[pt] Nesse trabalho, nós apresentamos novas aplicações para alguns controladores robustos, tais como as abordagens SMC e STA. O principal objetivo é conseguir executar uma navegação autônoma precisa em campos agrícolas, usando robôs móveis com rodas, equipados com uma câmera monocular fixa. Primeiro, nós projetamos uma abordagem de controle robusto baseado em servo-visão, a fim de lidar com imprecisões do modelo e perturbações da trajetória no espaço da imagem. Além disso, uma abordagem de controle robusto baseada em cascata, é aplicada, na qual, a malha de realimentação externo está conectada com uma malha de realimentação interna para lidar com os efeitos de todas as fontes de perturbação. Desse modo, uma abordagem robusta de rastreamento de trajetória, baseada em super-twisting, é aplicada para estabilização de movimento afim de garantir o sucesso da tarefa de seguir uma linha de cultivo considerando os efeitos de derrapagem das rodas e derrapagem lateral do veículo. A plataforma ROS-Gazebo, um simulador de robótica de código aberto, foi utilizada para realização de simulações computacionais 3D usando um robô móvel do tipo differentialdrive e um ambiente ad-hoc projetado para cultivo em linha. A eficácia e a viabilidade dos controladores robustos são avaliadas analisando simulações numéricas e métricas de desempenho, tais como: (i) o Erro Quadrático Médio (EQM) e (ii) o Desvio Absoluto Médio (DAM). Além disso, nós veremos nos resultados, que em geral, só é possível ter estabilidade, utilizando os controladores rosbustos. / [en] In this work, we present a new application for some robust controllers, such as SMC and STA approaches. The main idea is to perform autonomous navigation in agricultural fields accurately using wheeled mobile robots, equipped with a fixed monocular camera . Here, we consider the existence of uncertainties in the parameters of the robot-camera system and external disturbances caused by high driving velocities, sparse plants, and uneven terrains. First, we design a robust image-based visual servoing approach to deal with model inaccuracies and trajectory perturbations in the image space. In addition, a cascade-based robust control approach is applied, in which the outer vision feedback loop is connected with an inner pose feedback loop to deal with the effects of all disturbances sources. Then, a robust trajectory tracking approach based on the super-twisting algorithm is applied for motion stabilization to ensure the successful execution of row crop following tasks under wheel slippage and vehicle sideslip. ROSGazebo platform, an open-source robotics simulator, was used to perform 3D computer simulation using a differential-drive mobile robot and an adhoc designed row-crop environment. The effectiveness and feasibility of the robust controllers are evaluated by analyzing numerical simulations and performance metrics, such as: (i) the root-mean square error (RMSE) and (ii) the mean-absolute deviation (MAD). Furthermore, we will see in results, that in general, it is only possible to have stability, using robust controllers.

No description available.

With the development of autonomous vehicles, more and more technology is introduced into the automotive industry. Ethernet has found its way into the vehicle network, and it is forced to coexist with the well-established CAN bus. In terms of data acquisition, the presence of a mixed network brings challenges with significant changes in network architecture. This thesis explores CAN-Ethernet gateways as a replacement for the PCIe bus CAN transceivers utilized in today's logging systems, with the purpose to improve the adaptability of the autonomous logging system. A CAN-Ethernet gateway was implemented using Kvaser's DIN Rail SE400S-X10in an experimental comparison against the PCIe logging solution, including both classical CAN and CAN FD communication. In addition, a case study on the benefits and drawbacks with implementing an Ethernet architecture was performed, utilizing semi-structured interviews. It was concluded that a CAN-Ethernet gateway provides a robust solution in relation to data loss. Throughout the tests, the message loss rate was 0% for both logging solutions. However, CAN-Ethernetlogging