<p>Today, computer science and robotics have been highly developed. Simultaneous Localization and Mapping (SLAM) is widely used in mobile robot navigation, game design, and autonomous vehicles. It can be said that in the future, most scenarios where mobile robots are applied will require localization and mapping. Among them, the construction of three-dimensional(3D) maps is particularly important for environment visualization which is the focus of this research.</p> <p>In this project, the data used for visualization was collected using a vision sensor. The data collected by the vision sensor is processed by ORB-SLAM2 to generate the 3D cloud point maps of the environment. Because, there are a lot of noise in the map points cloud, filters are used to remove the noise. The generated map points are processed by the straight-through filter to cut off the points out of the specific range. Statistical filters are then used to remove sparse outlier noise. Thereafter, in order to improve the calculation efficiency and retain the necessary terrain details, a voxel filter is used for downsampling. In order to improve the composition effect, it is necessary to appropriately increase the sampling amount to increase surface smoothness. Finally, the processed map points are visualized using Octomap. The implementation utilizes the services provided by the Robot Operating System (ROS). The powerful Rviz software on the ROS platform is used. The processed map points as cloud data are published in ROS and visualized using Octomap. </p> <p>Simulation results confirm that Octomap can show the terrain details well in the 3D visualization of the environment. After the simulations, visualization experiments for two environments of different complexity are performed. The experimental results show that the approach can mitigate the influence of noise on the visualization results to a certain extent. It is shown that for static high-precision point clouds, Octomap provides a good visualization. The simulation and experimental results demonstrate the applicably of the approach to visualize 3D map points for the purpose of autonomous navigation.</p><br>

The graph-based formulation of the navigation problem is establishing itself as one of the standard ways to formulate the navigation problem within the sensor fusion community. It enables a convenient way to access information from previous positions which can be used to enhance the estimate of the current position.To restrict working memory usage, map partitioning can be used to store older parts of the map on a hard drive, in the form of submaps. This limits the number of previous positions within the active map. This thesis examines the effect that map partitioning information loss has on the state of the art positioning algorithm iSAM2, both in open routes and when loop closure is achieved. It finds that larger submaps appear to cause a smaller positional error than smaller submaps for open routes. The smaller submaps seem to give smaller positional error than larger submaps when loop closure is achieved. The thesis also examines how the density of landmarks at the partition point affects the positional error, but the obtained result is mixed and no clear conclusions can be made. Finally it reviews some loop closure detection algorithms that can be convenient to pair with the iSAM2 algorithm.

