With the envisioned age of Internet of Things (IoTs), different aspects of Intelligent Transportation System (ITS) will be linked so as to advance road transportation safety, ease congestion of road traffic, lessen air pollution, improve passenger transportation comfort and significantly reduce road accidents. In vehicular networks, regular exchange of current position, direction, speed, etc., enable mobile vehicle to foresee an imminent vehicle accident and notify the driver early enough in order to take appropriate action(s) or the vehicle on its own may take adequate preventive measures to avert the looming accident. Actualizing this concept requires use of shared media access protocol that is capable of guaranteeing reliable and timely broadcast of safety messages. This dissertation investigates the use of Network Coding (NC) techniques to enrich the content of each transmission and ensure improved high reliability of the broadcasted safety messages with less number of retransmissions. A Code Aided Retransmission-based Error Recovery (CARER) protocol is proposed. In order to avoid broadcast storm problem, a rebroadcasting vehicle selection metric η, is developed, which is used to select a vehicle that will rebroadcast the received encoded message. Although the proposed CARER protocol demonstrates an impressive performance, the level of incurred overhead is fairly high due to the use of complex rebroadcasting vehicle selection metric. To resolve this issue, a Random Network Coding (RNC) and vehicle clustering based vehicular communication scheme with low algorithmic complexity, named Reliable and Enhanced Cooperative Cross-layer MAC (RECMAC) scheme, is proposed. The use of this clustering technique enables RECMAC to subdivide the vehicular network into small manageable, coordinated clusters which further improve transmission reliability and minimise negative impact of network overhead. Similarly, a Cluster Head (CH) selection metric F(j) is designed, which is used to determine and select the most suitably qualified candidate to become the CH of a particular cluster. Finally, in order to investigate the impact of available radio spectral resource, an in-depth study of the required amount of spectrum sufficient to support high transmission reliability and minimum latency requirements of critical road safety messages in vehicular networks was carried out. The performance of the proposed schemes was clearly shown with detailed theoretical analysis and was further validated with simulation experiments.

