Timing-Based Localization using Multipath Information PDF Download
Are you looking for read ebook online? Search for your book and save it on your Kindle device, PC, phones or tablets. Download Timing-Based Localization using Multipath Information PDF full book. Access full book title Timing-Based Localization using Multipath Information by Andreas Bergström. Download full books in PDF and EPUB format.
Author: Andreas Bergström
Publisher: Linköping University Electronic Press
ISBN: 9179299172
Category :
Languages : en
Pages : 140
Get Book Here
Book Description
The measurements of radio signals are commonly used for localization purposes where the goal is to determine the spatial position of one or multiple objects. In realistic scenarios, any transmitted radio signal will be affected by the environment through reflections, diffraction at edges and corners etc. This causes a phenomenon known as multipath propagation, by which multiple instances of the transmitted signal having traversed different paths are heard by the receiver. These are known as Multi-Path Components (MPCs). The direct path (DP) between transmitter and receiver may also be occluded, causing what is referred to as non-Line-of-Sight (non-LOS) conditions. As a consequence of these effects, the estimated position of the object(s) may often be erroneous. This thesis focuses on how to achieve better localization accuracy by accounting for the above-mentioned multipath propagation and non-LOS effects. It is proposed how to mitigate these in the context of positioning based on estimation of the DP between transmitter and receiver. It is also proposed how to constructively utilize the additional information about the environment which they implicitly provide. This is all done in a framework wherein a given signal model and a map of the surroundings are used to build a mathematical model of the radio environment, from which the resulting MPCs are estimated. First, methods to mitigate the adverse effects of multipath propagation and non-LOS conditions for positioning based on estimation of the DP between transmitter and receiver are presented. This is initially done by using robust statistical measurement error models based on aggregated error statistics, where significant improvements are obtained without the need to provide detailed received signal information. The gains are seen to be even larger with up-to-date real-time information based on the estimated MPCs. Second, the association of the estimated MPCs with the signal paths predicted by the environmental model is addressed. This leads to a combinatorial problem which is approached with tools from multi-target tracking theory. A rich radio environment in terms of many MPCs gives better localization accuracy but causes the problem size to grow large—something which can be remedied by excluding less probable paths. Simulations indicate that in such environments, the single best association hypothesis may be a reasonable approximation which avoids the calculation of a vast number of possible hypotheses. Accounting for erroneous measurements is crucial but may have drawbacks if no such are occurring. Finally, theoretical localization performance bounds when utilizing all or a subset of the available MPCs are derived. A rich radio environment allows for good positioning accuracy using only a few transmitters/receivers, assuming that these are used in the localization process. In contrast, in a less rich environment where basically only the DP/LOS components are measurable, more transmitters/receivers and/or the combination of downlink and uplink measurements are required to achieve the same accuracy. The receiver’s capability of distinguishing between multiple MPCs arriving approximately at the same time also affects the localization accuracy.
Author: Andreas Bergström
Publisher: Linköping University Electronic Press
ISBN: 9179299172
Category :
Languages : en
Pages : 140
Get Book Here
Book Description
The measurements of radio signals are commonly used for localization purposes where the goal is to determine the spatial position of one or multiple objects. In realistic scenarios, any transmitted radio signal will be affected by the environment through reflections, diffraction at edges and corners etc. This causes a phenomenon known as multipath propagation, by which multiple instances of the transmitted signal having traversed different paths are heard by the receiver. These are known as Multi-Path Components (MPCs). The direct path (DP) between transmitter and receiver may also be occluded, causing what is referred to as non-Line-of-Sight (non-LOS) conditions. As a consequence of these effects, the estimated position of the object(s) may often be erroneous. This thesis focuses on how to achieve better localization accuracy by accounting for the above-mentioned multipath propagation and non-LOS effects. It is proposed how to mitigate these in the context of positioning based on estimation of the DP between transmitter and receiver. It is also proposed how to constructively utilize the additional information about the environment which they implicitly provide. This is all done in a framework wherein a given signal model and a map of the surroundings are used to build a mathematical model of the radio environment, from which the resulting MPCs are estimated. First, methods to mitigate the adverse effects of multipath propagation and non-LOS conditions for positioning based on estimation of the DP between transmitter and receiver are presented. This is initially done by using robust statistical measurement error models based on aggregated error statistics, where significant improvements are obtained without the need to provide detailed received signal information. The gains are seen to be even larger with up-to-date real-time information based on the estimated MPCs. Second, the association of the estimated MPCs with the signal paths predicted by the environmental model is addressed. This leads to a combinatorial problem which is approached with tools from multi-target tracking theory. A rich radio environment in terms of many MPCs gives better localization accuracy but causes the problem size to grow large—something which can be remedied by excluding less probable paths. Simulations indicate that in such environments, the single best association hypothesis may be a reasonable approximation which avoids the calculation of a vast number of possible hypotheses. Accounting for erroneous measurements is crucial but may have drawbacks if no such are occurring. Finally, theoretical localization performance bounds when utilizing all or a subset of the available MPCs are derived. A rich radio environment allows for good positioning accuracy using only a few transmitters/receivers, assuming that these are used in the localization process. In contrast, in a less rich environment where basically only the DP/LOS components are measurable, more transmitters/receivers and/or the combination of downlink and uplink measurements are required to achieve the same accuracy. The receiver’s capability of distinguishing between multiple MPCs arriving approximately at the same time also affects the localization accuracy.
