Submitted by Maria Aparecida (cidazen@gmail.com) on 2017-04-24T13:52:10Z No. of bitstreams: 1 Julio Cesar da Silva Soares.pdf: 1645821 bytes, checksum: e32d7384c0a6f1999bc7eb190dcd7a05 (MD5) / Made available in DSpace on 2017-04-24T13:52:10Z (GMT). No. of bitstreams: 1 Julio Cesar da Silva Soares.pdf: 1645821 bytes, checksum: e32d7384c0a6f1999bc7eb190dcd7a05 (MD5) Previous issue date: 2017-03-22 / FAPEMA / Urban roads are very complex. The increase in the flow of vehicles in the cities has contributed to traffic accidents. Researches for accident reduction show that the traffic lights are effective in reducing accidents. Traffic lights can minimize the occurrence of accidents at intersections and crosswalks. The implementation of traffic light signals shows significant advantages, otherwise reveals some problems such as the failure to detect road signs by drivers on urban roads. This fact is related to excessive visual information, the stress of the drivers and/or eyestrain makes the drivers lose their attention. These reasons motivated researches about intelligent vehicles. This work aims to develop a methodology to detect and recognize traffic lights, to be applied in smart vehicles. This methodology can contribute to the Advanced Driver Support Systems (ADAS), which assists drivers, especially those with partial vision impairment. Image processing techniques are used to develop the detection methodology. Back project and global thresholding are combined to find light points. Local thresholding techniques are applied to calculate the symmetry between the radius and the center of the light points to segment the traffic light body. The first step got an average rate of 99% of detection. The features of the traffic lights are extracted using Haralick texture measures, with the inclusion of color and shape information. The data generated by the feature extraction step were preprocessed using the SMOTE technique to balance the database. The recognition and identification of the traffic lights state are made by an artificial neural network using Multilayer-Perceptron (MLP). The backpropagation learning algorithm are used in the network training. The validation results show an average recognition rate of 98%. / As vias urbanas estão cada vez mais complexas e o acréscimo no fluxo de veículos nas cidades de médio e grande porte vem contribuindo para a elevação do número de acidentes. Pesquisas apontam que os sinais de trânsito são eficientes na redução do número de acidentes. A implantação de sinais de trânsito apresentam vantagens relevantes, mas por outro lado revelam alguns problemas, como a dificuldade na detecção de sinais de trânsito pelos condutores em vias urbanas. Este fato está relacionado à quantidade de informações visuais nas vias, ao estresse dos motoristas e/ou à fadiga visual destes, que fazem os motoristas desviarem sua atenção da sinalização. Estas razões motivaram muitas pesquisas nos últimos anos, sobre o tema veículos inteligentes. Assim, o presente trabalho propõe uma metodologia para detectar e reconhecer semáforos de trânsito para ser aplicada em veículos inteligentes, podendo contribuir para os Advanced Driver Support Systems - ADAS (Sistema Avançado de Auxílio ao Motorista), e que auxilie os motoristas, em especial aqueles com deficiência parcial da visão. Além disso, o sistema desenvolvido é capaz de identificar o estado do semáforo e indicar ao condutor se ele deve parar ou prosseguir, contribuindo assim para a redução de acidentes de transito. Para o desenvolvimento do algoritmo de detecção, utilizaram-se técnicas de processamento de imagens, através de histograma retroprojetado e limiarização global para detectar pontos de luzes. A limiarização local é aplicada para o cálculo de simetria entre o raio e o centro dos pontos de luzes, com a finalidade de segmentar o corpo do semáforo, onde se obteve uma taxa média de detecção de 99%. As características dos semáforos foram extraídas utilizando os atributos de Haralick, com a inclusão de informações de cor e forma. Os dados gerados pela extração de características foram pré-processados utilizando a técnica de SMOTE para balancear a base de dados. O reconhecimento e a identificação do estado do semáforo foram realizados por uma rede neural artificial do tipo Multilayer Perceptron (MLP). No treinamento da rede se utilizou o algoritmo de aprendizagem backpropagation e a separação de dados para treinamento e validação. Os resultados da validação mostraram uma taxa média de reconhecimento de 98%.