Käppalaverket är ett avloppsreningsverk som är beläget på Lidingö. Avloppsreningsverket renar idag avloppsvatten från över en halv miljon människor. Varje år produceras det omkring 30 000 ton slam på Käppalaverket. Det näringsrika slammet används sedan som gödsel som sprids på åkermarken. Från 2007 fram till 2012 har kopparhalten stigit i det producerade slammet på Käppalaverket. Det har resulterat i att mängden koppar har överstigit det uppsatta gränsvärdet för att få sprida slam på åkermark (300 g koppar/ha och år). Därmed får endast en begränsad mängd av det producerade slammet spridas på åkermarken. Gränsvärdet är satt för att skyddamark- och vattenlevande organismer. Den tvåvärda kopparjonen har nämligen en toxisk effekt på organismerna. Även Bromma och Henriksdals avloppsreningsverk har haft en liknande ökning i kopparhalt hos slammet, men inte lika kraftig som Käppalaverket. Den största delen av den koppar som hamnar i slammet förutspås komma från kopparrör i hushållen. Därför har flera olika parametrar undersökts för att se hur mycket påverkan de har på korrosionen av kopparrören. pH, alkalinitet och naturligt organiskt material (NOM) är de parametrar som mest påverkar mängden koppar som löses ut genom korrosion i dricksvattnet. En sänkning av pH leder till högre halt koppar i dricksvattnet, men en sänkning av alkalinitet och NOM-halt bidrar istället till en lägre halt koppar i dricksvattnet. Även temperatur och stilleståndstid har betydelse för hur snabbt korrosionen av koppar sker. Majoriteten av Käppalas medlemskommuner får sitt dricksvatten från Görvälnverket. Lovö och Norsborgs vattenverklevererar dricksvatten till övriga kommuner inom Käppalas upptagningsområde. Hos Görvälnverket har alkaliniteten stigit svagt från år 2000 fram till idag. Denna stigning borde endast påverka i ytterst liten grad på den höga kopparhalten i Käppalaverkets slam. På övriga parametrar som anses ha påverkan på kopparhalten i slammet har det inte setts någon tydlig tendens. Ökningen av kopparhalter i slammet tyder istället på andra orsaker. I februari 2009 bytte Käppalaverket tillsammans med Bromma och Henriksdals reningsverk analyslaboratorium. Bytet av laboratorium tyckts påverka analysresultaten så att analysresultaten från det nya laboratoriet uppvisar en högre halt koppar i slammet. Dessutom har standardavvikelsen varit större för de analysresult som erhållits efter laboratoriebytet jämfört med analysresultaten före laboratoriebytet. Analysosäkerheten har emellertid varierat mellan åren då kopparhalten islammet steg. På Stockholm Vattens laboratorium dit proverna skickades innan laboratoriebytet var analysosäkerheten väldigt ospecifikt angivet. Enligt analysrapporten hade provresultatet en analysosäkerhet mellan 15-40 %. Proverna skickades efter laboratoriebytet till Eurofinslaboratorium i Lidköping. Där har analysosäkerheten på slamproverna minskat med tiden. Första tiden efter bytet var analysosäkerheten på proverna 30 % därefter har osäkerheten minskat till 15 %. Anledningen till att Käppalaverket haft en kraftigare stigning än Bromma och Henriksdalsreningsverk är oviss. En förklaring skulle kunna vara att slamkonditioneringsprocessen Kemicond påverkade analysresultatet eftersom en viss koncentrationsförändring skulle skett då processen var i drift. Men kopparhalten har inte sjunkit så kraftigt som förväntat efter att Kemicond avvecklades. Emellertid har den totala mängden koppar in till Käppalaverket ökat under tidsperioden då kopparhalten i slammet varit högre. Det tyder på att det är någon annan bakomliggande orsak till ökningen av koppar i slammet. / Käppala wastewater treatment plant (WWTP) is located in Lidingö and currently treats water from over half a million people. Every year the Käppala plant produces about 30 000 t sludge. The nutritious sludge is then used as fertilizer on farmland. From 2007 until 2012 the copper content has increased in the produced sludge. Eventually the amount of copper has exceeded the allowed limit for spreading sludge on farmland (300 g copper/ha and year). So only a limited amount of sludge can be spread on the farmland. The limit is set to protect the soil and aquatic organisms since the divalent copper ion has toxic effects on the organisms. Even Bromma and Henriksdal WWTP have had the same increase in copper content in the sludge, but not as highas the Käppala plant. The largest part of the copper that ends up in the sludge is predicted to originates from copper pipes in households. That is the reason why a number of different parameters has been investigated in order to see how much each parameter affects the amount of copper that is dissolved in the drinking water. The parameters that affect the most are; pH, alkalinity andnatural organic matter (NOM). When pH is decreasing, more copper will dissolve to the drinking water, but when the alkalinity and the NOM content is increasing, more copper will dissolve to the drinking water. Even the temperature and the stagnation time is important for how fast corrosion of copper pipes will occurs. The Görväln plant purifies and supplies drinking water to the majority of the member municipalities of the Käppala plant. Lovö’s and Norsborg’swaterworks supply drinking water to the other municipalities within the catchment area of Käppala. The alkalinity has slightly increased in the Görväln plant from 2000 to present. This increase should only affect in a very small degree on the increase in copper content in the sludge of Käppala plant. There is no clear trend on the other parameters that is considered to have an influence on the copper content in the sludge. The sharp increase in copper content has probably other causes. In February 2009, the Käppala plant, together with Bromma and Henriksdal WWTP, changed analysis laboratory. The change of laboratory seemed to affect the results of the analysis giving higher values of copper contentin the sludge. The standard deviation is also greater in the analysis results of the samples after the change of laboratory compared to with the analysis results before the change of laboratory. However, the uncertainty of the analysis has varied over the years when the rising in the copper content in the sludge occurred. At Stockholm Vatten’s laboratory where samples were sent before the laboratory change was the uncertainty of the analysis very unspecific given. According to the analysis report, the result of the sample had an uncertainty of the analysis between 15-40 %. The samples were after the laboratory change sent to Eurofins laboratory in Lidköping. Here the uncertainty of the analysis has been diminished with the time. After the first years after the change the uncertainty of the analysis was 30 %. But after that the uncertainty of the analysis has declined to 15 %. The reason why the Käppala plant had a stronger increase of copper content in the sludge than in Bromma and Henriksdal WWTP is unknown. One explanation could be that the sludge conditioning process Kemicond affected the analysis result since a certain change in concentration could have occurred when the process was in operation. However, the copper content has not decreased as much as expected after Kemicond was liquidated. But the total amount of copper in to the Käppala plant has increased during the period when the copper content in the sludge was higher. This indicates that there are other underlying reasons for the increase of copper in the sludge.