Author: Robin Forsling
Publisher: Linköping University Electronic Press
ISBN: 9179297242
Category :
Languages : en
Pages : 110
Get Book Here
Book Description
Sensor networks consist of sensors (e.g., radar and cameras) and processing units (e.g., estimators), where in the former information extraction occurs and in the latter estimates are formed. In decentralized estimation information extracted by sensors has been pre-processed at an intermediate processing unit prior to arriving at an estimator. Pre-processing of information allows for the complexity of large systems and systems-of-systems to be significantly reduced, and also makes the sensor network robust and flexible. One of the main disadvantages of pre-processing information is that information becomes correlated. These correlations, if not handled carefully, potentially lead to underestimated uncertainties about the calculated estimates. In conservative estimation the unknown correlations are handled by ensuring that the uncertainty about an estimate is not underestimated. If this is ensured the estimate is said to be conservative. Neglecting correlations means information is double counted which in worst case implies diverging estimates with fatal consequences. While ensuring conservative estimates is the main goal, it is desirable for a conservative estimator, as for any estimator, to provide an error covariance which is as small as possible. Application areas where conservative estimation is relevant are setups where multiple agents cooperate to accomplish a common objective, e.g., target tracking, surveillance and air policing. The first part of this thesis deals with theoretical matters where the conservative linear unbiased estimation problem is formalized. This part proposes an extension of classical linear estimation theory to the conservative estimation problem. The conservative linear unbiased estimator (CLUE) is suggested as a robust and practical alternative for estimation problems where the correlations are unknown. Optimality criteria for the CLUE are provided and further investigated. It is shown that finding an optimal CLUE is more complicated than finding an optimal linear unbiased estimator in the classical version of the problem. To simplify the problem, a CLUE that is optimal under certain restrictions will also be investigated. The latter is named restricted best CLUE. An important result is a theorem that gives a closed form solution to a restricted best CLUE. Furthermore, several conservative estimation methods are described followed by an analysis of their properties. The methods are shown to be conservative and optimal under different assumptions about the underlying correlations. The second part of the thesis focuses on practical aspects of the conservative approach to decentralized estimation in configurations where the communication channel is constrained. The diagonal covariance approximation is proposed as a data reduction technique that complies with the communication constraints and if handled correctly can be shown to preserve conservative estimates. Several information selection methods are derived that can reduce the amount of data being transmitted in the communication channel. Using the information selection methods it is possible to decide what information other actors of the sensor network find useful.
