This thesis is concerned with the problem of situational awareness in autonomous vehicles. In this context, situational awareness refers to the ability of an autonomous vehicle to perceive the road layout ahead, interpret the implied semantics and gain an awareness of its surrounding - thus reading the road ahead. Autonomous vehicles require a high level of situational awareness in order to operate safely and efficiently in real-world dynamic environments. A system is therefore needed that is able to model the expected road layout in terms of semantics, both under normal and roadwork conditions. This thesis takes a three-pronged approach to this problem: Firstly, we consider reading the road surface. This is formulated in terms of probabilistic road marking classification and interpretation. We then derive the road boundaries using only a 2D laser and algorithms based on geometric priors from Highway Traffic Engineering principles. Secondly, we consider reading the road scene. Here, we formulate a roadwork scene recognition framework based on opponent colour vision in humans. Finally, we provide a data representation for situational awareness that unifies reading the road surface and reading the road scene. This thesis therefore frames situational awareness in autonomous vehicles in terms of both static and dynamic road semantics - and detailed formulations and algorithms are discussed. We test our algorithms on several benchmarking datasets collected using our autonomous vehicle on both rural and urban roads. The results illustrate that our road boundary estimation, road marking classification, and roadwork scene recognition frameworks allow autonomous vehicles to truly and meaningfully read the semantics of the road ahead, thus gaining a valuable sense of situational awareness even at challenging layouts, roadwork sites, and along unknown roadways.

Computerised regulation for disturbed traffic in metro-type railways is proposed. Previous work has used optimisation techniques to minimise disbenefits to passengers, such as waiting time and journey time, in the objective function. The particular disbenefit of trains being halted between stations is introduced in this thesis, in combination with those already mentioned. An effective method in real operations for preventing trains being halted between stations is to hold trains already at stations and to allow running trains to reach the next station when a particular train departure is delayed. The proposed algorithm uses this ‘stop-all-trains-at-once’ philosophy combined with optimisation ideas, in a sequentially structured approach. A further consideration from real operations is the fact that it is not possible to predict precisely when the delayed train will re-start. Estimates of the re-starting time will improve as the delay increases, and the proposed scheme takes this into account. Numerous simulations were undertaken to investigate the performance of the regulation algorithm. It is shown that the proposed regulation algorithm is effective in reducing the disbenefit to passengers from disturbed traffic for various characteristic metros with different passenger flows.

A mathematical model is developed which predicts if a person or a pushchair is destabilised by a train's slipstream. The model simulates the mean slipstream velocity time history using the theories of potential flow, boundary layer growth and wake decay. The turbulence-induced fluctuations are reproduced with an autoregressive model. A randomised person is generated and subjected to the simulated slipstream, and their response is modelled by a simple solid object and a mass-spring-damper system. If the slipstream forces cause the person to be displaced by a critical distance the person is destabilised. A randomised pushchair is also generated and positioned so as to be capable of being destabilised by either toppling over or moving along the ground on its wheels. A toppling pushchair is modelled as a simple solid object in a similar manner to that of a person, and a pushchair will move along the ground on its wheels if the slipstream force is greater than the frictional force. Greater numbers of destabilised people and pushchairs are associated with the slipstream of a freightliner than that of a passenger train, increasing train speed and decreasing distance from the train side.

The oxidation resistance of four prototype single crystal nickel-based superalloys for industrial gas turbine applications is studied. All contain greater quantities of Cr than in most existing single crystal superalloys; two are alloyed with Si, one with Re. To explain the results, the factors known to influence the rate of Al\(_2\)O\(_3\) scale formation are considered. Models are developed to predict whether any given alloy composition will form a continuous Al\(_2\)O\(_3\) scale. These are used to rationalise the dependence of Al\(_2\)O\(_3\) scale formation on alloy composition in these systems. The mechanical behaviour of a new single crystal nickel-based superalloy for industrial gas turbine applications is also studied under creep and out-of-phase thermomechanical fatigue (TMF) conditions. Neutron diffraction methods and thermodynamic modelling are used to quantify the variation of the gamma prime (\(\gamma\)’) strengthening phase around the \( \gamma\)’ solvus temperature; these aid the design of primary ageing heat treatments to develop either uniform or bimodal microstructures of the \( \gamma\)’ phase. During TMF, localised shear banding occurs with the \( \gamma\)’ phase penetrated by dislocations; however during creep the dislocation activity is restricted to the matrix phase. The factors controlling TMF resistance are rationalised.