There has been an increasing interest in visual sensors and vision-based solutions for single and multi-robot systems. Vision-based sensors, e.g., traditional RGB cameras, grant rich semantic information and accurate directional measurements at a relatively low cost; however, such sensors have two major drawbacks. They do not generally provide reliable depth estimates, and typically have a limited field of view. These limitations considerably increase the complexity of controlling multiagent systems. This thesis studies some of the underlying problems in vision-based multiagent control and mapping. The first contribution of this thesis is a method for restoring bearing rigidity in non-rigid networks of robots. We introduce means to determine which bearing measurements can improve bearing rigidity in non-rigid graphs and provide a greedy algorithm that restores rigidity in 2D with a minimum number of added edges. The focus of the second part is on the formation control problem using only bearing measurements. We address the control problem for consensus and formation control through non-smooth Lyapunov functions and differential inclusion. We provide a stability analysis for undirected graphs and investigate the derived controllers for directed graphs. We also introduce a newer notion of bearing persistence for pure bearing-based control in directed graphs. The third part is concerned with the bearing-only visual homing problem with a limited field of view sensor. In essence, this problem is a special case of the formation control problem where there is a single moving agent with fixed neighbors. We introduce a navigational vector field composed of two orthogonal vector fields that converges to the goal position and does not violate the field of view constraints. Our method does not require the landmarks' locations and is robust to the landmarks' tracking loss. The last part of this dissertation considers outlier detection in pose graphs for Structure from Motion (SfM) and Simultaneous Localization and Mapping (SLAM) problems. We propose a method for detecting incorrect orientation measurements before pose graph optimization by checking their geometric consistency in cycles. We use Expectation-Maximization to fine-tune the noise's distribution parameters and propose a new approximate graph inference procedure specifically designed to take advantage of evidence on cycles with better performance than standard approaches. These works will help enable multi-robot systems to overcome visual sensors' limitations in collaborative tasks such as navigation and mapping.

Localization is critical for self-awareness of any autonomous system and is an important part of the autonomous system stack which consists of many phases including sensing, perceiving, planning and control. In the sensing phase, data from on board sensors are collected, preprocessed and passed to the next phase. The perceiving phase is responsible for self awareness or localization and situational awareness which includes multi-objects detection and scene understanding. After the autonomous system is aware of where it is and what is around it, it can use this knowledge to plan for the path it can take and send control commands to pursue this path. In this proposal, we focus on the localization part of the autonomous stack using camera images. We deal with the localization problem from different perspectives including single images and videos. Starting with the single image pose estimation, our approach is to propose systems that not only have good localization accuracy, but also have low space and time complexity. Firstly, we propose SurfCNN, a low cost indoor localization system that uses SURF descriptors instead of the original images to reduce the complexity of training convolutional neural networks (CNN) for indoor localization application. Given a single input image, the strongest SURF features descriptors are used as input to 5 convolutional layers to find its absolute position and orientation in arbitrary reference frame. The proposed system achieves comparable performance to the state of the art using only 300 features without the need for using the full image or complex neural networks architectures. Following, we propose SURF-LSTM, an extension to the idea of using SURF descriptors instead the original images. However, instead of CNN used in SurfCNN, we use long short term memory (LSTM) network which is one type of recurrent neural networks (RNN) to extract the sequential relation between SURF descriptors. Using SURF-LSTM, We only need 50 features to reach comparable or better results compared with SurfCNN that needs 300 features and other works that use full images with large neural networks. In the following research phase, instead of using SURF descriptors as image features to reduce the training complexity, we study the effect of using features extracted from other CNN models that were pretrained on other image tasks like image classification without further training and fine tuning. To learn the pose from pretrained features, graph neural networks (GNN) are adopted to solve the single image localization problem (Pose-GNN) by using these features representations either as features of nodes in a graph (image as a node) or converted into a graph (image as a graph). The proposed models outperform the state of the art methods on indoor localization dataset and have comparable performance for outdoor scenes. In the final stage of single image pose estimation research, we study if we can achieve good localization results without the need for training complex neural network. We propose (Linear-PoseNet) by which we can achieve similar results to the other methods based on neural networks with training a single linear regression layer on image features from pretrained ResNet50 in less than one second on CPU. Moreover, for outdoor scenes, we propose (Dense-PoseNet) that have only 3 fully connected layers trained on few minutes that reach comparable performance to other complex methods. The second localization perspective is to find the relative poses between images in a video instead of absolute poses. We extend the idea used in SurfCNN and SURF-LSTM systems and use SURF descriptors as feature representation of the images in the video. Two systems are proposed to find the relative poses between images in the video using 3D-CNN and 2DCNN-RNN. We show that using 3D-CNN is better than using the combination of CNN-RNN for relative pose estimation. / Graduate