Author: Marcelo Nogueira de Sousa
Publisher: BoD – Books on Demand
ISBN: 3863602447
Category : Technology & Engineering
Languages : en
Pages : 270
Get Book Here
Book Description
The precise localization of radio frequency (RF) transmitters in outdoor environments has been an important research topic in various fields for several years. Nowadays, the functionalities of many electronic devices are based on the position data of a radiofrequency transmitter using a Wireless Sensor Network (WSN). Spatially separated sensor scan measure the signal from the transmitter and estimate its location using parameters such as Time Of Arrival (ToA), Time Difference Of Arrival (TDOA), Received Signal Strength (RSS) or Direction Of Arrival (DOA). However, certain obstacles in the environment can cause reflection, diffraction, or scattering of the signal. This so called multipath effect affects the measurements for the precise location of the transmitter. Previous studies have discarded multipath information and have not considered it valuable for locating the transmitter. Some studies used ray tracing (RT) to create position fingerprints, without reference measurements, in a simulated scenario. Others tested this concept with real measurement data, but this proved to be a more cumbersome method due to practical problems in the outdoor environment. This thesis exploits the concept of Channel Impulse Response (CIR) to address the problem of precision in outdoor localization environments affected by multipath. The study aims to fill the research gap by combining multipath information from simulation with real measurements in a machine learning framework. The research question was whether the localization could be improved by combining real measurements with simulations. We propose a method that uses the multipath fingerprint information from RT simulation with reference transmitters to improve the location estimation. To validate the effectiveness of the proposed method, we implemented a TDoA location system enhanced with multipath fingerprints in an outdoor scenario. This thesis investigated suburban and rural areas using well-defined reflective components to characterize the localization multipath pattern. The results confirm the possibility of using multipath effects with real measurements to enhance the localization in outdoor situations. Instead of rejecting the multipath information, we can use them as an additional source of information.
Author: Gustav Zetterqvist
Publisher: Linköping University Electronic Press
ISBN: 9180758304
Category :
Languages : en
Pages : 93
Get Book Here
Book Description
Direction of arrival (DOA) estimation is a well-established problem in signal processing. It involves determining the direction from which a signal reaches a sensor array, and is fundamental in applications like radar, sonar, and acoustics. Traditionally, DOA estimation relies on comparing the time of arrival of the signal across different sensors in the array. However, this approach is sensitive to the time difference of arrival (TDOA) between sensors, which can be challenging to estimate accurately. Additionally, precise synchronization among the sensors is essential, but this can be difficult to achieve in certain environments or applications. In this thesis, we explore a novel approach to DOA estimation based on the received signal power at the sensors. The method exploits the directional sensitivity of the microphones in the array, which defines how effectively each microphone captures sound from different directions. To model the directional sensitivity, we use a Fourier series (FS) model. The model is then used to estimate the DOA of a sound source across various environments, and for different types of signals. The parametric model enables Cramér-Rao lower bound (CRLB) analysis of the DOA estimation problem. Our findings demonstrate that the directional sensitivity exhibits a significant variation in accordance with the frequency content of the signal, and we exploit this to estimate the DOA for different types of sounds. The proposed method has been validated with a range of signals, including gunshots, elephant trumpets, sirens, and female screams. The results show that the developed method achieves high accuracy in estimating the DOA for the above-mentioned signals. Furthermore, the method performs similarly well in outdoor scenarios with realistic background noise levels. When compared to state-of-the-art DOA estimation techniques, our approach performs better or equally well for the investigated sounds. A key advantage of this method is that it does not require any TDOA measurement between the microphones, enabling the design of smaller, more compact devices. This opens up new possibilities for estimating DOA in environments where traditional methods are impractical. A limitation, however, is that the method requires knowledge of the microphone’s directional sensitivity, which necessitates calibration in an anechoic chamber. Nevertheless, this calibration has proven to be robust, and only needs to be performed once to create a model applicable across different environments. Additionally, this thesis explores a different application of DOA estimation, where geophones are used to estimate the DOA to elephants. As elephants move, they generate ground