introduced additional delay into the system. For the tests on a truck simulation rig the mean additional delay from a CAN-Ethernet gateway, compared to the existing PCIe-CAN logging, was 2 ms. Moreover, some spikes occurred and in a number of cases it could be up to 6 ms in additional delay compared to the existing PCIe logging. It was also proven difficult to time synchronize the gateway with the autonomous logging system, and unknown delays had an impact. Relevant metrics were obtained from relative measurements of side-by-side logging between the PCIe and CAN-Ethernet communication. The standard deviation and fluctuation of the delay were relevant metrics, since smaller fluctuations made the delay more predictable and real-time compatible. A CAN-Ethernet deployment may create a more complex architecture in general, and as of now has limitations for real-time systems. On the other hand, it may offer significant possibilities in future development of a more adaptable and scalable logging system. / Med utvecklingen av autonoma fordon har mer och mer teknologi introducerats inom fordonsindustrin. Ethernet har funnit sin väg in i fordonsnätverket och tvingas existera sida vid sida med den väletablerade CAN-bussen. För dataloggning orsakar närvaron av ett blandat nätverk (med både CAN och Ethernet) utmaningar i samband med stora förändringar inom nätverksarkitektur. Det här examensarbetet utforskar nätslussar för CAN-Ethernet som en ersättare till PCIe-bussens CAN-sändtagare som används i dagens loggningssystem. En CAN-Ethernet-nätsluss implementerades genom att använda Kvasers DIN Rail SE400S-X10 i en experimentell jämförelse med PCIe-loggningssystemet, och inkluderade både klassisk CAN samt CAN FD kommunikation. I tillägg gjordes en fallstudie om fördelar och nackdelar med att implementera en Ethernet-arkitektur,vilken grundades på semi-strukturerade intervjuer. Slutsatsen var att CAN-Ethernet-nätslussar tillhandahåller en robust lösning i förhållande till dataförlust. Under alla testerna var meddelandeförlusten 0% hos båda loggningsmetoderna. Däremot introducerade CAN-Ethernet-loggning en ökad fördröjning till systemet. För testerna på lastbilsriggsimulatorn var fördröjningen 2 ms jämfört med PCIe-CAN-loggningen. Dessutom förekom spikar i fördröjningen och i vissa fall resulterade fördröjningen i upp mot 6 ms, jämfört mot den befintliga PCIe-loggningen. Det visade sig även vara svårt att tidssynkronisera nätslussen med det autonoma loggningssystemet och okända fördröjningar hade en inverkan. Relevanta mått erhölls från relativa mätningar av jämsides loggning mellan PCIe och CAN-Ethernet kommunikation. Standardavvikelsen och fluktuation av fördröjningen var relevanta mått eftersom mindre fluktuationer resulterade i en mer förutsägbar samt realtidskompatibel fördröjning. Användningen av CAN-Ethernet kan, i allmänhet, resultera i en mer avancerad arkitektur och har i dagsläget begränsningar inom realtidssystem. Å andra sidan kan CAN-Ethernet erbjuda markanta möjligheter inom framtida utveckling av ett mer modulärt och skalbart loggningssystem.

This project’s goal was to assess both the challenges of implementing the Deep Q-Learning algorithm to create an autonomous car in the CARLA simulator, and the driving performance of the resulting model. An agent was trained to follow waypoints based on two main approaches. First, a camera-based approach, which allowed the agent to gather information about the environment from a camera sensor. The image along with other driving features were fed to a convolutional neural network. Second, an approach focused purely on following the waypoints without the camera sensor. The camera sensor was substituted for an array containing the agent’s angle with respect to the upcoming waypoints along with other driving features. Even though the camera-based approach was the best during evaluation, no approach was successful in consistently following the waypoints of a straight route. To increase the performance of the camera-based approach more training episodes need to be provided. Furthermore, both approaches would greatly benefit from experimentation and optimization of the model’s neural network configuration and its hyperparameters.