L’essor des transports à une échelle globale s’inscrit dans un phénomène de mondialisation et 90% des biens échangés au travers le monde sont effectués par voie maritime. La maritimisation se définit comme un processus de dépendance économique des états au trafic maritime due à ce phénomène de mondialisation (Vigarié 1983). Ce phénomène transforme les territoires, et principalement les espaces côtiers mais a également comme conséquence une territorialisation des espaces maritimes, à savoir une utilisation et gestion des espaces pour les activités humaines. L'utilisation de cet espace a augmenté depuis 1970 et principalement le transport, que ce soit via porte-conteneurs ou tanker (Rodrigue, Comtois, and Slack 2013). Cela forme ainsi un réseau de transport maritime et d’approvisionnement lié à l’emplacement des ports dans le monde et des routes maritimes les reliant. Cependant ce réseau s’étend sur un espace vaste et dynamique, l’espace maritime, qui peut générer des risques pour l’usage de l’espace à des fins de transports.Ce réseau d’approvisionnement est vulnérable face à certains événements (perturbations) pouvant affecter la performance du réseau d’approvisionnement (Achurra-Gonzalez et al. 2016) : les tempêtes, les pénuries, la piraterie maritime, les fermetures de canaux sont des événements ayant une influence sur l’efficacité d’un réseau d’approvisionnement. Cette vulnérabilité est liée à plusieurs facteurs (Wagner and Neshat 2010) : les variations de l’offre et de la demande, les risques économiques, et à la structure du réseau et de l’espace maritime, les risques géographiques. L’ensemble de ces risques peuvent être mesurés et quantifiés dans les caractéristiques topologiques, géométriques et relationnelles des graphes. Un graphe est un ensemble de noeuds relié par un ensemble de lien. Les graphes sont utilisés pour la formalisation des réseaux, notamment des réseaux de transports (Ducruet and Lugo 2013).Afin de tester les perturbations sur les caractéristiques du réseau, un système de simulation est mis en place. Pour cela, les perturbations sont modélisées par des objets spatio-temporels ayant une influence sur l’accessibilité à l’espace des navires. Au travers d’un système multi-agents (Ferber 1997), ces navires prennent des décisions face aux perturbations (changement de route, de destination, report ou annulation du voyage). Ces comportements individuels, une fois agrégés, par la pondération des noeuds et des liens du graphe, permettent d’évaluer les variations topologiques et géométriques du graphe en fonction des perturbations dans l’espace maritime, grâce à des indicateurs utilisés pour mesurer la vulnérabilité des réseaux (Gleyze 2005). Ainsi cette recherche permet de mieux comprendre l’influence de ces perturbations sur le fonctionnement d’un réseau d’approvisionnement dans un espace maritime. / Transportation on a global scale is due to the globalization process. Today, 90% of goods in the world are traded by sea. Maritimization is defined as a process of increased economic dependence of states on maritime traffic (Vigarié 1983). This phenomenon transforms territories, mainly the coastal areas, but also results in a territorialization of maritime spaces, which means a use and a management of maritime space for human activities. The use of this space for transportation has increased since 1970, whether through container ships or tanker (Rodrigue, Comtois, and Slack 2013). These trips have created a network which linked the locations of the ports in the world by the sea routes of shipping. However, this network extends over a vast and dynamic space, maritime space, which can generate risks for the use of space for transportation.This supply network is vulnerable to certain events (disruptions) that may affect the performance of the supply network (Achurra-Gonzalez et al., 2016): storms, shortages or maritime piracy have an influence on the efficiency of a supply network. This vulnerability is linked to several factors (Wagner and Neshat 2010): variations in supply and demand, which are economical risks, and the structure of the network and maritime space, which are geographical risks. All these can be measured and quantified by the topological, geometric and relational properties of graphs. A graph is a set of nodes connected by a link set. The graphs are used for the formalization of networks, notably transport networks (Ducruet and Lugo 2013).To test the disruptions on the characteristics of the network, a simulation system is used. In the simulations, disruptions are modeled by space-time objects having an influence on the accessibility of the vessels. Through a multi-agent system (Ferber 1997), these ships make decisions about disruptions (change of route, destination, postponement or cancellation of the trip). These individual behaviors, once aggregated by the weighting of the nodes and links of the graph, allow evaluating the topological and geometric variations which represent the influence of disruptions on the martitime network. Theses variations can be measured by indicators of the vulnerability of network (Gleyze 2005). This research allows us to improve the understanding on the influence of these disruptions on a network within a dynamic space.