In this thesis, detection of traffic signs is studied. Since, both color and shape properties of traffic signs are distinctive / these two properties have been employed for detection. Detection using color properties is studied in two different color domains in order to examine and compare the advantages and the disadvantages of these domains for detection purposes. In addition to their color information, shape information is also employed for detection purpose. Edge information (obtained by using the Sobel Operator) of the images/frames is considered as search domain to find triangular, rectangular, octagonal and circular traffic signs. In order to improve the performance of detection process a joint implementation of shape and color based algorithms is utilized. Two different methods have been used v in order to combine these two features. Both of the algorithms help reducing the number of pixels to check whether they belong to a sign or not. This, of course, reduces the processing time of detection process. Each utilized algorithm is tested and compared with the others by using both static images from different sources and video streams. Images having adverse properties are used in order to state algorithms response for some specific conditions such as bad illumination and shadow. After implementation, results show that joint implementation of the color and shape based detection algorithms produces more accurate results. Moreover, joint implementation reduces the processing time of the detection process when compared to application of algorithms individually since it diminishes the search domain.

Μία από τις πιο γοργά αναπτυσσόμενες περιοχές της επιστήμης των υπολογιστών είναι η ανάπτυξη έξυπνων συστημάτων που υποστηρίζουν τις αποφάσεις των χρηστών και παρέχουν ένα ευρύ πεδίο υπηρεσιών. Η εξυπνάδα τους βασίζεται στην παρακολούθηση και αποκωδικοποίηση των αναγκών και την προσομοίωση της συμπεριφοράς του χρήστη. Το αντικείμενο της παρούσας διατριβής είναι η παρουσίαση ενός νέου, αξιόπιστου συστήματος δυναμικής αξιολόγησης της συμπεριφοράς του οδηγού και υποστήριξης αυτού σε πραγματικό χρόνο. Πιο συγκεκριμένα, δίνεται έμφαση στην ανοικτή, προσανατολισμένη προς τις υπηρεσίες (service-oriented) αρχιτεκτονική του, στους κανόνες που το διέπουν και στο υλικό και το λογισμικό που του επιτρέπουν να παρέχει διαλειτουργικές υπηρεσίες. Εφαρμόζεται η συστημική προσέγγιση που αρχίζει με τα στοιχεία εισόδου. Τα στοιχεία αυτά αφορούν βασικά τον οδηγό: προσωπικά στοιχεία, προφίλ, καλούς χειρισμούς κ.ά., το αυτοκίνητο: ταχύτητα, επιτάχυνση, επιβράδυνση, γωνία τιμονιού, αριθμός στροφών κινητήρα, σχέση μετάδοσης στο κιβώτιο ταχυτήτων, μοντέλο και τύπος οχήματος και το περιβάλλον: GPS, RFID, κάμερες, αισθητήρες, ασύρματες και δορυφορικές επικοινωνίες κ.ά. Συνεχίζουμε με την ευφυή - αλγοριθμική, στατιστική κ.λπ. - επεξεργασία αυτών των στοιχείων (α) για να εκτιμήσουμε την τρέχουσα κατάσταση του οδηγού και του οχήματος στις συγκεκριμένες περιβαλλοντικές συνθήκες και (β) για να κατανοήσουμε τη συμπεριφορά και να υποστηρίξουμε παθητικά ή ενεργά τον οδηγό κατά τη διάρκεια ενός ταξιδιού. Παράγουμε πρωτότυπα αποτελέσματα, δηλαδή χρήσιμη πληροφορία και πιθανές συμβουλές προς τον οδηγό του οχήματος, στοιχεία για την συμπεριφορά του οδηγού με σκοπό την περαιτέρω χρήση αυτών των πληροφοριών από άλλους φορείς, όπως ασφαλιστικές, εταιρείες, ελεγκτικά όργανα του κράτους κ.λπ. Βέβαια, οι δυνατότητες που προσφέρονται από το προτεινόμενο σύστημα μπορούν να οδηγήσουν σε οφέλη και για τις ασφαλιστικές εταιρείες οι οποίες καλούνται να εκσυγχρονίζουν συνεχώς τον τρόπο με τον οποίο καθορίζουν το ύψος των ασφαλίστρων. Επιπλέον, το προτεινόμενο σύστημα θα μπορούσε να χρησιμοποιηθεί από εταιρείες που διαθέτουν στόλο οχημάτων, προκειμένου να επαληθεύουν και να ελέγχουν την ικανότητα των οδηγών σε πραγματικό χρόνο. Τέλος, το προτεινόμενο σύστημα θα μπορούσε να χρησιμοποιηθεί από το υπουργείο μεταφορών, την τροχαία, τους φορείς τοπικής διοίκησης κ.