<p dir="ltr">Highway safety continues to pose a serious challenge to the social sustainability of transportation systems, and initiatives are being pursued at all levels of government to reduce the high fatality count of 42,000. At the same time, it is sought to ensure higher travel efficiency in order to increase economic productivity. The emergence of automated transportation provides great promise to mitigate these ills of the transportation sector that have persisted for so many decades. With regards to safety, such promise is rooted in the capability of autonomous vehicles to self-drive some or all of the time, thus reducing the impact of inherently errant human driving to which 95% of all crashes have been attributed. With regards to mobility, such promise is guided by the capability of the autonomous vehicle to carry out path planning, navigation, and vehicle controls in ways that are far more efficient than the human brain, thereby facilitating mobility and reducing congestion-related issues such as delay, emissions, driver frustration, and so on.</p><p dir="ltr">Unfortunately, the two key outcomes (safety and mobility) are reciprocal in the sense that navigation solutions that enhance safety generally tend to reduce mobility, and vice versa. As such, there is a need to assign values explicit to these performance criteria in order to develop balanced solutions for AV decisions. Most existing machine-learning-based path planning algorithms derive these weights using a learning approach. Unfortunately, the stability of these weights across time, individuals, and trip types, is not guaranteed. It is necessary to develop weights and processes that are trip situation-specific. Secondly, user trust in automation remains a key issue, given the relatively recent emergence of this technology and a few highly-publicized crashes, which has led to reservations among potential users.</p><p dir="ltr">To address these research questions, this thesis identifies various situational contexts of the problem, identifies the alternatives (the viable trajectories by fitting curves between the vehicle maneuver’s initial and final positions), develops the decision criteria (safety, mobility, comfort), carries out weighting of the criteria to reflect their relative significance, and scales the criteria to develop dimensionless equivalents of their raw values. Finally, a process for amalgamating the overall impacts of each driving decision alternative is developed based on the weighted and scaled criteria, to identify the best decision (optimal trajectory path). This multi-criteria decision making (MCDM) problem involves the collection of data through questionnaire surveys.</p><p dir="ltr">The weights obtained early in the MCDM process could be integrated into any one of two types of planning algorithms. First, they could be incorporated into interpolating curve-based planning algorithms, to identify the optimal trajectory based on human preferences. Additionally, they can be integrated into optimization-based planning algorithms to allocate weights to the various functions used.</p><p dir="ltr">Overall, this research aims to align the behavior of autonomous vehicles closely with human-driven vehicles, serving two primary purposes: first, facilitating their seamless coexistence on mixed-traffic roads and second, enhancing public acceptance of autonomous vehicles.</p>

The aim of this project was to compare two control algorithms designed to steer an autonomous vehicle. The comparison was made using a simulated environment to evaluate the performance of both controllers. The simulation used in this project was designed in Python and used an algorithm which randomly constructed roads from predefined road segments to create paths for the vehicle to follow. In this environment the Linear Time Varying (LTV)-Model Predictive Controller (MPC) and Pure Pursuit Controller (PPC) algorithms were evaluated. The thesis compared how well they follow paths, the average control cost of completing tasks, how well they handle input constraints, and the computational time for each algorithm. The data was collected by driving along three sets of randomly generated roads with both control algorithms. One set mostly straight, one with some turns, and one with mostly turns. An Analysis of Variance (ANOVA) test was used to make the comparison between the performance of the two algorithms. The results showed that both algorithms performed well. The PPC had low computation time and used less control, but it also had larger position errors. The LTV-MPC had higher computation time, but smaller position errors at the cost of larger control values. The conclusion is that the MPC is preferable if computational capabilities are available. Room for future work exists in the form of comparing additional controller types for autonomous vehicles and exploring different tuning parameters for the MPC controller. The simulation could also be expanded to more accurately