Travel time is one of the most important inputs in many Intelligent Transport Systems (ITS). As a result, this information needs to be accurate and dynamic in both spatial and temporal dimensions. For the estimation of travel time, data from fixed sensors such as Inductive Loop Detectors (ILD) and cameras have been widely used since the 1960 s. However, data from fixed sensors may not be sufficiently reliable to estimate travel time due to a combination of limited coverage and low quality data resulting from the high cost of implementing and operating these systems. Such issues are particularly critical in the context of Less Developed Countries, where traffic levels and associated problems are increasing even more rapidly than in Europe and North America, and where there are no pre-existing traffic monitoring systems in place. As a consequence, recent developments have focused on utilising moving sensors (i.e. probe vehicles and/or people equipped with GPS: for instance, navigation and route guidance devices, mobile phones and smartphones) to provide accurate speed, positioning and timing data to estimate travel time. However, data from GPS also have errors, especially for positioning fixes in urban areas. Therefore, map-matching techniques are generally applied to match raw positioning data onto the correct road segments so as to reliably estimate link travel time. This is challenging because most current map-matching methods are suitable for high frequency GPS positioning data (e.g. data with 1 second interval) and may not be appropriate for low frequency data (e.g. data with 30 or 60 second intervals). Yet, many moving sensors only retain low frequency data so as to reduce the cost of data storage and transmission. The accuracy of travel time estimation using data from moving sensors also depends on a range of other factors, for instance vehicle fleet sample size (i.e. proportion of vehicles equipped with GPS); coverage of links (i.e. proportion of links on which GPS-equipped vehicles travel); GPS data sampling frequency (e.g. 3, 6, 30, 60 seconds) and time window length (e.g. 5, 10 and 15 minutes). Existing methods of estimating travel time from GPS data are not capable of simultaneously taking into account the issues related to uncertainties associated with GPS and spatial road network data; low sampling frequency; low density vehicle coverage on some roads on the network; time window length; and vehicle fleet sample size. Accordingly this research is based on the development and application of a methodology which uses GPS data to reliably estimate travel time in real-time while considering the factors including vehicle fleet sample size, data sampling frequency and time window length in the estimation process. Specifically, the purpose of this thesis was to first determine the accurate location of a vehicle travelling on a road link by applying a map-matching algorithm at a range of sampling frequencies to reduce the potential errors associated with GPS and digital road maps, for example where vehicles are sometimes assigned to the wrong road links. Secondly, four different methods have been developed to estimate link travel time based on map-matched GPS positions and speed data from low frequency data sets in three time windows lengths (i.e. 5, 10 and 15 minutes). These are based on vehicle speeds, speed limits, link distances and average speeds; initially only within the given link but subsequently in the adjacent links too. More specifically, the final method draws on weighted link travel times associated with the given and adjacent links in both spatial and temporal dimensions to estimate link travel time for the given link. GPS data from Interstate I-880 (California, USA) for a total of 73 vehicles over 6 hours were obtained from the UC-Berkeley s Mobile Century Project. The original GPS dataset which was broadcast on a 3 second sampling frequency has been extracted at different sampling frequencies such as 6, 30, 60 and 120 seconds so as to evaluate the performance of each travel time estimation method at low sampling frequencies. The results were then validated against reference travel time data collected from 4,126 vehicles by high resolution video cameras, and these indicate that factors such as vehicle sample size, data sampling frequency, vehicle coverage on the links and time window length all influence the accuracy of link travel time estimation.

This research seeks to investigate how additional data sources can be used within traffic control systems to reduce average delay and improve reliability of journey time. Current state of the art urban traffic control systems do not take full advantage of the improved granularity of data available as they use traditional, static detection methods such as inductive loops, infra-red and radar. Therefore further research was required to fully understand what new data sources are available, how they could be used and if there are any potential benefits for traffic control systems. The transport industry is moving into an era of data abundance as more people use smart phones, satellite navigation systems, Wi-Fi and Bluetooth devices. These richer data sources could provide additional information (vehicle location, speed and destination data) but it is currently unknown as to whether they can improve the performance of the road network. Much of the research in this thesis has been published through conference and journal papers. A novel traffic control algorithm called DEMA was developed during this research, which can significantly outperform MOVA (a leading industrial control algorithm) through reducing average delay by up to 34% when additional data sources are incorporated into the decision process. DEMA uses vehicle location, speed and turning intention information to select the most suitable stage for minimising delay. Also a study was conducted to determine if turning intention information could be predicted from outside of a vehicle, which is a previously un-researched area. The results demonstrated that people could correctly predict turning intention with a 70% median success rate when the vehicles were 50 metres from the junction. The outcomes of this research could have a significant impact on the future of urban traffic control systems as new data sources become more readily available in the transport industry.