ά. / One of the fastest growing areas of computer science is the development of intelligent systems that support user decisions and provide a wide range of services. Their intelligence is based on monitoring and decoding of real needs, as well as the simulation of end user’s behavior. The object of this Thesis is the presentation of a new, integrated system for dynamic evaluation of driver behavior. In particular, we emphasize at its open, service-oriented architecture, the incorporated set of rules and the system hardware and software which allow it to provide interoperability. We apply the systematic approach that begins with the input data plus requirements. These data mainly concern the driver: personal data, profile, good practices etc., the vehicle: speed, acceleration, deceleration, steering angle, engine speed, gear ratio in gearbox, model and type, and the environment: GPS, RFID, cameras, wireless and satellite communications, etc. Then, these data are processed applying an intelligent-algorithmic, statistical etc.- approach in order (a) to evaluate the current state of the driver and the car in certain environmental conditions, and (b) to understand the behavior and passively or actively support the driver during a travel. We produce original results, i.e. useful information and possible recommendations to the driver of the vehicle, data concerning the driver behavior and thus, this information can be further used by others, such as insurance companies, audit institutions of state etc. We place great emphasis on cutting-edge technologies that are applied to achieve the required feedback, parameterization and adaptation of the system. Of course, the capabilities offered by the proposed system can lead to clear benefits for various organizations, like insurance companies, which are required to continually update their price policy, companies that have a fleet of vehicles, in order to verify the ability of the drivers and support them in real time. Finally, the proposed system could be used by the Ministry of Transport, the traffic police, the local authorities etc.

A human-centered approach when developing new support systems in vehicles has the potential to enable the driver to make safe decisions in the transition between manual and automatic control. However, careful considerations have to be taken. Not only would the design of the systems, in terms of interface be important, but also what kind of activities the systems support. The aim of this study was to identify an appropriate activity to support the cognitive processes for truck drivers, develop an interface for this activity, and evaluate it in driving situations. This was executed in three sub-studies: the Pre-study, the Design-study, and the Evaluation study. In the Pre-study, the aim was to investigate for what kind of driver-related activity distribution and long haulage truck drivers need a driver support and interface. This was investigated via contribution from truck drivers, HMI/Ergonomics experts, as well as engineers. The activity chosen to support was detecting objects around the vehicle. However, reconsiderations were made due to constrains in the simulator. Suggested by Scania’s Vehicle Ergonomics group a holistic system was chosen; an interface approach enabling for more technologies to be included within the same interface, reducing the amount of modalities a driver can be exposed to. The Design-study addressed the aim of designing an interface for the Holistic system with truck drivers’ cognitive workload in focus. A LED-prototype was built running along the window edges inside the cab of Shania’s Vehicle Ergonomics groups’ simulator, to create warning signal concepts. Literature findings, the LED-prototype, and the simulator were used in an iterative process to design and improve warning signal concepts, until two final concepts were created. The holistic system informs of hazards around and near the vehicle by lighting the area risky objects occurs to guide drivers’ attention and this was done either with 1) the informative display or, 2) the directional display. The Informative display conveys information of a hazard location and type, and the Directional display exclusively conveys information of the hazard location. The Evaluation study explored how drivers were affected by, and how they perceived, the holistic interface design regarding mental workload and hazard detection. A user simulator test was designed to collect data within the areas of ‘Event detection’, ‘Workload’, ‘Driving performance’ and ‘Subjective opinion’. Fourteen professional truck drivers assessed three conditions: 1) Baseline (driving without a system), 2) the Informative display, and, 3) the Directional display, while being exposed to potential hazards. To