Orientador: Ely Carneiro de Paiva / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Mecânica / Made available in DSpace on 2018-08-27T21:55:57Z (GMT). No. of bitstreams: 1 MartinezArias_RonaldRicardo_M.pdf: 6818308 bytes, checksum: d91e74cb8827b9ee03bcb0e84ccc9bd0 (MD5) Previous issue date: 2014 / Resumo: No presente trabalho, o modelo dinâmico do dirigível do Projeto AURORA (Gomes e Ramos, 1998), (Azinheira et al., 2001) e (Azinheira et al., 2008) é adaptado para considerar 4 motores ao invés de 2 apenas. Além disso, esses motores possuem acionamento elétrico (DC-Brushless) ao invés de propulsão por motor de combustão. Esses motores poderão trabalhar em acionamento diferencial, ou seja, motores frente-trás ou esquerda-direita com forças de propulsão diferentes, com a finalidade de gerar momentos e forças que complementem os demais atuadores do dirigível, como as superfícies de cauda ou leme. Duas inovações principais caracterizam esse novo sistema. O primeiro ponto é a utilização de quatro propulsores vetorizáveis ao invés de apenas dois como o usual. O segundo ponto é a angulação de 20 [graus] presente na fixação dos propulsores. Essa angulação faz com que, ao ser vetorizado para cima, cada propulsor gere uma componente de força lateral, além das componentes longitudinal e vertical. Se a intensidade da força gerada em cada propulsor for a mesma, obviamente as forças laterais geradas num par de propulsores se cancelam, e temos uma situação parecida com a atuação clássica de dirigíveis, gerando forças apenas para frente e para cima. Mas se, por outro lado, as intensidades de força nos propulsores de um dado par (dianteiro ou traseiro) forem diferentes, então forças resultantes laterais (bem como momentos) podem ser obtidos no CV (Centro de Volume) do dirigível. Assim, se o par de propulsores dianteiros gera uma componente lateral de forças para a direita, por exemplo, e o par traseiro gera uma componente lateral para a esquerda, então temos a geração de um momento de guinada positivo (horário), supondo obviamente que os motores encontram-se vetorizados. Essa é a chamada propulsão diferencial esquerda-direita que permite a geração de momentos de guinada em baixas velocidades, de tal forma compensar a baixa eficiência aerodinâmica da cauda com pouca incidência de ar (vento relativo). Outra possibilidade é a geração da propulsão diferencial dianteira-traseira, onde ambos propulsores dianteiros fornecem força de igual intensidade, mas de amplitude diferente daquela gerada pelos propulsores traseiros. As forças laterais são canceladas, mas um novo momento de arfagem pode ser gerado, e com um grau de liberdade a mais do que na situação onde se usava apenas dois propulsores vetorizáveis. Podemos obter inclusive um momento de arfagem no dirigível e ainda mantê-lo na posição "nivelada" de ângulo de arfagem (pitch) nulo. O uso da propulsão diferencial dianteira-traseira permite também obter uma mudança contínua e suave nas forças e momentos ao se variar a velocidade de operação do dirigível (airspeed). Evita-se assim a transição brusca de atuação que é observada quando o dirigível, na configuração clássica de apenas dois motores, passa das baixas velocidades (usando vetorização) para as altas velocidades (dispensando vetorização). Dessa forma, tanto a propulsão diferencial lateral (esquerda-direita), como a longitudinal (traseira-dianteira) obtida com essa configuração inédita, permite gerar momentos e forças que complementam os demais atuadores do dirigível, como as superfícies de cauda ou leme. Ressalta-se que as diferentes configurações de propulsão motora como essa proposta aqui considerada (de domínio do Projeto DRONI) poderão aumentar a eficiência e desempenho das abordagens de controle linear e não linear já desenvolvidas previamente no âmbito do Projeto AURORA / Abstract: In this paper masters, the dynamic model of AURORA Project airship (Gomes e Ramos, 1998), (Azinheira et al., 2001) and (Azinheira et al., 2008) is adapted to consider 4 engines instead of 2. Furthermore, those engines as electric start (DC-Brushless) instead of propulsion combustion engine. Engines can work with differential start, that is, front-back or left-right engines with different thrust forces. It has the purpose of generating forces and torques which complement other airship actuators, such as the tail and rudder surfaces. Two main innovations characterize this new system. First, it uses four thrusters with thrust vector control instead of just two as usual. Second, it is the 20 [degrees] angulation that is present in the thrusters setting. This angulation allows each thruster generates a lateral force component besides the longitudinal and vertical components when thrusters are vectorized upward. If the amount of force generated in each thruster is the same, obviously, the addition of lateral forces generated in a pair of thrusters will be zero, and we will have a similar situation with the classic airship performance, generating forces only forward and up. But if the amount of force on thrusters on a given pair (front or back) are different, then resulting lateral force (and torques) can be obtained on airship CV (Volume Center). Thus, if the front pair of thrusters generates a lateral force component to the right, for example, and the rear pair generates a lateral component to the left, we have a generation of yaw torque positive (clockwise), obviously assuming that the engines are vectorized. This is called the differential thrust left-right which allows the generation of yaw torques on low speeds, in order to compensate a low aerodynamic efficiency of tail with little air effect (relative wind). Another possibility is the generation of the differential thrust front-back, where both front thrusters provide equal amount of force, but with a different amplitude than the force generated by the rear thrusters. The lateral forces are canceled, but a new pitch torque can be generated, and with a degree of freedom more than in the situation where it was used only two thrusters with thrust vector control. We can get even a pitch torque on the airship and still keeping it on null pitch angle position. The use of diferential thrust front-back allows also getting a slight and continuous change on the forces and torques when is varied the airship operation speed (airspeed). It avoids the abrupt transition of performance that is observed when the airship, on classic configuration of only two engines, goes through from low speeds (using vectoring) to high speeds (dispensing vectorization). Thus, both the differential thrust lateral (left-right) as the longitudinal (front-back) obtained with this configuration unprecedented, it allows to generate torques and forces that complementing the other airship actuators, such as tail surfaces or rudder. It should be noted that different confiurations of thrust (Project DRONI) may increase the efficiency and performance of linear control approaches and nonlinear previously carried out under the AURORA Project / Mestrado / Mecanica dos Sólidos e Projeto Mecanico / Mestre em Engenharia Mecânica