The potential for binder rich bituminous road surfaces to deliver low levels of dry friction was first noted in 1944. Using established test methods exploiting basic principles of physics first tested in criminal court in the 1940s (and still in use today) it has been possible to show statistically that modern negative textured road surfaces (NTS) deliver levels of dry friction significantly below those considered typical by collision investigators for the more traditional positive textured surfaces (PTS). NTS surfaces are shown to perform relatively worse in the absence of ABS (Anti-Blockier System, Anti-lock braking) than PTS equivalents such as Hot Rolled Asphalt (HRA). Skid tests undertaken on DRY NTS surfaces with ABS braking have been shown to manifest momentary low levels of deceleration similar to those experienced during NOABS tests on the same surfaces and to generate dash like skid marks atypical of ABS tests on DRY PTS surfaces. The ratio of peak to sliding friction also appears lower for dry NTS surfaces than for Dry PTS surfaces documented in the literature. Using high-speed video and false colour infrared imaging it has been possible to see the low friction phenomenon termed “bituplaning”. Vehicles equipped with ABS have also been shown to suffer momentary “bituplanes” resulting in less than optimum performance. Tyre deformation during dry skidding on NTS appears reduced in relation to a PTS equivalent.

This research has examined the application of machine learning and statistical methods for developing roadside particle (number/mass concentrations) prediction models that can be used for air quality management. Data collected from continuous monitoring stations including pollutants, traffic and meteorological variables were used for training the models. A hybrid feature selection method involving Genetic Algorithms and Random Forests was successfully used in selecting the most relevant predictor variables for the models from the variables selected based on their correlation with the PM\(_+\), PM\(_{2.5}\) and PNC concentrations. The study found that the hybrid feature selection can be used with both statistical and machine learning methods to produce less expensive and more efficient air quality prediction models. Among the machine learning models studied the Boosted Regression Trees (BRT), Random Forests (RF), Extreme Learning Machines (ELM) and Deep Learning Algorithms were found to be the most suitable for the predictions of roadside PM\(_+\), PM\(_{2.5}\), and PNC concentrations. The machine learning models performed better than the ADMS-road model in spatiotemporal predictions involving monitoring sites locations. Moreover, they performed much better in predicting the concentrations in street Canyons. The ANN and BRT were found to be suitable for air quality management applications involving traffic management scenarios.

Dans cette thèse, nous nous intéressons au problème du suivi en convoi, désigné en anglais par le terme platooning, où un train de robots essaie de suivre un chemin décrit par le leader. Ce chemin, n'étant pas prédéfini mais généré au cours du suivi, est inconnu de tous les robots suiveurs. Dans ce travail, nous choisissons une approche décentralisée locale où chaque robot du convoi observe son voisinage et calcule son contrôle de façon à avoir un suivi stable (absence d'oscillations) et précis (erreur latérale aussi faible que possible). Cette thèse étudie plus précisément le comportement latéral d'un robot du convoi et propose deux contrôleurs s'appuyant sur la mémorisation du chemin suivi par son prédécesseur. Un premier algorithme de contrôle Memo-LAT (Memorization and Look-Ahead Target) calcule une commande latérale continue en utilisant une loi de contrôle analytique. La stabilité de Memo-LAT n'étant pas toujours garantie, nous proposons l'algorithme de contrôle NOC (Non-Oscillatory Convergence) qui prend en compte la courbure du chemin à suivre dans le calcul du comportement latéral. NOC combine une approche géométrique avec une recherche heuristique pour calculer une commande discrète permettant au robot de suivre avec précision le chemin de son prédécesseur sans oscillation. / This thesis deals with the platooning problem which aims to concieve a control algorithm allowing a convoy of vehicles to follow their leader's path. This path, which is initially undefined and unknown to all the following robots, is generated as the leader moves. In this thesis, we choose a local decentralized approach in which each robot of the platoon uses its local perceptions to compute its own commands aiming to achieve a stable (no oscillations) and precise (with a lateral error as small as possible) platooning. More precisely, this thesis studies the lateral behavior of a platoon's robot and introduces two controllers based on the memorization of the robot's predecessor's path. The first algorithm, Memo-LAT (Memorization and Look-Ahead Target), computes a continuous lateral command using an analytic control law. As the stability of Memo-LAT is not always guaranteed, we present NOC (Non-Oscillatory Convergence), a control algorithm which takes into account the path's curvature in the robot's lateral behavior's computation. NOC combines a geometric approach to a heuristic search method to compute a discrete command allowing the robot to follow precisely and without oscillations its predecessor's path