This thesis addresses the problem of energy management of a hybrid electric power unit for an autonomous vehicle. We introduce, evaluate, and discuss the idea of autonomous-specific energy management strategy. This method is an optimization-based strategy which improves the powertrain fuel economy by exploiting motion planning data. First, to build a firm base for further evaluations, we will develop a high-fidelity system-level model for our case study using MATLAB/Simulink. This model mostly concerns about energy-related aspects of the powertrain and the vehicle. We will derive and implement the equations for each of the model subsystems. We derive model parameters using available data in the literature or online. Evaluation of the developed model shows acceptable conformity with the actual dynamometer data. We will use this model to replace the built-in rule-based logic with the proposed strategy and assess the performance.\par Second, since we are considering an optimization-based approach, we will develop a novel convex representation of the vehicle and powertrain model. This translates to reformulating the model equations using convex functions. Consequently, we will express the fuel-efficient energy management problem as the convex optimization problem. We will solve the optimization problem using dedicated numerical solvers. Extracting the control inputs using this approach and applying them on the high-fidelity model provides similar results to dynamic programming in terms of fuel consumption but in substantially less amount of time. This will act as a pivot for the subsequent real-time analysis.\par Third, we will perform a proof-of-concept for the autonomous-specific energy management strategy. We implement an optimization-based path and trajectory planning for a vehicle in the simplified driving scenario of a racing track. Accordingly, we use motion planning data to obtain the energy management strategy by solving an optimization problem. We will let the vehicle to travel around the circuit with the ability to perceive and plan up to an observable horizon using the receding horizon approach. Developed approach for energy management strategy shows a substantial reduction in the fuel consumption of the high-fidelity model, compared to the rule-based controller. / Thesis / Master of Science in Mechanical Engineering (MSME) / The automotive industry is on the verge of groundbreaking transformations as a result of electrification and autonomous driving. Electrified autonomous car of the future is sustainable, energy-efficient, more convenient, and safer. In addition to the advantages of electrification and autonomous driving individually, the intersection and interaction of these mainstreams provide new opportunities for further improvements on the vehicles. Autonomous cars generate an unprecedented amount of real-time data due to excessive use of perception sensors and processing units. This thesis considers the case of an autonomous hybrid electric vehicle and presents the novel idea of autonomous-specific energy management strategy. Specifically, this thesis is a proof-of-concept, a trial to exploit the motion planning data for a self-driving car to improve the fuel economy of the hybrid electric power unit by adopting a more efficient energy management strategy. With the ever-increasing number of autonomous hybrid electric vehicles, particularly in the self-driving fleets, the presented method shows an extremely promising potential to reduce the fuel consumption of these vehicles.

Recent advancements in time-optimal trajectory planning, control, and state estimation for autonomous vehicles have paved the way for the emerging field of autonomous racing. In the last 5-10 years, this form of racing has become a popular and challenging testbed for autonomous driving algorithms, aiming to enhance the safety and performance of future intelligent vehicles. In autonomous racing, the main goal is to develop real-time algorithms capable of autonomously maneuvering a vehicle around a racetrack, even in the presence of moving opponents. However, as a vehicle approaches its handling limits, several challenges arise for online trajectory planning and control. The vehicle dynamics become nonlinear and hard to capture with low-complexity models, while fast re-planning and good generalization capabilities are crucial to execute optimal maneuvers in unforeseen scenarios. These challenges leave several open research questions, three of which will be addressed in this thesis. The first explores developing accurate yet computationally efficient vehicle models for online time-optimal trajectory planning. The second focuses on enhancing learning-based methods for trajectory planning, control, and state estimation, overcoming issues like poor generalization and the need for large amounts of training data. The third investigates the optimality of online-executed trajectories with simplified vehicle models, compared to offline solutions of minimum-lap-time optimal control problems using high-fidelity vehicle models. This thesis consists of four parts, each of which addresses one or more of the aforementioned research questions, in the fields of time-optimal vehicle trajectory planning, control and state estimation. The first part of the thesis presents a novel artificial race driver (ARD), which autonomously learns to drive a vehicle around an obstacle-free circuit, performing online time-optimal vehicle trajectory planning and control. The following research questions are addressed in this part: How optimal is the trajectory executed online by an artificial agent that drives a high-fidelity vehicle model, in comparison with a minimum-lap-time optimal control problem (MLT-OCP), based on the same vehicle model and solved offline? Can the artificial agent generalize to circuits and conditions not seen during training? ARD employs an original neural network with a physics-driven internal structure (PhS-NN) for steering control, and a novel kineto-dynamical vehicle model for time-optimal trajectory planning. A new learning scheme enables ARD to progressively learn the nonlinear dynamics of an unknown vehicle. When tested on a high-fidelity model of a high-performance car, ARD achieves very similar