reflect real world conditions. / Målet med detta projekt var att jämföra två kontrollalgoritmer avsedda för att styra en självkörande bil. Jämförelsen gjordes med hjälp av en simulering som utformades i Python. Den använde sig av en algoritm som slumpmässigt satte ihop vägar från förkonstruerade delar för att skapa banor för den självkörande bilen att följa. I denna miljö har vi testat två algoritmer, en LTV-MPC och en PPC. Vi jämförde hur pass väl de följer banor som skall likna riktiga vägar, hur mycket styrning de använder sig av för att bedöma energianvändning, hur väl de förhåller sig till begränsningar på acceleration och styrning, och den beräkningstiden som krävdes för att köra vår algoritm. Datan samlades genom att köra längs med tre grupper av slumpmässigt genererade vägar med båda kontrollalgoritmerna. En grupp innehöll huvudsakligen raka sträckor, en innehöll en del svängar, och en innehöll mycket svängar. ANOVA-testet användes för att göra jämförelsen mellan resultatet av dessa två algoritmer. Resultatet visade att båda algoritmer presterar väl. PPCn hade låg beräkningstid och mindre styrvärden, men större positionsfel. MPCn hade högre beräkningstid och större styrvärden, men mindre positionsfel. Slutsatsen är att MPCn är att föredra om beräkningsmöjligheterna finns tillgängliga. Det finns utrymme för framtida arbete i form av att jämföra fler kontrollalgoritmer och att utforska fler parameter justeringar för MPCn. Utöver det finns det även utrymme för en simulation som reflekterar verkligheten noggrannare.

This dissertation presents novel approaches to optimizing convolutional neural network (CNN) architectures for connected autonomous vehicle (CAV) workload on edge, tailored to surmount the challenges inherent in cooperative perception under the stringent resource constraints of edge devices (an endpoint on the network, the interface between the data center and the real world). Employing a modular methodology, this research utilizes the insights from granular examination of CAV perception workloads on edge platforms, identifying and analyzing critical bottlenecks. Through memory contention-aware neural architecture search (NAS), coupled with multi-objective optimization (MOO) and the Non-dominated Sorting Genetic Algorithm II (NSGA-II), this work dynamically optimizes CNN architectures, focusing on reducing memory cost, layer configuration and parameter optimization to reach set hardware constraints whilst maintaining a target precision performance. The results of this exploration are significant, achieving a 63% reduction in memory usage while maintaining a precision rate above 80% for CAV relevant object classes. This dissertation makes novel contributions to the field of edge computing in CAVs, offering a scalable and automated pipeline framework for dynamically obtaining an optimized model for given constraints, thus enabling CAV workloads on edge. In future research, this dissertation also opens multiple different venues for areas of integration. The modular aspect of the pipeline allows for security, privacy, scalability, and energy constraints to be added natively. Through detailed layer by layer analysis and refinement, this dissertation can ensure that CAVs can fully utilize any suitable edge device for the workload requested to realize autonomous driving for everyone.

The biggest shift in the automotive industry lies ahead. Autonomous vehicles create both curiosity and skepticism among drivers and people around. Autonomous vehicles, more specifically trucks, will not be utterly self-driving overnight. The whole transformation will take place in different phases. When a vehicle does not need a driver behind the wheel, new needs will arise. This is where this project comes into play. On behalf of Scania, a new driver’s seat shall be developed for new needs from the drivers for autonomous trucks of type 4. The project was carried out at Scania’s design department for cabin interiors. The project aimed to develop new needs for the future autonomous level 4 trucks in order to develop a driver’s seat that meets these needs. The project began with a planning phase in which the goals and the time frame for the project were set up. The project was then implemented in four different phases inspired by CDIO (n.d.). The work began with a benchmarking on existing trucks and passenger cars, but also on the future visions of different competitors regarding autonomous vehicles. Much work was put into understanding theories and interpreting relevant information. The users were used early in the project in the form of interviews, observations, and a survey that reached 299 truck drivers. The work then continued with various forms of brainstorming both within the project group and together with engineers from the group at Scania. The final work contained a CAD model of both prototype, CAID models of the final design, and a prototype scale of 1:1. The final result of the project is a new driver’s seat