vibrations, and these signals can be captured by geophones. We show that a traditional delay-and-sum beamformer can accurately estimate the DOA of elephants at distances up to 40 meters. By determining when elephants are approaching and from which direction, park rangers can take early measures to avoid conflicts between humans and elephants, which is a major problem in some parts of the world. Förmågan att höra var ett ljud kommer ifrån, något vi ofta tar för givet, kallas för riktningsuppfattning. Den gör det möjligt för oss att snabbt avgöra om någon ropar på oss och från vilket håll ljudet kommer. Denna förmåga är viktig för att kunna orientera sig i omgivningen och uppfatta hot eller andra viktiga ljud. Våra öron samarbetar genom att jämföra hur ljud når varje öra, både när det gäller ljudets intensitet och hur lång tid det tar för ljudet att nå dem. Det här kallas för interaural tids- och nivåskillnad. Vissa ljud kan dock vara svåra att uppfatta, till exempel om ljudet är kort och impulsivt, eller om det är i en stadsmiljö med mycket bakgrundsljud och reflektioner. I den här avhandlingen undersöker vi nya metoder för att uppskatta ljudets riktning. Vi använder mikrofoner för att mäta ljudet och beräknar därefter riktningen som ljudet kommer ifrån. Traditionella metoder fokuserar på tidsskillnaden mellan ljud som registreras i olika mikrofoner. Vi tar istället en annan väg och undersöker hur ljudets styrka kan användas för att avgöra riktningen, oavsett tidsskillnader mellan mikrofonerna. Vår metod bygger på att vi skapar en modell av mikrofonernas riktningskänslighet, det vill säga hur väl de uppfattar ljud från olika håll. Modellen skapas genom att mäta mikrofonens riktningskänslighet i ett ekofritt rum. Genom att först mäta detta i en kontrollerad miljö, utan ekon, kan vi sedan använda modellen för att beräkna ljudriktningen i mer varierande miljöer och för olika typer av ljud. Till exempel har vi använt ljud såsom pistolskott, elefanttrumpeter, sirener och skrik för att testa vår metod. Resultaten visar att vår metod kan beräkna riktningar med hög noggrannhet för de ovan nämnda ljuden, även i en utomhusmiljö med mer realistiska nivåer av bakgrundsljud. När vi jämfört vår metod med traditionella metoder, presterar vår lösning lika bra eller bättre för de testade ljuden. En stor fördel med vår metod är att den inte kräver att mikrofonerna är placerade på ett visst avstånd från varandra, vilket innebär att vi kan bygga mindre och mer kompakta enheter. Detta kan leda till nya typer av produkter för att identifiera ljudriktningar i olika situationer. En nackdel är dock att mikrofonernas riktningskänslighet måste kalibreras i ett ljudlabb, men denna kalibrering har visat sig vara robust och det räcker att utföra en kalibrering som kan användas i flera olika miljöer. I avhandlingen inkluderas även en annan tillämpning av riktningsskattning, nämligen att uppskatta riktningen till elefanter med hjälp av geofoner som mäter vibrationer i marken. Elefanter är stora djur som skapar tydliga vibrationer i marken när de går. Genom att mäta dessa vibrationer med geofoner kan vi uppskatta riktningen till elefanten. Vi visar att traditionella metoder kan uppskatta riktningen med hög noggrannhet på ett avstånd upp till 40 meter. Genom att avgöra när elefanter närmar sig människor och varifrån de kommer kan parkvakter vidta åtgärder för att undvika konflikter mellan människor och elefanter, vilket är ett stort problem i vissa delar av världen.
Author: Daniel Arnström
Publisher: Linköping University Electronic Press
ISBN: 9179296920
Category :
Languages : en
Pages : 45
Get Book Here
Book Description
In model predictive control (MPC) an optimization problem has to be solved at each time step, which in real-time applications makes it important to solve these efficiently and to have good upper bounds on worst-case solution time. Often for linear MPC problems, the optimization problem in question is a quadratic program (QP) that depends on parameters such as system states and reference signals. A popular class of methods for solving such QPs is active-set methods, where a sequence of linear systems of equations is solved. The primary contribution of this thesis is a method which determines which sequence of subproblems a popular class of such active-set algorithms need to solve, for every possible QP instance that might arise from a given linear MPC problem (i.e, for every possible state and reference signal). By knowing these sequences, worst-case bounds on how many iterations, floating-point operations and, ultimately, the maximum solution time, these active-set algorithms require to compute a solution can be determined, which is of importance when, e.g, linear MPC is used in safety-critical applications. After establishing this complexity certification method, its applicability is extended by showing how it can be used indirectly to certify the complexity of another, efficient, type of active-set QP algorithm which reformulates the QP as a nonnegative least-squares method. Finally, the proposed complexity certification method is extended further to situations when enhancements to the active-set algorithms are used, namely, when they are terminated early (to save computations) and when outer proximal-point iterations are performed (to improve numerical stability).
