Traffic responsive mode of operation with its two mechanisms, threshold-based and pattern matching, is considered one of the effective and efficient signal control modes. This operation mode is underutilized due to its cumbersome configuration procedure. The research presented in this thesis aims to give some guidelines regarding traffic responsive and issues that might improve the system performance. Four different issues related to traffic responsive are considered: The first issue is the generation of different traffic scenarios that drive the design of the system. This point is not limited to traffic responsive only but it is more general for different traffic engineering applications that need different traffic scenarios. The second issue is presenting an approach to implement traffic responsive control mode of operation in a large arterial network in Northern Virginia. Pattern matching mechanism is used for this application. Compared to time-of-day control mode, traffic responsive control saves up to 26.94% of the average delay and 21.13% of average number of stops for Reston Parkway network. The third issue is an attempt to improve the current threshold mechanism by relaxing the threshold constraints and using variable thresholds for different levels of plan selection parameters. The last issue is a study for the pedestrian effect on the performance of networks operating by traffic responsive control. The effects of pedestrian calls and pedestrian phases on traffic responsive control are compared and the results shows that pedestrian calls are better for low pedestrian volumes while pedestrian phases are better for high pedestrian volumes. / Master of Science

Modern camera-based vehicle safety systems heavily rely on machine learning and consequently require large amounts of training data to perform reliably. However, collecting and annotating the needed data is an extremely expensive and time-consuming process. In addition, it is exceptionally difficult to collect data that covers critical scenarios. This thesis investigates to what extent synthetic data can replace real-world data for these scenarios. Since only a limited amount of data consisting of such real-world scenarios is available, this thesis instead makes use of proxy scenarios, e.g. situations when pedestrians are located closely in front of the vehicle (for example at a crosswalk). The presented approach involves training a detector on real-world data where all samples of these proxy scenarios have been removed and compare it to other detectors trained on data where the removed samples have been replaced with various degrees of synthetic data. A method for generating and automatically and accurately annotating synthetic data, using features in the CARLA simulator, is presented. Also, the domain gap between the synthetic and real-world data is analyzed and methods in domain adaptation and data augmentation are reviewed. The presented experiments show that aligning statistical properties between the synthetic and real-world datasets distinctly mitigates the domain gap. There are also clear indications that synthetic data can help detect pedestrians in critical traffic situations / <p>Examensarbetet är utfört vid Institutionen för teknik och naturvetenskap (ITN) vid Tekniska fakulteten, Linköpings universitet</p>

Pendant les deux dernières décennies, les technologies de réseaux ad-hoc de véhicules (VANETs : Vehicular Ad-Hoc Networks) ont été développées sous l’impulsion du monde de la recherche comme de l’industrie, étant donnés les liens des VANETs avec la sécurité routière, l’internet des objets (IoT/WoT : Internet of Things/Web of Things) pour les systèmes de transport intelligents (ITS : Intelligent Transportation Systems), les villes intelligentes et les villes vertes. Composant essentiel des VANETs, les protocoles de communication inter-véhicules (IVC : Inter-Vehicle Communication) font face à des défis techniques, en particulier à cause de la diversité des applications dans lesquelles ils sont impliqués. Dans cette thèse, après une présentation des VANETs et de l’état de l’art des IVC, nous proposons un protocole de dissémination de données, TrAD, conçu pour diffuser de manière efficiente des messages d’une source vers les véhicules présents dans la zone d’intérêt (ROI : Range of Interest). TrAD se base sur les états du trafic routier et du trafic réseau pour adapter localement la stratégie et les paramètres de transmission des données afin d’optimiser les performances des applications qui l’utilisent. De plus, un algorithme de classification des clusters locaux de véhicules est conçu pour permettre l’usage de TrAD sur autoroute aussi bien qu’en ville. Pour éviter l’encombrement des canaux de communication, un mécanisme illustratif de contrôle de la congestion reposant sur une approche distribuée est utilisé. Trois protocoles IVC de l’état de l’art ont été comparés à TrAD dans des scénarios réalistes de simulation, basés sur différentes villes réelles, différents trajets et densités véhiculaires. Les performances de TrAD surpassent celles des protocoles de référence en termes de taux de délivrance des paquets (PDR : Packet Delivery Ratio), nombre de transmissions et latence. De plus, nous montrons que TrAD est tolérant, dans une certaine mesure, aux erreurs sur les données GPS. Pour s’assurer de la qualité des simulations, nous avons étudié le modèle de déplacement employé dans le simulateur de trafic, puis couplé ce dernier au simulateur de réseau, afin que les deux s’échangent des informations en temps-réel. Grâce à la compréhension acquise lors de l’analyse du modèle de déplacement, nous avons pu développer un simulateur de conduite de tramway pour la T2C (Transports en Commun de l’agglomération Clermontoise). Des tests menés sur le matériel roulant nous ont permis d’élaborer des modèles de déplacement fidèles correspondants aux diverses situations rencontrées par le tramway. L’affichage de la simulation est assuré par un flux vidéo ajusté plutôt que des images de synthèse, ce qui permet de limiter le coût de développement tout en garantissant un certain réalisme dans l’affichage. Ce projet est soutenu par la T2C pour une durée de deux ans. / In the last two decades, Vehicular Ad hoc Network (VANETs) were developed significantly by both academic institute and industries association, since VANETs originate from traffic safety and are also an important application of Internet of Things / Web of Things (IoT/WoT) for Intelligent Transportation System (ITS), Intelligent Vehicles and Smart Cities. As an essential component of VANETs, Inter-Vehicle Communication (IVC) protocols face many critical challenges, in particular, because they relate to various specific applications. In this thesis, after elaborating on related knowledge of VANETs and state-of-the-art of IVC protocols, we propose a data dissemination protocol for vehicular networking, named TrAD, to disseminate efficiently warning messages from a source to vehicles in a range of interest (ROI). TrAD considers the status of road traffic and network traffic to adapt locally the strategy and the parameters of transmissions in order to optimize the global performance of IVC application. Moreover, a local vehicular cluster classification algorithm is designed to support TrAD to be performed in both highway and urban scenarios. In addition, an illustrative congestion control mechanism is used to avoid channel congestion using a distributed approach. Three state-of-the-art IVC protocols have been compared with TrAD by means of realistic simulations. The performance of all those protocols is evaluated quantitatively in various scenarios by taking into account different real road maps, trafic routes and vehicular densities. Compared with the reference protocols, TrAD gains an outstanding overall performance in terms of packet delivery ratio, number of transmissions and delay. Furthermore, TrAD also can tolerate a reasonable degree of GPS drift while achieving efficient data dissemination. In order to ensure the quality of simulations, we deeply investigated the mobility model of road traffic simulator, and then performed the bidirectionally coupled simulation in which the network simulator and the road trafic simulator can exchange information in real-time. Upon understanding of the mobility model, we obtained a chance to develop a low-cost tram simulator for the local public transportation provider, the T2C (Transports en Commun de l’agglomération Clermontoise). We attempt to design accurate mobility models from different scenarios for the specific type of tram used by T2C. Real world trials are carried out to explore the key parameters required by theoretical deduction for our mobility model. Moreover, the display GUI relies on a video stream, rather than 3D graphics, which can reduce the cost while guaranteeing the quality of service. This project was supported for two years by T2C.