with the possibility of pushing the seat almost three times further back than the current seat. It can now be done when the driver’s seat is part of the bed. During the user study and the brainstorming, new needs were taken from the perspective of the sun being able to adapt to three different positions; rest, drive, work. The new driver’s seat now gives the drivers this opportunity. The result of this project may be more effective in driving the driver, which benefits both Scania and the customers in the form of the drivers being able to drive longer than previously allowed. / Det största skiften inom fordonsbranschen ligger framför oss. Autonoma fordon skapar både nyfikenhet och en skepsis bland förare och människor runt omkring. Autonoma fordon, mer specifikt lastbilar, kommer inte bli helt självkörande under en natt. Hela för- vandlingen kommer ske i olika faser. När ett fordon inte behöver en förare bakom ratten kommer nya behov uppkomma. Det är här det här projektet kommer in i bilden. På uppdrag av Scania, ska en ny förarstol utvecklas för nya behov från förarna för autonoma lastbilar av typen nivå 4. Projektet är ett examensarbete gjort av två studenter vid utbildningen civilingenjör inom teknisk design med inriktning produktutveckling, vid Luleå tekniska universitet. Projektet genomfördes på Scanias konstruktionsavdelning för hyttint- eriör. Målet för projektet var att ta fram nya behov för framtidens autonoma nivå 4 lastbilar för att sedan utveckla en förarstol som uppfyller dessa behov. Projektet började med en planeringsfas där målen och tidsramen för projektet sattes upp. Projektet genomfördes sedan i fyra olika faser inspirerade av CDIO (n.d.). Arbetet började med att en benchmarking gjordes på befintliga lastbilar och personbilar men även på olika konkurrenters framtidsvisioner gällande autonoma fordon. Mycket arbete lades på att förstå teorier och tolka relevant infor- mation. Användarna användes tidigt i projektet i form av intervjuer, observationer och en enkät som nådde ut till 299 lastbilsförare. Arbetet fortsatte sedan med olika former av brainstorming både inom projektgruppen och tillsammans med ingenjörer från gruppen på Scania. Slutgiltiga arbetet innehöll CAD-modeller av både prototyp, CAID-modeller av slutgiltig design samt en prototyp i skala 1:1. Det slutgiltiga resultatet av projektet är en ny förarstol med möjligheten att skjuta bak stolen nästan tre gånger längre än vad som tidigare var möjligt. Det kan nu göras då förarstolen är en del av sängen. Under användarstudien och brainstormingen togs nya behov fram i from av att solen ska ha möjlighet att anpassas till tre olika lägen; vila, köra, arbeta. Den nya förarstolen ger nu förarna den här möjligheten. Resultatet av det här projektet kan komma att effektivisera föraryrket, vilket gynnar både Scania och kunderna i form av att förarna kommer kunna köra längre än vad tidigare varit tillåtet.

The word “Perception” seems to be intuitive and maybe the most straightforward problem for the human brain because as a child we have been trained to classify images, detect objects, but for computers, it can be a daunting task. Giving intuition and reasoning to a computer which has mere capabilities to accept commands and process those commands is a big challenge. However, recent leaps in hardware development, sophisticated software frameworks, and mathematical techniques have made it a little less daunting if not easy. There are various applications built around to the concept of “Perception”. These applications require substantial computational resources, expensive hardware, and some sophisticated software frameworks. Building an application for perception for the embedded system is an entirely different ballgame. Embedded system is a culmination of hardware, software and peripherals developed for specific tasks with imposed constraints on memory and power. Therefore, the applications developed should keep in mind the memory and power constraints imposed due to the nature of these systems.Before 2012, the problems related to “Perception” such as classification, object detection were solved using algorithms with manually engineered features. However, in recent years, instead of manually engineering the features, these features are learned through learning algorithms. The game-changing architecture of Convolution Neural Networks proposed in 2012 by Alex K, provided a tremendous momentum in the direction of pushing Neural networks for perception. This thesis is an attempt to develop a convolution neural network architecture for embedded systems, i.e. an architecture that has a small model size and competitive accuracy. Recreate state-of-the-art architectures using fire module’s concept to reduce the model size of the architecture. The proposed compact models are feasible for deployment on embedded devices such as the Bluebox 2.0. Furthermore, attempts are made to integrate the compact Convolution Neural Network with object detection pipelines.

