h small sc een. To con e he ad ices o oice, we use he TTS lib a y ha suppo s And oid since e sion 2.1. 7 Conclusions In his pape , we ha e p esen ed he a chi ec u e o an eco- d i ing assis an de eloped inside he ARTEMISA p ojec . The p oposed eco-d i ing assis an e alua es eco-d i ing ules whose e ec i eness has been widely es ed. On he o he hand, he eco-d i ing assis an is di ec ly in luencing he d i e h ough he speed pa ame e . In addi ion, he use o he assis an inc eases sa e y be- cause he p oposed speed o he ehicle is sui ed and adap ed o he cu en condi ions o he oad. A la ge p o- po ion o acciden s o a ic a e due o an inadequa e speed The p oposed assis an uns on an And oid mobile de ice. Today's mobile de ices a e sui able o modeling he en i- onmen due o o hei mul iple ne wo k connec ions (Blue- oo h, Wi-Fi, UTMS) and senso s (GPS, Ligh Senso , Ac- cele a ions senso ). In addi ion, he cos o implemen ing he solu ion is low. We can ind And oid mobile de ices o less han $ 100. Fu he mo e, he p oposed a chi ec u e could be easily mo ed o ano he pla o m. As u u e wo k, we wan o imp o e he eco d i ing assis- an ca ying ou an exchange o in o ma ion be ween he ehicles on he oad. In his way, we can issue new eco- d i ing ad ices based on he an icipa ion. Fo example, i a ehicle is ci cula ing a an unusual speed and i issues his in o ma ion o ehicles ha ollow, hese can shape slow- downs mo e smoo hly. Acknowledgmen s The esea ch leading o hese esul s has ecei ed unding om he ARTEMISA p ojec TIN2009-14378-C02-02 wi h- in he Spanish "Plan Nacional de I+D+I", om he Eu ope- an Union's Se en h F amewo k P og amme managed by REA-Resea ch Execu i e Agency (FP7/2007-2013) unde g an ag eemen n° 286533 and om he Spanish unded HAUS IPT-2011-1049-430000 p ojec . Re e ences [Mu aki and Kanoh, 2008] “Wide-a ea a ic signal con ol using p edic ed a ic based on eal- ime in o ma ion,” P oceedings o he 11 h In e na ional IEEE Con e ence on In elligen T anspo a ion Sys ems, Beijing, China, Oc obe 12-15, 2008. [Ba bé and Boy, 2006], “On-boa d sys em design o op i- mize ene gy managemen ,” P oceedings o he Eu opean Annual Con e ence on Human Decision-Making and Manual Con ol (EAM'06), Valenciennes, F ance, Sep- embe 27-29, 200The Risk Obse a o y. Ins i u e o se- 9&RUFREDHWDO$57(0,6$$UFKLWHFWXUHRIDQHFRGULYLQJDVVLVWDQWEDVHGRQWKHDQWLFLSDWLRQ  cu i y s udies (IDES), «T a ic: Ci iliza ion o Ba ba- ism,» 2006. [Mie lo e al., 2004], «D i ing s yle and a ic measu es- in luence on ehicle emissions and uel consump ion,» Jou nal o Au omobile Enginee ing, 2004. [Koskinen, 2008], “Imp o ing ehicle uel economy and educing emissions by d i ing echnique,” P oceedings o he15 h ITS Wo ld Cong ess, New Yo k, No 15-20, 2008 S uc u e and In e p e a ion o Compu e P o- g ams. MIT P ess, Camb idge, Massachuse s, 1985. [E ik G. Lind eld e al., 2025], “S a egies o a oad anspo sys em based on enewable esou ces – The case o an impo -independen Sweden in 2025”, Ap- plied Ene gy, Volume 87, Issue 6, June 2010, Pages 1836-1845. [E icsson, 2001], “Independen d i ing pa e n ac o s and hei in luence on uel-use and exhaus emission ac o s” T anspo a ion Resea ch Pa D: T anspo , 2001 – Else ie , 325-345. [Johansson e al, 2003], «Impac o EcoD i ing on emis- sions» In e na ional Scien i ic Symposium on T anspo and Ai Pollu ion, A ignon, 2003. [Kuhle and Ka ens, 1978], “Imp o ed d i ing cycle o es ing au omo i e exhaus emissions”. SAE Technical Pape Se ies 780650. 1978. [Bo iboonsomsin e al, 2009], «Co Eco-D i ing: Pi-lo E alua ion o D i ing Beha io Changes among U.S. D i e s» 2009. [Klaue e al., 2006] The Impac o D i e Ina en ion On Nea -C ash / C ash Risk : An Analysis Using he 100- Ca Na u alis ic D i ing S udy Da a. Analysis, (Ap il), 224. Na ional Highway T a ic Sa e y Adminis a ion. [Riene e al., 2010], "Subliminal ib o- ac ile based no i i- ca ion o CO2 economy while d i ing", 2nd In e na ion- al Con e ence on Au omo i e Use In e aces and In e - ac i e Vehicula Applica ions Au omo i eUI 2010), No- embe 11-12, 2010, Pi sbu gh, Pennsyl ania, USA, ACM, ISBN: 978-1-4503-0437-5, pp. 92-101, 2010. [Saboohi and Fa zaneh, 2009], «Model o de eloping an eco-d i ing s a egy o a passenge ehicle based on he leas uel consump ion,» Applied Ene gy, ol. 86, nº 10, pp. 1925-1932, 2009. [Casa ola e al., 2010], "E icien gea shi ing s a egies o g een d i ing policies," Ame ican Con ol Con e ence (ACC), 2010, ol., no., pp.4331-4336, June 30 2010-July 2 2010. [Ba h and Bo iboonsomsin, 2009] Ene gy and emissions impac s o a eeway-based dynamic eco-d i ing sys em, T anspo a ion Resea ch Pa D: T anspo and En i- onmen , Volume 14, Issue 6, Augus 2009, Pages 400- 410, ISSN 1361-9209, 10.1016/j. d.2009.01.004. [Ke e al., 2010], ”Fuel-e icien d i ing s a egies o high- way ehicles”, in P oc. Indus ial Elec onics and Appli- ca ions, pp.2070 - 2075, 2010. [A emisa P oyec , 2012]. URL: h p://www.gas .i .uc3m.es/a emisa/index.php/Po ada, [Las access: May 2012]. [DGT, 2012], URL: h p://in oca .dg .es/e a ic/ ss_ odas.xml. [Las access: May 2012]. [AEMET, 2012] URL: h p://www.aeme .es/es/po ada. [Las access: May 2012]. [OpenCV p ojec , 2012], “The OpenCV documen a ion”. URL: h p://openc .willowga age.com/wiki/. [Las ac- cess: May 2012]. [Goda a y e al., 2000], “In e acing o he on-boa d diag- nos ic sys em”. Vehicula Technology Con e ence, 2000. [OBD2 Adap e , 2012], URL: h p://www.scan ool.ne /. [Las access: May 2012]. 9&RUFREDHWDO$57(0,6$$UFKLWHFWXUHRIDQHFRGULYLQJDVVLVWDQWEDVHGRQWKHDQWLFLSDWLRQ 