Achieving robot autonomy is an extremely challenging task and it starts with developing algorithms that help the robot understand how humans perceive the environment around them. Once the robot understands how to make sense of its environment, it is easy to make efficient decisions about safe movement. It is hard for robots to perform tasks that come naturally to humans like understanding signboards, classifying traffic lights, planning path around dynamic obstacles, etc. In this work, we take up one such challenge of motion segmentation using Light Detection and Ranging (LiDAR) point clouds. Motion segmentation is the task of classifying a point as either moving or static. As the ego-vehicle moves along the road, it needs to detect moving cars with very high certainty as they are the areas of interest which provide cues to the ego-vehicle to plan it's motion. Motion segmentation algorithms segregate moving cars from static cars to give more importance to dynamic obstacles. In contrast to the usual LiDAR scan representations like range images and regular grid, this work uses a modern representation of LiDAR scans using permutohedral lattices. This representation gives ease of representing unstructured LiDAR points in an efficient lattice structure. We propose a machine learning approach to perform motion segmentation. The network architecture takes in two sequential point clouds and performs convolutions on them to estimate if 3D points from the first point cloud are moving or static. Using two temporal point clouds help the network in learning what features constitute motion. We have trained and tested our learning algorithm on the FlyingThings3D dataset and a modified KITTI dataset with simulated motion.

This dissertation first conducts a study on raw-data level cooperative perception for enhancing the detection ability of self-driving systems for connected autonomous vehicles (CAVs). A LiDAR (Light Detection and Ranging sensor) point cloud-based 3D object detection method is deployed to enhance detection performance by expanding the effective sensing area, capturing critical information in multiple scenarios and improving detection accuracy. In addition, a point cloud feature based cooperative perception framework is proposed on edge computing system for CAVs. This dissertation also uses the features' intrinsically small size to achieve real-time edge computing, without running the risk of congesting the network. In order to distinguish small sized objects such as pedestrian and cyclist in 3D data, an end-to-end multi-sensor fusion model is developed to implement 3D object detection from multi-sensor data. Experiments show that by solving multiple perception on camera and LiDAR jointly, the detection model can leverage the advantages from high resolution image and physical world LiDAR mapping data, which leads the KITTI benchmark on 3D object detection. At last, an application of cooperative perception is deployed on edge to heal the live map for autonomous vehicles. Through 3D reconstruction and multi-sensor fusion detection, experiments on real-world dataset demonstrate that a high definition (HD) map on edge can afford well sensed local data for navigation to CAVs.

abstract: Vehicular automation and autonomy are emerging fields that are growing at an exponential rate, expected to alter the very foundations of our transportation system within the next 10-25 years. A crucial interaction has been born out this new technology: Human and automated drivers operating within the same environment. Despite the well- known dangers of automobiles and driving, autonomous vehicles and their consequences on driving environments are not well understood by the population who will soon be interacting with them every day. Will an improvement in the understanding of autonomous vehicles have an effect on how humans behave when driving around them? And furthermore, will this improvement in the understanding of autonomous vehicles lead to higher levels of trust in them? This study addressed these questions by conducting a survey to measure participant’s driving behavior and trust when in the presence of autonomous vehicles. Participants were given several pre-tests to measure existing knowledge and trust of autonomous vehicles, as well as to see their driving behavior when in close proximity to autonomous vehicles. Then participants were presented with an educational intervention, detailing how autonomous vehicles work, including their decision processes. After examining the intervention, participants were asked to repeat post-tests identical to the ones administered before the intervention. Though a significant difference in self-reported driving behavior was measure between the pre-test and post- test, there was no significant relation found between improvement in scores on the education intervention knowledge check and driving behavior. There was also no significant relation found between improvement in scores on the education intervention knowledge check and the change in trust scores. These findings can be used to inform autonomous vehicle and infrastructure design as well as future studies of the effects of autonomous vehicles on human drivers in experimental settings. / Dissertation/Thesis / Masters Thesis Human Systems Engineering 2019