Cette thèse est consacrée à étudier les applications potentielles de véhicules autonomes et communications V2X pour construire les systèmes de transport intelligents. Premièrement, le comportement de caravane dans un environnement de véhicule connecté est étudié. Un algorithme de commande de caravane est conçu pour obtenir l'espacement sécuritaire ainsi que la conformité de la vitesse et de l'accélération. Deuxièmement, à plus grande échelle, les caravanes autour d'une intersection sont considérées. Le débit pendant une période de signal de trafic peut être amélioré en tirant profit de la capacité redondante de la route. Dans diverses contraintes, les véhicules peuvent choisir d'accélérer et rejoindre la caravane précédente ou à décélérer de déroger à l'actuel. Troisièmement, une intersection sans signalisation en VANET est considérée. Dans des conditions de faible trafic, les véhicules peuvent réguler leur vitesse avant d'arriver à l'intersection en fonction du temps d'occupation de la zone de conflit (TOZC) stocké au niveau du gestionnaire, afin qu'ils puissent traverser l'intersection sans collision ni arrêt. Le délai peut être réduit en conséquence. Enfin, un algorithme de gestion d'intersection autonome universelle, qui peut fonctionner même avec le trafic lourd, est développé. Le véhicule cherche à sécuriser les fenêtres entrant dans le TOZC. Ensuite, sur la base des fenêtres trouvées et le mouvement du véhicule qui précède, les trajectoires des véhicules peuvent être planifiées en utilisant une méthode de programmation dynamique segmentée. Tous les algorithmes conçus sont testés et vérifiés avec succès par des simulations dans scénarios différents / This thesis is devoted to study the potential applications of autonomous vehicles and V2X communications to construct the intelligent transportation systems. Firstly, the behavior of platoon in connected vehicle environment is studied. A platoon control algorithm is designed to obtain safe spacing as well as accordance of velocity and acceleration for vehicles in the same lane. Secondly, in larger scale, the platoons around an intersection are considered. The throughput in a traffic signal period can be improved by taking advantage of the redundant road capacity. Within diverse constraints, vehicles can choose to accelerate to join in the preceding platoon or to decelerate to depart from the current one. Thirdly, an unsignalized intersection in VANET is considered. In light traffic conditions, vehicles can regulate their velocities before arriving at the intersection according to the conflict zone occupancy time (CZOT) stored at the manager, so that they could get through the intersection without collision or stop. The delay can be reduced accordingly. Finally, an universal autonomous intersection management algorithm, which can work even with heavy traffic, is developed. The vehicle searches for safe entering windows in the CZOT. Then based on the found windows and the motion of preceding vehicle, the trajectories of vehicles can be planned using a segmented dynamic programming method. All the designed algorithms are successfully tested and verified by simulations in various scenarios

In this thesis the cubic spline guidance law and its variants are derived. A detailed analysis is carried out to find the initial conditions for successful interception. The results are applied to three dimensional guidance design and for solving waypoint following problems. The basic cubic spline guidance law is derived for intercepting a stationary target at a desired impact angle in a surface-to-surface engagement scenario. The guidance law is obtained using an inverse method, from a cubic spline curve based trajectory. For overcoming the drawbacks of the basic cubic spline guidance law, it is modified by introducing an additional parameter. This modification has an interesting feature that the guidance command can be obtained using a single cubic spline polynomial even for impact angles greater than π/2, while resulting in substantial improvement in the guidance performance in terms of lateral acceleration demand and length of the trajectory. For imparting robustness to the cubic spline guidance law, in the presence of uncertainties and acceleration saturation, an explicit guidance expression is also derived. A comprehensive capturability study of the proposed guidance law is carried out. The capturability for the cubic spline guidance law is defined in terms of the set of all feasible initial conditions for successful interception. This set is analytically derived and its dependence on various factors, such as initial engagement geometry and interceptor capability, are also established. The basic cubic spline guidance and its variants are also derived for a three dimen- sional scenario. The novelty of the present work lies in the particular representation of the three dimensional cubic spline curve and the adoption of the analytical results available for two dimensional cubic spline guidance law. This