results as an MLT-OCP, based on the same vehicle model and solved offline. When tested on a 1:8 vehicle prototype, ARD achieves similar lap times as an offline optimization problem. Thanks to its physics-driven architecture, ARD generalizes well to unseen circuits and scenarios, and is robust to unmodeled changes in the vehicle’s mass. The second part of the thesis deals with online time-optimal trajectory planning for dynamic obstacle avoidance. The research questions addressed in this part are: Can time-optimal trajectory planning for dynamic obstacle avoidance be performed online and with low computational times? How optimal is the resulting trajectory? Can the planner generalize to unseen circuits and scenarios? At each planning step, the proposed approach builds a tree of time-optimal motion primitives, by performing a sampling-based exploration in a local mesh of waypoints. The novel planner is validated in challenging scenarios with multiple dynamic opponents, and is shown to be computationally efficient, to return near-time-optimal trajectories, and to generalize well to new circuits and scenarios. The third part of the thesis shows an application of time-optimal trajectory planning with optimal control and PhS-NNs in the context of autonomous parking. The research questions addressed in this part are: Can an autonomous parking framework perform fast online trajectory planning and tracking in real-life parking scenarios, such as parallel, reverse and angle parking spots, and unstructured environments? Can the framework generalize to unknown variations in the vehicle’s parameters and road adherence, and operate with measurement noise? The autonomous parking framework employs a novel penalty function for collision avoidance with optimal control, a new warm-start strategy and an original PhS-NN for steering control. The framework executes complex maneuvers in a wide range of parking scenarios, and is validated with a high-fidelity vehicle model. The framework is shown to be robust to variations in the vehicle’s mass and road adherence, and to operate with realistic measurement noise. The fourth and last part of the thesis develops novel kinematics-structured neural networks (KS-NNs) to estimate the vehicle’s lateral velocity, which is a key quantity for time-optimal trajectory planning and control. The KS-NNs are a special type of PhS-NNs: their internal structure is designed to incorporate the kinematic principles, which enhances the generalization capabilities and physical explainability. The research questions addressed in this part are: Can a neural network-based lateral velocity estimator generalize well when tested on a vehicle not used for training? Can the network’s parameters be physically explainable? The approach is validated using an open dataset with two race cars. In comparison with traditional and neural network estimators of the literature, the KS-NNs improve noise rejection, exhibit better generalization capacity, are more sample-efficient, and their structure is physically explainable.

Le travail présenté dans la thèse s’inscrit dans le projet de recherche partenarial ANR-ABV 2009 dont l’objet est la conception d’un système de conduite automatisée à basse vitesse. Il décrit et analyse les principes d’un contrôle partagé d’un véhicule automobile entre un conducteur humain et un copilote électronique (E-copilote). L’objectif est de mettre en place une coopération Homme-Machine efficace entre le conducteur et l’E-copilote. Un des enjeux est notamment de permettre au conducteur d’interagir avec l’E-copilote de façon continue pour pouvoir exécuter les manœuvres qu’il souhaite sans nécessiter la désactivation ni être gêné par l’E-copilote. Cet enjeu répond au besoin de prise en compte des actions du conducteur entreprises pour pallier celles du E-copilote dans certaines situations par exemple éviter un obstacle non perçu par le système. L’objectif dans ce cas est de garantir le confort au conducteur ainsi que sa conscience du mode engagé (système actif ou pas). Le conducteur et l’E-copilote agissant simultanément sur le système de direction, chacun doit être conscient des actions de l’autre : une communication bidirectionnelle est essentielle. Pour atteindre cet objectif, nous avons retenu les interactions haptiques à travers le système de direction du véhicule. Le couple appliqué par le conducteur sur volant est utilisé par l’E-copilote pour prendre en compte ces actions de la même façon que le couple produit par l’E-copilote est ressenti par le conducteur et utilisé pour comprendre le comportement du système. D’autres aspects essentiels pour la coopération H-M ont également été abordés : l’´étude des changements de modes de fonctionnement du système ainsi que l’IHM via laquelle le conducteur interagit avec le système. / The work presented in the thesis is part of the research partnership project ANRABV 2009 which aims is to design an automated low-speed driving. It describes and analyzes the principles of shared control of a motor vehicle between a human driver and an electronic copilot (E-copilot). The objective is to establish effective human-machine cooperation between the driver and E-copilot. One issue is particular to allow the driver to interact with the E-copilot continuously in order to perform maneuvers he wants without requiring deactivation neither constrained by E-copilot. This issue addresses the need for consideration of driver actions taken to remedy those of E-copilot for example avoiding undetected obstacle by the system while ensuring operator comfort and the driver situation awareness. The driver and E-co-pilot acting simultaneously on the steering system, everyone must be aware of the actions of the other: twoway communication is essential. To achieve this goal, we used the haptic interactions through the steering system of the vehicle. The torque applied by the driver on the steering wheel is used by the E-copilot to take into account these actions as the torque produced by the E-copilot is felt by the driver and used to understand the system’s behavior. Other key issues for the Human-Machine Cooperation were also discussed: the study of changes in modes of operation of the system and HMI via which the driver interact with the system.