further increase workload, a secondary task was performed at the end of each condition. The results showed that the Informative display did not only result in more ‘Detection hits’, instances when a driver responded to a present hazard, but also significantly decreased reaction time to detect a hazard. However, in terms of acceptance, the two concepts were considered equally preferred. As the Informative display showed to be more efficient in terms of hazard detection, this should be investigated further. A holistic interface enables for more systems to be included within the same interface, reducing the amount of alarms and modalities drivers are exposed to if designed skillfully. Thus, more support systems can be included in future vehicles, without causing unnecessary distraction when applying a holistic interface approach. / Ett människocentrerat förhållningsätt vid utveckling av nya stödsystem i fordon möjliggör för förare att ta säkra beslut i övergången mellan manuell kontroll och automation. Men noggranna överväganden måste tas. Inte bara systemets utförande i form av gränssnittet är av stor vikt, utan även vilken typ av aktivitet som stöds. Syftet med denna studie var att identifiera en lämplig aktivitet att stödja lastbilsförares kognitiva processer, utveckla ett gränssitt för denna aktivitet och utvärdera gränssnittet i en körsituation. Detta utfördes i tre substudier: Förstudien, Designstudien samt Utvärderingsstudien. Förstudiens syfte var att undersöka för vilken typ av körrelaterad aktivitet distributions- och långtransportförare behövde ett förarstöd och gränssitt. Detta undersöktes med bidrag från lastbilsförare, HMI/Ergonomi experter samt ingenjörer. Den valda aktiviteten blev upptäcka objekt framför och kring lastbilen. Dock ändrades den valda aktiviteten på grund av begräsningar i simulatorn. Förslaget från Scanias Ergonomigrupp för förarhytten blev ett Holistiskt system istället; en gränssnittsstrategi som möjliggör att fler tekniker och system att inkluderas i samma gränssnitt, vilket minskar antalet modaliteter en förare kan bli utsatt för.  Designstudien behandlar syftet beträffande utformningen av gränssnittet för det holistiska systemet med avseende på lastbilsförares kognitiva belastning. En LED-prototyp byggdes, denna löpte längs med fönsterkanten i förarhytten på Scanias Ergonomigrupps simulator, för att skapa varningssignals-koncept. Resultat från litertur, LED-prototypen och simulatorn användes i en iterativ process för att utveckla och förbättra varningssignalerna. Det holistiska systemet informerar om faror runt fordonet genom att tända ljus i det område riskfyllda objekt upptäckts för att leda förarens uppmärksamhet och detta görs med något av de två utvecklade koncepten: 1) det informativa varningskonceptet eller 2) det riktningsgivande konceptet. Det informativa konceptet förmedlar information om farans placering och typ, medan det riktningsgivande varningskonceptet enbart förmedlar information om farans placering. Utvärderingsstudien utforskade hur förare påverkades av och hur de upplevde det holistiska gränssnittet med avseende på mentalbelastning och upptäckten av faror. Ett användartest i en simulatorutvecklades för att samla in data inom områdena Upptäckt av faror, Mentalbelastning, Körförmåga samt Subjektiv uppfattning. Fjorton professionella förare bedömde tre tillstånd: 1) Baslinje (körning utan ett system), 2) det informativa varningskonceptet och 3) det riktningsgivande varningskonceptet, medan de blev utsatta för potentiella faror. För att öka den mentala belastningen utfördes en sekundäruppgift vid slutet av varje tillstånd. Resultaten visade att det Informativa varningskonceptet inte enbart resulterade i fler upptäckta faror, tillfällen då förare reagerade på en närvarande fara, utan även signifikant minskade reaktionstider att upptäcka faror. Däremot föredrogs båda koncepten i samma utsträckning med avseende på acceptans. Då det informativa varningskonceptet visades sig mest effektivt gällande upptäckten av faror borde denna undersökas vidare. Ett holistiskt gränssitt möjliggör för fler system att inkluderas i samma gränssitt och minskar mängden alarm och modaliteter som en förare kan utsättas för om det designas skickligt. Om ett holistiskt gränssnitt tillämpas kan därmed fler stödsystem innefattas i framtida fordon utan att orsaka oönskad distraktion.