enables selection of the boundary condition at launch for given terminal boundary condition and also in avoiding the singularities associated with the inverse method based guidance laws. For establishing the feasibility of the guidance laws in the real world, the rigid body dynamics of the interceptor is presented as a 6 degrees-of-freedom model. Further, using a simplified model, elementary autopilots are also designed. The successful interception of the target in the presence of the rigid body dynamics proves practical applicability of the cubic spline based guidance laws. Finally, the theory developed in the first part of the thesis is applied to solve the waypoint following problem. A smooth path is designed for transition of vehicle velocity from incoming to outgoing direction. The approach developed is similar to Dubins’ path, as it comprises line–cubic spline–line segments. The important feature of this method is that the cubic spline segments are fitted such that the path curvature is bounded by a pre-specified constrained value and the acceleration demand for following the smooth path obtained by this method, gradually increases to the maximum value and then decreases. This property is advantageous from a practical point of view. All the results obtained are verified with the help of numerical simulations which are included in the thesis. The proposed cubic spline guidance law is conceptually simple, does not use linearised kinematic equations, is independent of time-to-go es- timates, and is also computationally inexpensive.

Tekniska innovationer förknippas alltmer med modernt, utvecklade IT-komponenter i dagens samhällsutveckling. Detta framträder framförallt inom fordonsindustrin där självkörande/automatiserade fordon tagit ett rejält språng på den globala fordonsmarknaden. Automatiserade fordon anses av många som framtidens teknik inom områden som framförallt berör kollektivtrafik och varutransporter, men visionen om en helt automatiserad trafik lämnas inte oberörd. Men då ny, främmande teknologi även kan ses som avskräckande för gemene man - vad krävs då för att ändra på detta synsätt? Hur kan det säkerställas att tekniska innovationer såsom automatiserade fordon accepteras av allmänheten och används i praktiken? Dessa är frågor som legat till grund för föreliggande rapport som syftat till att presentera en studie som undersöker vilka faktorer som har en inverkan på användningen av automatiserade fordon. Detta har genomförts genom en omfattande litteratursökning av tidigare och aktuell forskning inom ämnet, samt intervjuer och enkätundersökningar för att ta del av synpunkter och erfarenheter från befintliga användare av automatiserad teknologi. Forskningen har resulterat i en kunskapsprodukt som utgörs av en tabell innehållandes ovan nämnda faktorer med en kort beskrivning, typ av faktor, samt en motivering till varför de bedömts till att ha en inverkan på användningen av automatiserade fordon. Tabellen har därtill kompletterats av en relationsmodell som beskriver hur faktorer samverkar och är beroende av varandra. Resultatet ska med förhoppningar kunna användas som stöd till framtida studier och forskningsprojekt, samt ge inspiration till företag och organisationer som bedriver verksamhet med koppling till automatiserade fordon och dess användning. / Technological innovations are increasingly associated with modern, developed IT-capabilities in today’s societal evolution. This is seen primarily in the automobile industry where self-driving/automated vehicles has taken a great leap on the global automobile market. Automated vehicles are considered by many to be the technology of the future which mainly involves public transport and transport of goods, but the vision of a completely automated traffic is not left untouched. But because new, foreign technology also can be seen as detrimental to people - what is then required to change this approach? How can it be ensured that technical innovations such as automated vehicles are accepted by the public and used in practice? These are issues that gave rise to the following report with a purpose to present a study that investigates which factors that has an impact on the use of automated vehicles. This has been accomplished through an extensive literature search of previous and current research of the topic, along with interviews and surveys to acquire personal opinions and experiences from existing users of automated technology. The research has eventuated in a knowledge product represented by a table of above mentioned factors including a brief description, type of factor, and a justification to as why they have been assessed as having an impact on the use of automated vehicles. Furthermore, the table has been supplemented with a relational model that describes how factors interact and depend on each other. Hopefully, the result can be used as support for future studies and research projects, as well as inspire companies and organizations engaged in automated vehicles and their use.