[ES] En la actualidad, uno de los principales desafíos a los que se enfrentan las grandes áreas metropolitanas es la congestión provocada por el tráfico, la cual se ha convertido en un problema importante al que se enfrentan las autoridades de cada ciudad. Para abordar este problema es necesario implementar una solución eficiente para controlar el tráfico que genere beneficios para los ciudadanos, como reducir los tiempos de viaje de los vehículos y, en consecuencia, el consumo de combustible, el ruido, y la contaminación ambiental. De hecho, al analizar adecuadamente la demanda de tráfico, es posible predecir las condiciones futuras del tráfico, y utilizar esa información para la optimización de las rutas tomadas por los vehículos. Este enfoque puede ser especialmente efectivo si se aplica en el contexto de los vehículos autónomos, que tienen un comportamiento más predecible, lo cual permite a los administradores de la ciudad mitigar los efectos de la congestión, como es la contaminación, al mejorar el flujo de tráfico de manera totalmente centralizada. La validación de este enfoque generalmente requiere el uso de simulaciones que deberían ser lo más realistas posible. Sin embargo, lograr altos grados de realismo puede ser complejo cuando los patrones de tráfico reales, definidos a través de una matriz de Origen/Destino (O-D) para los vehículos en una ciudad, son desconocidos, como ocurre la mayoría de las veces. Por lo tanto, la primera contribución de esta tesis es desarrollar una heurística iterativa para mejorar el modelado de la congestión de tráfico; a partir de las mediciones de bucle de inducción reales hechas por el Ayuntamiento de Valencia (España), pudimos generar una matriz O-D para la simulación de tráfico que se asemeja a la distribución de tráfico real. Si fuera posible caracterizar el estado del tráfico prediciendo las condiciones futuras del tráfico para optimizar la ruta de los vehículos automatizados, y si se pudieran tomar estas medidas para mitigar de manera preventiva los efectos de la congestión con sus problemas relacionados, se podría mejorar el flujo de tráfico en general. Por lo tanto, la segunda contribución de esta tesis es desarrollar una Ecuación de Predicción de Tráfico para caracterizar el comportamiento en las diferentes calles de la ciudad en términos de tiempo de viaje con respecto al volumen de tráfico, y aplicar una regresión logística a esos datos para predecir las condiciones futuras del tráfico. La tercera y última contribución de esta tesis apunta directamente al nuevo paradigma de gestión de tráfico previsto, tratándose de un servidor de rutas capaz de manejar todo el tráfico en una ciudad, y equilibrar los flujos de tráfico teniendo en cuenta las condiciones de congestión del tráfico presentes y futuras. Por lo tanto, realizamos un estudio de simulación con datos reales de congestión de tráfico en la ciudad de Valencia (España), para demostrar cómo se puede mejorar el flujo de tráfico en un día típico mediante la solución propuesta. Los resultados experimentales muestran que nuestra solución, combinada con una actualización frecuente de las condiciones del tráfico en el servidor de rutas, es capaz de lograr mejoras sustanciales en términos de velocidad promedio y tiempo de trayecto, ambos indicadores de un menor grado de congestión y de una mejor fluidez del tráfico. / [CA] En l'actualitat, un dels principals desafiaments als quals s'enfronten les grans àrees metropolitanes és la congestió provocada pel trànsit, que s'ha convertit en un problema important al qual s'enfronten les autoritats de cada ciutat. Per a abordar aquest problema és necessari implementar una solució eficient per a controlar el trànsit que genere beneficis per als ciutadans, com reduir els temps de viatge dels vehicles i, en conseqüència, el consum de combustible, el soroll, i la contaminació ambiental. De fet, en analitzar adequadament la demanda de trànsit, és possible predir les condicions futures del trànsit, i utilitzar aqueixa informació per a l'optimització de les rutes preses pels vehicles. Aquest enfocament pot ser especialment efectiu si s'aplica en el context dels vehicles autònoms, que tenen un comportament més predictible, i això permet als administradors de la ciutat mitigar els efectes de la congestió, com és la contaminació, en millorar el flux de trànsit de manera totalment centralitzada. La validació d'aquest enfocament generalment requereix l'ús de simulacions que haurien de ser el més realistes possible. No obstant això, aconseguir alts graus de realisme pot ser complex quan els patrons de trànsit reals, definits a través d'una matriu d'Origen/Destinació (O-D) per als vehicles en una ciutat, són desconeguts, com ocorre la majoria de les vegades. Per tant, la primera contribució d'aquesta tesi és desenvolupar una heurística iterativa per a millorar el modelatge de la congestió