Author:
Publisher:
ISBN: 9781728195070
Category :
Languages : en
Pages :
Get Book Here
Book Description
Author: Y. Jade Morton
Publisher: John Wiley & Sons
ISBN: 1119458412
Category : Science
Languages : en
Pages : 1170
Get Book Here
Book Description
Covers the latest developments in PNT technologies, including integrated satellite navigation, sensor systems, and civil applications Featuring sixty-four chapters that are divided into six parts, this two-volume work provides comprehensive coverage of the state-of-the-art in satellite-based position, navigation, and timing (PNT) technologies and civilian applications. It also examines alternative navigation technologies based on other signals-of-opportunity and sensors and offers a comprehensive treatment on integrated PNT systems for consumer and commercial applications. Volume 1 of Position, Navigation, and Timing Technologies in the 21st Century: Integrated Satellite Navigation, Sensor Systems, and Civil Applications contains three parts and focuses on the satellite navigation systems, technologies, and engineering and scientific applications. It starts with a historical perspective of GPS development and other related PNT development. Current global and regional navigation satellite systems (GNSS and RNSS), their inter-operability, signal quality monitoring, satellite orbit and time synchronization, and ground- and satellite-based augmentation systems are examined. Recent progresses in satellite navigation receiver technologies and challenges for operations in multipath-rich urban environment, in handling spoofing and interference, and in ensuring PNT integrity are addressed. A section on satellite navigation for engineering and scientific applications finishes off the volume. Volume 2 of Position, Navigation, and Timing Technologies in the 21st Century: Integrated Satellite Navigation, Sensor Systems, and Civil Applications consists of three parts and addresses PNT using alternative signals and sensors and integrated PNT technologies for consumer and commercial applications. It looks at PNT using various radio signals-of-opportunity, atomic clock, optical, laser, magnetic field, celestial, MEMS and inertial sensors, as well as the concept of navigation from Low-Earth Orbiting (LEO) satellites. GNSS-INS integration, neuroscience of navigation, and animal navigation are also covered. The volume finishes off with a collection of work on contemporary PNT applications such as survey and mobile mapping, precision agriculture, wearable systems, automated driving, train control, commercial unmanned aircraft systems, aviation, and navigation in the unique Arctic environment. In addition, this text: Serves as a complete reference and handbook for professionals and students interested in the broad range of PNT subjects Includes chapters that focus on the latest developments in GNSS and other navigation sensors, techniques, and applications Illustrates interconnecting relationships between various types of technologies in order to assure more protected, tough, and accurate PNT Position, Navigation, and Timing Technologies in the 21st Century: Integrated Satellite Navigation, Sensor Systems, and Civil Applications will appeal to all industry professionals, researchers, and academics involved with the science, engineering, and applications of position, navigation, and timing technologies. pnt21book.com
Author: Fredrik Ljungberg
Publisher: Linköping University Electronic Press
ISBN: 9179298400
Category :
Languages : en
Pages : 144
Get Book Here
Book Description
As marine vessels are becoming increasingly autonomous, having accurate simulation models available is turning into an absolute necessity. This holds both for facilitation of development and for achieving satisfactory model-based control. When accurate ship models are sought, it is necessary to account for nonlinear hydrodynamic effects and to deal with environmental disturbances in a correct way. In this thesis, parameter estimators for nonlinear regression models where the regressors are second-order modulus functions are analyzed. This model class is referred to as second-order modulus models and is often used for greybox identification of marine vessels. The primary focus in the thesis is to find consistent estimators and for this an instrumental variable (IV) method is used. First, it is demonstrated that the accuracy of an IV estimator can be improved by conducting experiments where the input signal has a static offset of sufficient amplitude and the instruments are forced to have zero mean. This two-step procedure is shown to give consistent estimators for second-order modulus models in cases where an off-the-shelf applied IV method does not, in particular when measurement uncertainty is taken into account. Moreover, it is shown that the possibility of obtaining consistent parameter