La plupart des logiciels contrôlant les véhicules intelligents traite de la compréhension de la scène. De nombreuses méthodes existent actuellement pour percevoir les obstacles de façon automatique. La majorité d’entre elles emploie ainsi les capteurs extéroceptifs comme des caméras ou des lidars. Cette thèse porte sur les domaines de la robotique et de la fusion d’information et s’intéresse aux systèmes d’information géographique. Nous étudions ainsi l’utilité d’ajouter des cartes numériques, qui cartographient le milieu urbain dans lequel évolue le véhicule, en tant que capteur virtuel améliorant les résultats de perception. Les cartes contiennent en effet une quantité phénoménale d’information sur l’environnement : sa géométrie, sa topologie ainsi que d’autres informations contextuelles. Dans nos travaux, nous avons extrait la géométrie des routes et des modèles de bâtiments afin de déduire le contexte et les caractéristiques de chaque objet détecté. Notre méthode se base sur une extension de grilles d’occupations : les grilles de perception crédibilistes. Elle permet de modéliser explicitement les incertitudes liées aux données de cartes et de capteurs. Elle présente également l’avantage de représenter de façon uniforme les données provenant de différentes sources : lidar, caméra ou cartes. Les cartes sont traitées de la même façon que les capteurs physiques. Cette démarche permet d’ajouter les informations géographiques sans pour autant leur donner trop d’importance, ce qui est essentiel en présence d’erreurs. Dans notre approche, le résultat de la fusion d’information contenu dans une grille de perception est utilisé pour prédire l’état de l’environnement à l’instant suivant. Le fait d’estimer les caractéristiques des éléments dynamiques ne satisfait donc plus l’hypothèse du monde statique. Par conséquent, il est nécessaire d’ajuster le niveau de certitude attribué à ces informations. Nous y parvenons en appliquant l’affaiblissement temporel. Étant donné que les méthodes existantes n’étaient pas adaptées à cette application, nous proposons une famille d’opérateurs d’affaiblissement prenant en compte le type d’information traitée. Les algorithmes étudiés ont été validés par des tests sur des données réelles. Nous avons donc développé des prototypes en Matlab et des logiciels en C++ basés sur la plate-forme Pacpus. Grâce à eux nous présentons les résultats des expériences effectués en conditions réelles. / This thesis is situated in the domains of robotics and data fusion, and concerns geographic information systems. We study the utility of adding digital maps, which model the urban environment in which the vehicle evolves, as a virtual sensor improving the perception results. Indeed, the maps contain a phenomenal quantity of information about the environment : its geometry, topology and additional contextual information. In this work, we extract road surface geometry and building models in order to deduce the context and the characteristics of each detected object. Our method is based on an extension of occupancy grids : the evidential perception grids. It permits to model explicitly the uncertainty related to the map and sensor data. By this means, the approach presents also the advantage of representing homogeneously the data originating from various sources : lidar, camera or maps. The maps are handled on equal terms with the physical sensors. This approach allows us to add geographic information without imputing unduly importance to it, which is essential in presence of errors. In our approach, the information fusion result, stored in a perception grid, is used to predict the stateof environment on the next instant. The fact of estimating the characteristics of dynamic elements does not satisfy the hypothesis of static world. Therefore, it is necessary to adjust the level of certainty attributed to these pieces of information. We do so by applying the temporal discounting. Due to the fact that existing methods are not well suited for this application, we propose a family of discoun toperators that take into account the type of handled information. The studied algorithms have been validated through tests on real data. We have thus developed the prototypes in Matlab and the C++ software based on Pacpus framework. Thanks to them, we present the results of experiments performed in real conditions.

<div><p>Analyzing behaviors of objects on the road is a complex task that requires data from various sensors and their fusion to recreate 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.</p> <p> 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. </p> <p>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. </p><br></div>