de trànsit; a partir dels mesuraments de bucle d'inducció reals fetes per l'Ajuntament de València (Espanya), vam poder generar una matriu O-D per a la simulació de trànsit que s'assembla a la distribució de trànsit real. Si fóra possible caracteritzar l'estat del trànsit predient les condicions futures del trànsit per a optimitzar la ruta dels vehicles automatitzats, i si es pogueren prendre aquestes mesures per a mitigar de manera preventiva els efectes de la congestió amb els seus problemes relacionats, es podria millorar el flux de trànsit en general. Per tant, la segona contribució d'aquesta tesi és desenvolupar una Equació de Predicció de Trànsit per a caracteritzar el comportament en els diferents carrers de la ciutat en termes de temps de viatge respecte al volum de trànsit, i aplicar una regressió logística a aqueixes dades per a predir les condicions futures del trànsit. La tercera i última contribució d'aquesta tesi apunta directament al nou paradigma de gestió de trànsit previst. Es tracta d'un servidor de rutes capaç de manejar tot el trànsit en una ciutat, i equilibrar els fluxos de trànsit tenint en compte les condicions de congestió del trànsit presents i futures. Per tant, realitzem un estudi de simulació amb dades reals de congestió de trànsit a la ciutat de València (Espanya), per a demostrar com es pot millorar el flux de trànsit en un dia típic mitjançant la solució proposada. Els resultats experimentals mostren que la nostra solució, combinada amb una actualització freqüent de les condicions del trànsit en el servidor de rutes, és capaç d'aconseguir millores substancials en termes de velocitat faig una mitjana i de temps de trajecte, tots dos indicadors d'un grau menor de congestió i d'una fluïdesa millor del trànsit. / [EN] Currently, one of the main challenges that large metropolitan areas have to face is traffic congestion, which has become an important problem faced by city authorities. To address this problem, it becomes necessary to implement an efficient solution to control traffic that generates benefits for citizens, such as reducing vehicle journey times and, consequently, use of fuel, noise and environmental pollution. In fact, by properly analyzing traffic demand, it becomes possible to predict future traffic conditions, and to use that information for the optimization of the routes taken by vehicles. Such an approach becomes especially effective if applied in the context of autonomous vehicles, which have a more predictable behavior, thus enabling city management entities to mitigate the effects of traffic congestion and pollution by improving the traffic flow in a city in a fully centralized manner. Validating this approach typically requires the use of simulations, which should be as realistic as possible. However, achieving high degrees of realism can be complex when the actual traffic patterns, defined through an Origin/Destination (O-D) matrix for the vehicles in a city, are unknown, as occurs most of the times. Thus, the first contribution of this thesis is to develop an iterative heuristic for improving traffic congestion modeling; starting from real induction loop measurements made available by the City Hall of Valencia, Spain, we were able to generate an O-D matrix for traffic simulation that resembles the real traffic distribution. If it were possible to characterize the state of traffic by predicting future traffic conditions for optimizing the route of automated vehicles, and if these measures could be taken to preventively mitigate the effects of congestion with its related problems, the overall traffic flow could be improved. Thereby, the second contribution of this thesis was to develop a Traffic Prediction Equation to characterize the different streets of a city in terms of travel time with respect to the vehicle load, and applying logistic regression to those data to predict future traffic conditions. The third and last contribution of this thesis towards our envisioned traffic management paradigm was a route server capable of handling all the traffic in a city, and balancing traffic flows by accounting for present and future traffic congestion conditions. Thus, we perform a simulation study using real data of traffic congestion in the city of Valencia, Spain, to demonstrate how the traffic flow in a typical day can be improved using our proposed solution. Experimental results show that our proposed solution, combined with frequent updating of traffic conditions on the route server, is able to achieve substantial improvements in terms of average travel speeds and travel times, both indicators of lower degrees of congestion and improved traffic fluidity. / Finally, I want to thank the Ecuatorian Republic through the "Secretaría de Educación Superior, Ciencia, Tecnología e Innovación" (SENESCYT), for granting me the scholarship to finance my studies. / Zambrano Martínez, JL. (2019). Efficient Traffic Management in Urban Environments [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/129865 / TESIS