estimators for models of this type depends on how process disturbances enter the system and on the amount of prior knowledge about the disturbances’ probability distributions that is available. In cases where the first-order moments are known, the aforementioned approach gives consistent estimators even when disturbances enter the system before the nonlinearity. In order to obtain consistent estimators in cases where the first-order moments are unknown, a framework for estimating the first and second-order moments alongside the model parameters is suggested. The idea is to describe the environmental disturbances as stationary stochastic processes in an inertial frame and to utilize the fact that their effect on a vessel depends on the vessel’s attitude. It is consequently possible to infer information about the environmental disturbances by over time measuring the orientation of a vessel they are affecting. Furthermore, in cases where the process disturbances are of more general character it is shown that supplementary disturbance measurements can be used for achieving consistency. Different scenarios where consistency can be achieved for instrumental variable estimators of second-order modulus models are demonstrated, both in theory and by simulation examples. Finally, estimation results obtained using data from a full-scale marine vessel are presented. I takt med att marina farkoster blir mer autonoma ökar behovet av noggranna matematiska farkostmodeller. Modellerna behövs både för att förenkla utvecklingen av nya farkoster och för att kunna styra farkosterna autonomt med önskad precision. För att erhålla allmängiltiga modeller behöver olinjära hydrodynamiska effekter samt systemstörningar, främst orsakade av vind- och vattenströmmar, tas i beaktning. I det här arbetet undersöks metoder för att skatta okända storheter i modeller för marina farkoster givet observerad data. Undersökningen gäller en speciell typ av olinjära modeller som ofta används för att beskriva marina farkoster. Huvudfokus i arbetet är att erhålla konsistens, vilket betyder att de skattade storheterna ska anta rätt värden när mängden observerad data ökar. För det används en redan etablerad statistisk metod som baseras på instrumentvariabler. Det visas först att noggrannheten i modellskattningsmetoden kan förbättras om datainsamlingsexperimenten utförs på ett sätt så att farkosten har signifikant nollskild hastighet och instrumentvariablernas medelvärde dras bort. Den här tvåstegslösningen påvisas vara fördelaktig vid skattning av parametrar i den ovan nämnda modelltypen, framför allt då mätosäkerhet tas i beaktning. Vidare så visas det att möjligheten att erhålla konsistenta skattningsmetoder beror på hur mycket kännedom om systemstörningarna som finns tillgänglig på förhand. I fallet då de huvudsakliga hastigheterna på vind- och vattenströmmar är kända, räcker den tidigare nämnda tvåstegsmetoden bra. För att även kunna hantera det mer generella fallet föreslås en metod för att skatta de huvudsakliga hastigheterna och de okända modellparametrarna parallellt. Denna idé baserar sig på att beskriva störningarna som stationära i ett globalt koordinatsystem och att anta att deras effekt på en farkost beror på hur farkosten är orienterad. Genom att över tid mäta och samla in data som beskriver en farkosts kurs, kan man således dra slutsatser om de störningar som farkosten påverkas av. Utöver detta visas det att utnyttjande av vindmätningar kan ge konsistens i fallet med störningar av mer generell karaktär. Olika scenarion där konsistens kan uppnås visas både i teori och med simuleringsexempel. Slutligen visas också modellskattningsresultat som erhållits med data insamlad från ett fullskaligt fartyg.
Author: Erik Hedberg
Publisher: Linköping University Electronic Press
ISBN: 9179297404
Category : Electronic books
Languages : en
Pages : 83
Get Book Here
Book Description
The two topics at the heart of this thesis are how to improve control of industrial manipulators and how to reason about the role of models in automatic control. On industrial manipulators, two case studies are presented. The first investigates estimation with inertial sensors, and the second compares control by feedback linearization to control based on gain-scheduling. The contributions on the second topic illustrate the close connection between control and estimation in different ways. A conceptual model of control is introduced, which can be used to emphasize the role of models as well as the human aspect of control engineering. Some observations are made regarding block-diagram reformulations that illustrate the relation between models, control and inversion. Finally, a suggestion for how the internal model principle, internal model control, disturbance observers and Youla-Kucera parametrization can be introduced in a unified way is presented.