Title: Automatic On-Line Calibration and Calibration Monitoring of Cameras and Lidars Author: Jaroslav Moravec Department: Department of Software and Computer Science Education Supervisor: doc. RNDr. Elena Šikudová, Ph.D., Department of Software and Computer Science Education Abstract: Cameras and LiDARs are important devices in the automotive indus- try as their combination provides useful information (3D coordinates of a point, its colour and intensity) for perception, localization, mapping and prediction. Successful data fusion and interpretation requires accurate calibration of intrin- sic parameters of the sensors and their 6D relative pose. In this thesis, we present a target-less calibration method on three different calibration tasks. The solu- tion is based on a robust likelihood function constructed over the reprojection error of LiDAR edges relative to image edges. When the calibration slowly wears off, our online recalibration procedure can jointly follow the extrinsic calibration drift with an average error of 0.13◦ in rotation and 4 cm in translation. Based on this recalibration tool, we are also able to monitor the calibration and detect an abrupt decalibration in a couple of seconds. And we also present a fully automatic calibration routine that estimates both the extrinsic and intrinsic...

Deploying robots in unknown and complex areas for inspection tasks is becoming a real need for various application scenarios. Recently, there has been an increasing interest to develop and use autonomous aerial robots in environments such as urban voids and subterranean mine tunnels, aiming to decrease the human presence in dangerous or inaccessible areas. These areas are characterized by complete darkness and narrow tunnels, where the ground can often be rough and not traversible for mobile vehicles, thus the developments focus on Micro Aerial Vehicles (MAVs). MAVs are mechanically simple and agile platforms that can navigate through cluttered areas and have the potential to perform complex exploration tasks when equipped with proper onboard sensors. One of the key milestones in the development of autonomous robots is self-exploration. The definition of self-exploration according to [7] is "the act of moving through an unknown environment while building a map that can be used for subsequent navigation". By reaching this milestone, robots would be freed from the limitation of requiring already existing maps for navigation. In this thesis, a frontier-based exploration algorithm is established and evaluated to understand how such method could be used to reach the self-exploration milestone. By marking the border between what is known and unknown the method is able to determine the next desired position for the robot to expand the map. The resulting algorithm, together with a path planning method and 3-dimensional mapping framework, the method was tested and examined in simulated environments with different levels of complexity.

No description available.

Intelligent transportation systems (ITS) have become a popular method to enhance the safety and efficiency of transportation. Pedestrians, as an essential participant of ITS, are very vulnerable in a traffic collision, compared with the passengers inside the vehicle. In order to protect the safety of all traffic participants and enhance transportation efficiency, the novel autonomous vehicles are required to detect pedestrians accurately and timely. In the area of pedestrian detection, deep learning-based pedestrian detection methods have gained significant development since the appearance of powerful GPUs. A large number of researchers are paying efforts to improve the accuracy of pedestrian detection by utilizing the Convolutional Neural Network (CNN)-based detectors. In this thesis, we propose a one-stage anchor-free pedestrian detector named Bi-Center Network (BCNet), which is aided by the semantic features of pedestrians' visible parts. The framework of our BCNet has two main modules: the feature extraction module produces the concatenated feature maps that extracted from different layers of ResNet, and the four parallel branches in the detection module produce the full body center keypoint heatmap, visible part center keypoint heatmap, heights, and offsets, respectively. The final bounding boxes are converted from the high response points on the fused center keypoint heatmap and corresponding predicted heights and offsets. The fused center keypoint heatmap contains the semantic feature fusion of the full body and the visible part of each pedestrian. Thus, we conduct ablation studies and discover the efficiency of feature fusion and how visibility features benefit the detector's performance by proposing two types of approaches: introducing two weighting hyper-parameters and applying three different attention mechanisms. Our BCNet gains 9.82% MR-2 (the lower the better) on the Reasonable setup of the CityPersons dataset, compared to baseline model which gains 12.14% MR-2 . The experimental results indicate that the performance of pedestrian detection could be significantly improved because the visibility semantic could prompt stronger responses on the heatmap. We compare our BCNet with state-of-the-art models on the CityPersons dataset and ETH dataset, which shows that our detector is effective and achieves a promising performance.

No description available.