<p>This thesis presents solutions to some of the problems facing Autonomous Surface Vehicle (ASV) deployments in inland waterways through the development of navigational and control systems. Fluvial systems are one of the hardest inland waterways to navigate and are thus used as a use-case for system development. The systems are built to reduce the reliance on a-prioris during ASV operation. This is crucial for exceptionally dynamic environments such as fluvial bodies of water that have poorly defined routes and edges, can change course in short time spans, carry away and deposit obstacles, and expose or cover shoals and man-made structures as their water level changes. While navigation of fluvial systems is exceptionally difficult potential autonomous data collection can aid in important scientific missions in under studied environments.</p> <p><br></p> <p>The work has four contributions targeting solutions to four fundamental problems present in fluvial system navigation and control. To sense the course of fluvial systems for navigable path determination a fluvial segmentation study is done and a novel dataset detailed. To enable rapid path computations and augmentations in a fast moving environment a Dubins path generator and augmentation algorithm is presented ans is used in conjunction with an Integral Line-Of-Sight (ILOS) path following method. To rapidly avoid unseen/undetected obstacles present in fluvial environments a Deep Reinforcement Learning (DRL) agent is built and tested across domains to create dynamic local paths that can be rapidly affixed to for collision avoidance. Finally, a custom low-cost and deployable ASV, BREAM (Boat for Robotic Engineering and Applied Machine-Learning), capable of operating in fluvial environments is presented along with an autonomy package used in providing base level sensing and autonomy processing capability to varying platforms.</p> <p><br></p> <p>Each of these contributions form a part of a larger documented Fluvial Navigation Control Architecture (FNCA) that is proposed as a way to aid in a-priori free navigation of fluvial waterways. The architecture relates the navigational structures into high, mid, and low-level controller Guidance and Navigational Control (GNC) layers that are designed to increase cross vehicle and domain deployments. Each component of the architecture is documented, tested, and its application to the control architecture as a whole is reported.</p>

Several manufacturers of road vehicles today are working on developing autonomous vehicles. One subject that is often up for discussion when it comes to integrating autonomous road vehicles into the infrastructure is the safety aspect. There is in the context no common view of how safety should be quantified. As a contribution to this discussion we propose describing each potential hazardous event of a vehicle as a Semi-Markov Process (SMP). A reliability-based method for using the semi-Markov representation to calculate the probability of a hazardous event to occur is presented. The method simplifies the expression for the reliability using the Laplace-Stieltjes transform and calculates the transform of the reliability exactly. Numerical inversion algorithms are then applied to approximate the reliability up to a desired error tolerance. The method is validated using alternative techniques and is thereafter applied to a system for automated steering based on a real example from the industry. A desired evolution of the method is to involve a framework for how to represent each hazardous event as a SMP. / Flertalet tillverkare av vägfordon jobbar idag på att utveckla autonoma fordon. Ett ämne ofta på agendan i diskussionen om att integrera autonoma fordon på vägarna är säkerhet. Det finns i sammanhanget ingen klar bild över hur säkerhet ska kvantifieras. Som ett bidrag till denna diskussion föreslås här att beskriva varje potentiellt farlig situation av ett fordon som en Semi-Markov process (SMP). En metod presenteras för att via beräkning av funktionssäkerheten nyttja semi-Markov representationen för att beräkna sannolikheten för att en farlig situation ska uppstå. Metoden nyttjar Laplace-Stieltjes transformen för att förenkla uttrycket för funktionssäkerheten och beräknar transformen av funktionssäkerheten exakt. Numeriska algoritmer för den inversa transformen appliceras sedan för att beräkna funktionssäkerheten upp till en viss feltolerans. Metoden valideras genom alternativa tekniker och appliceras sedan på ett system för autonom styrning baserat på ett riktigt exempel från industrin. En fördelaktig utveckling av metoden som presenteras här skulle vara att involvera ett ramverk för hur varje potentiellt farlig situation ska representeras som en SMP.