<p>While traditional rigid robots are widely used in almost all applications today, their rigidity restricts the use of these robots in environments where interaction with the surroundings or humans is inevitable. This is where soft robots come into play. Due to their compliant and adaptable nature, these robots can safely interact with humans and traverse through unpredictable, cluttered environments. This research focuses on the navigation of a special class of soft growing robots called Vine robots. Vine robots can easily maneuver through tight spaces and rough terrain and have an added advantage of speed over general soft robots. In this work, we develop a model which localizes the Vine robot in an unknown surrounding by giving us the position of the tip of the robot at every instant. The model exploits the passive steering of growing robots using obstacle aided navigation. The robot is sensorized to record the orientation of the its tip and the total length it has grown to. This data along with the force generated on collision with the environment is used to localize the robot in space. The localization model is implemented using the sensor data. The accuracy of this model is then verified by comparing the tip position of the robot we have calculated with its predicted position and the actual position as measured by an overhead camera. It is concluded that the robot can be localized in an environment with a maximum error of 7.65 cm (10\%) when the total length the robot has grown to is 170 cm. </p>

La surveillance des émissions de polluants dans l’atmosphère constitue pour les industriels une problématique environnementale de premier ordre. Qu’elles soient ponctuelles (rejet de polluant par une cheminée) ou fugitives (fuites accidentelles de canalisations ou de stockages), la connaissance et la maîtrise de ces émissions est aujourd’hui nécessaire pour quantifier et réduire les cas échéant leur impact environnemental. Dans ce contexte, la modélisation de la dispersion atmosphérique est un outil d’analyse intéressant, permettant la surveillance d’un site industriel et la cartographie des concentrations autour du site. L’objectif de cette thèse était de développer un code de calcul opérationnel assurant le suivi des polluants sur un site industriel, en champ proche (prise en compte de la complexité du bâti) et avec des temps de calcul avoisinant le temps réel. Nous avons, au cours de ce travail de recherche, développé une approche appeléeFlow’Air-3D basée sur la constitution, en amont de toute situation opérationnelle, d’une base de données de champs de vent CFD calculés sur le site industriel étudié. En situation opérationnelle, la dispersion des polluants est modélisée avec un code de dispersion lagrangien, SLAM, également développé dans le cadre de cette thèse. Pour pouvoir mettre en place cette approche Flow’Air-3D, nous avons développé une méthodologie et une paramétrisation spécifique du modèle RANS-k-e pour représenter une couche limite de surface diabatique. Nous avons ensuite identifié les paramètres nécessaires à la construction de la base de données, ainsi que l’influence de la discrétisation et de l’interpolation de ces paramètres sur les champs de vent déterminés par cette approche. Finalement un code de dispersion lagrangien stochastique à particules, utilisant les champs de vent de la base de données, a été développé et partiellement validé sur quelques cas académiques simples (condition de mélange homogène, comparaison à la dispersion d’une bouffée gaussienne, etc.)Des essais en soufflerie, une approche eulérienne (effectuée avec FLUENT 6.3) et une première application de la méthodologie Flow’Air-3D/SLAM ont été menés sur le site pétrochimique de la raffinerie de Feyzin. Les comparaisons effectuées entre ces trois approches montrent le bon comportement du modèle SLAM. Les temps CPU mis en œuvre pour réaliser les calculs de dispersion lagrangien sont encourageants et montrent la faisabilité de notre approche sur un cas applicatif réel. / The monitoring of pollutant’s emissions in the atmosphere constitutes for the industrialists a main environmental issue. That they are punctual (emissions of pollutants by a chimney) or fugitive (accidental releases of drains or storages), the knowledge and the control of these emissions are important data to quantify and reduce their environmental impact. In this context, the modeling of atmospheric dispersion is an interesting tool of analysis for the monitoring of a site and this thesis has permitted the creation of an operational code ensuring the follow-up of pollutants on an industrial site, in close field and with computing times bordering the real time. We have, during this research, developed an approach called Flow'Air-3D consisting to create before any operational situation, a data base of CFD wind fields calculated on the studied industrial site. In operational situation a lagrangian code of dispersion, SLAM, developed in the thesis will ensure calculations of dispersion in a few seconds. To be able to set up this Flow' Air-3D approach, we have developed, during this thesis, a methodology and a specific parameterization of the k-e model to represent the atmospheric boundary layer with a CFD approach. Then, we have identified the parameters necessary to build the data base, as well as the influence of the discretization and the interpolation of these parameters on the wind fields resulting from this base. Finally a lagrangian code of dispersion (SLAM) using the precalculated wind fields of the data base was developed and partially validated on simple academic cases (well-mixed condition criteria, comparison with the dispersion of a Gaussian puff, etc.).Tests in wind tunnel, a eulerian approach (done with FLUENT 6.3) and a first application of the Flow' Air-3D/SLAM methodology were carried out on the petrochemical site of the refinery of Feyzin. The comparisons between these three approaches show the good behaviour of the model SLAM. The CPU times for the calculations of lagrangian dispersion are encouraging and show the feasibility of our approach on a real case.