Author: Magnus Malmström
Publisher: Linköping University Electronic Press
ISBN: 9179296807
Category :
Languages : en
Pages : 103
Get Book Here
Book Description
In science, technology, and engineering, creating models of the environment to predict future events has always been a key component. The models could be everything from how the friction of a tire depends on the wheels slip to how a pathogen is spread throughout society. As more data becomes available, the use of data-driven black-box models becomes more attractive. In many areas they have shown promising results, but for them to be used widespread in safety-critical applications such as autonomous driving some notion of uncertainty in the prediction is required. An example of such a black-box model is neural networks (NNs). This thesis aims to increase the usefulness of NNs by presenting an method where uncertainty in the prediction is obtained by linearization of the model. In system identification and sensor fusion, under the condition that the model structure is identifiable, this is a commonly used approach to get uncertainty in the prediction from a nonlinear model. If the model structure is not identifiable, such as for NNs, the ambiguities that cause this have to be taken care of in order to make the approach applicable. This is handled in the first part of the thesis where NNs are analyzed from a system identification perspective, and sources of uncertainty are discussed. Another problem with data-driven black-box models is that it is difficult to know how flexible the model needs to be in order to correctly model the true system. One solution to this problem is to use a model that is more flexible than necessary to make sure that the model is flexible enough. But how would that extra flexibility affect the uncertainty in the prediction? This is handled in the later part of the thesis where it is shown that the uncertainty in the prediction is bounded from below by the uncertainty in the prediction of the model with lowest flexibility required for representing true system accurately. In the literature, many other approaches to handle the uncertainty in predictions by NNs have been suggested, of which some are summarized in this work. Furthermore, a simulation and an experimental studies inspired by autonomous driving are conducted. In the simulation study, different sources of uncertainty are investigated, as well as how large the uncertainty in the predictions by NNs are in areas without training data. In the experimental study, the uncertainty in predictions done by different models are investigated. The results show that, compared to existing methods, the linearization method produces similar results for the uncertainty in predictions by NNs. An introduction video is available at https://youtu.be/O4ZcUTGXFN0 Inom forskning och utveckling har det har alltid varit centralt att skapa modeller av verkligheten. Dessa modeller har bland annat använts till att förutspå framtida händelser eller för att styra ett system till att bete sig som man önskar. Modellerna kan beskriva allt från hur friktionen hos ett bildäck påverkas av hur mycket hjulen glider till hur ett virus kan sprida sig i ett samhälle. I takt med att mer och mer data blir tillgänglig ökar potentialen för datadrivna black-box modeller. Dessa modeller är universella approximationer vilka ska kunna representera vilken godtycklig funktion som helst. Användningen av dessa modeller har haft stor framgång inom många områden men för att verkligen kunna etablera sig inom säkerhetskritiska områden såsom självkörande farkoster behövs en förståelse för osäkerhet i prediktionen från modellen. Neuronnät är ett exempel på en sådan black-box modell. I denna avhandling kommer olika sätt att tillförskaffa sig kunskap om osäkerhet i prediktionen av neuronnät undersökas. En metod som bygger på linjärisering av modellen för att tillförskaffa sig osäkerhet i prediktionen av neuronnätet kommer att presenteras. Denna metod är välbeprövad inom systemidentifiering och sensorfusion under antagandet att modellen är identifierbar. För modeller såsom neuronnät, vilka inte är identifierbara behövs det att det tas hänsyn till tvetydigheterna i modellen. En annan utmaning med datadrivna black-box modeller, är att veta om den valda modellmängden är tillräckligt generell för att kunna modellera det sanna systemet. En lösning på detta problem är att använda modeller som har mer flexibilitet än vad som behövs, det vill säga en överparameteriserad modell. Men hur påverkas osäkerheten i prediktionen av detta? Detta är något som undersöks i denna avhandling, vilken visar att osäkerheten i den överparameteriserad modellen kommer att vara begränsad underifrån av modellen med minst flexibilitet som ändå är tillräckligt generell för att modellera det sanna systemet. Som avslutning kommer dessa resultat att demonstreras i både en simuleringsstudie och en experimentstudie inspirerad av självkörande farkoster. Fokuset i simuleringsstudien är hur osäkerheten hos modellen är i områden med och utan tillgång till träningsdata medan experimentstudien fokuserar på jämförelsen mellan osäkerheten i olika typer av modeller.Resultaten från dessa studier visar att metoden som bygger på linjärisering ger liknande resultat för skattningen av osäkerheten i prediktionen av neuronnät, jämfört med existerande metoder.