Traditionellt sett har poesin varit talad, inte skriven, något som framförts på scen framför en publik. Den politiska användningen av estradpoesi är inget nytt påfund. Brukspoesi och tillfällespoesi i flera olika former har använts för att övertyga eller övertala publiken om olika samhällspolitiska frågor. Det retoriska talet har fått mycket uppmärksamhet för sin starka relation till politik, men hur fungerar det när talet är poesi? Denna uppsats vill fördjupa förståelsen av poesi som framförs på scen och hur den fungerar som en politisk handling. Detta görs genom att undersöka en specifik poesigenre: Poetry slam. Poetry slam är en tävling i estradpoesi som under de senaste trettio åren växt fram världen över, och många utnyttjar sin tid framför micken till att ta upp politiska åsikter eller personliga berättelser med en politisk agenda. Uppsatsen undersöker hur åtta olika poeter som tävlade i SM i Poetry slam 2019 använder sig av Poetry slam som en politisk plattform. Den undersöker på vilket sätt dikterna tar upp aktuella samhällspolitiska frågor, hur dessa frågor gestaltas samt hur olika strategiska manövreringar väcker identifikation hos publiken. Då dikterna måste vara skrivna av poeterna själva, men inte måste vara baserade på egna erfarenheter, blir det tydligt att poeterna genom dikterna konstruerar en identitet. Uppsatsen undersöker därför även hur detta används som ett medel för att övertyga eller övertala.

The aim of the thesis is to apply simultaneous localization and mapping (SLAM) to automated guided vehicles (AGVs) in a Robot Operating System (ROS) environment. Different sensor setups are used and evaluated. The SLAM applications used is the open-source solution Cartographer as well as Intel's own commercial SLAM in their T265 tracking camera. The different sensor setups are evaluated based on how well the localization will give the exact pose of the AGV in comparison to another positioning system acting as ground truth.

Indiana University-Purdue University Indianapolis (IUPUI) / Analyzing behaviors of objects on the road is a complex task that requires data from various sensors and their fusion to recreate the movement of objects with a high degree of accuracy. A data collection and processing system are thus needed to track the objects accurately in order to make an accurate and clear map of the trajectories of objects relative to various coordinate frame(s) of interest in the map. Detection and tracking moving objects (DATMO) and Simultaneous localization and mapping (SLAM) are the tasks that needs to be achieved in conjunction to create a clear map of the road comprising of the moving and static objects. These computational problems are commonly solved and used to aid scenario reconstruction for the objects of interest. The tracking of objects can be done in various ways, utilizing sensors such as monocular or stereo cameras, Light Detection and Ranging (LIDAR) sensors as well as Inertial Navigation systems (INS) systems. One relatively common method for solving DATMO and SLAM involves utilizing a 3D LIDAR with multiple monocular cameras in conjunction with an inertial measurement unit (IMU) allows for redundancies to maintain object classification and tracking with the help of sensor fusion in cases when sensor specific traditional algorithms prove to be ineffectual when either sensor falls short due to their limitations. The usage of the IMU and sensor fusion methods relatively eliminates the need for having an expensive INS rig. Fusion of these sensors allows for more effectual tracking to utilize the maximum potential of each sensor while allowing for methods to increase perceptional accuracy. The focus of this thesis will be the dock-less e-scooter and the primary goal will be to track its movements effectively and accurately with respect to cars on the road and the world. Since it is relatively more common to observe a car on the road than e-scooters, we propose a data collection system that can be built on top of an e-scooter and an offline processing pipeline that can be used to collect data in order to understand the behaviors of the e-scooters themselves. In this thesis, we plan to explore a data collection system involving a 3D LIDAR sensor and multiple monocular cameras and an IMU on an e-scooter as well as an offline method for processing the data to generate data to aid scenario reconstruction.