As computing power becomes more and more available, robotic systems are moving away from active sensors for environmental awareness and transitioning into passive vision sensors.  With the advent of teleoperation and real-time video tracking of dynamic environments, the need to stabilize video onboard mobile robots has become more prevalent. This thesis presents a digital stabilization method that incorporates inertial fusion with a Kalman filter.  The camera motion is derived visually by tracking SIFT features in the video feed and fitting them to an affine model.  The digital motion is fused with a 3 axis rotational motion measured by an inertial measurement unit (IMU) rigidly attached to the camera. The video is stabilized by digitally manipulating the image plane opposite of the unwanted motion. The result is the foundation of a robust video stabilizer comprised of both visual and inertial measurements.  The stabilizer is immune to dynamic scenes and requires less computation than current digital video stabilization methods. / Master of Science

Indiana University-Purdue University Indianapolis (IUPUI) / With passage of time and repeated usage of a system, component values that make up the system parameters change, causing errors in its functional output. In order to ensure the fidelity of the results derived from these systems it is thus very important to keep track of the system parameters while being used. This thesis introduces a method for tracking the existing system parameters while the system was being used using the inherent noise of its signal source. Kalman filter algorithm is used to track the inherent noise response to the system and use that response to estimate the system parameters. In this thesis this continuous characterization scheme has been used on a Universal Invertible Amplifier (UIA). Current biomedical research as well as diagnostic medicine depend a lot on shape profile of bio-electric signals of different sources, for example heart, muscle, nerve, brain etc. making it very important to capture the different event of these signals without the distortion usually introduced by the filtering of the amplifier system. The Universal Invertible Amplifier extracts the original signal in electrodes by inverting the filtered and compressed signal while its gain bandwidth profile allows it to capture from the entire bandwidth of bioelectric signals. For this inversion to be successful the captured compressed and filtered signals needs to be inverted with the actual system parameters that the system had during capturing the signals, not its original parameters. The continuous characterization scheme introduced in this thesis is aimed at knowing the system parameters of the UIA by tracking the response of its source noise and estimating its transfer function from that. Two types of source noises have been tried out in this method, an externally added noise that was digitally generated and a noise that inherently contaminates the signals the system is trying to capture. In our cases, the UIA was used to capture nerve activity from vagus nerve where the signal was contaminated with electrocardiogram signals providing us with a well-defined inherent noise whose response could be tracked with Kalman Filter and used to estimate the transfer function of UIA. The transfer function estimation using the externally added noise did not produce good results but could be improved by means that can be explored as future direction of this project. However continuous characterization using the inherent noise, a bioelectric signal, was successful producing transfer function estimates with minimal error. Thus this thesis was successful to introduce a novel approach for system characterization using bio-signal contamination.

<p> Bayesian filtering refers to the process of sequentially estimating the current state of a complex dynamic system from noisy partial measurements using Bayes' rule. This thesis considers Bayesian filtering as applied to an important class of state estimation problems, which is describable by a discrete-time nonlinear state-space model with additive Gaussian noise. It is known that the conditional probability density of the state given the measurement history or simply the posterior density contains all information about the state. For nonlinear systems, the posterior density cannot be described by a finite number of sufficient statistics, and an approximation must be made instead.</p> <p> The approximation of the posterior density is a challenging problem that has engaged many researchers for over four decades. Their work has resulted in a variety of approximate Bayesian filters. Unfortunately, the existing filters suffer from possible divergence, or the curse of dimensionality, or both, and it is doubtful that a single filter exists that would be considered effective for applications ranging from low to high dimensions. The challenge ahead of us therefore is to derive an approximate nonlinear Bayesian filter, which is theoretically motivated, reasonably accurate, and easily extendable to a wide range of applications at a minimal computational cost.</p> <p> In this thesis, a new approximate Bayesian filter is derived for discrete-time nonlinear filtering problems, which is named the cubature Kalman filter. To develop this filter, it is assumed that the predictive density of the joint state-measurement random variable is Gaussian. In this way, the optimal Bayesian filter reduces to the problem of how to compute various multi-dimensional Gaussian-weighted moment integrals. To numerically compute these integrals, a third-degree spherical-radial cubature rule is proposed. This cubature rule entails a set of cubature points scaling linearly with the state-vector dimension. The cubature Kalman filter therefore provides an efficient solution even for high-dimensional nonlinear filtering problems. More remarkably, the cubature Kalman filter is the closest known approximate filter in the sense of completely preserving second-order information due to the maximum entropy principle. For the purpose of mitigating divergence, and improving numerical accuracy in systems where there are apparent computer roundoff difficulties, the cubature Kalman filter is reformulated to propagate the square roots of the error-covariance matrices. The formulation of the (square-root) cubature Kalman filter is validated through three different numerical experiments, namely, tracking a maneuvering ship, supervised training of recurrent neural networks, and model-based signal detection and enhancement. All three experiments clearly indicate that this powerful new filter is superior to other existing nonlinear filters. </p> / Thesis / Doctor of Philosophy (PhD)

Position in the environment is essential in any autonomous system. As increased accuracy is required, the costs escalate accordingly. This paper presents a simple way to systematically integrate sensory data to provide a drivable and accurate position solution at a low cost. The data fusion is handled by a Kalman filter tracking five states and an undetermined number of asynchronous measurements. This implementation allows the user to define additional adjustments to improve the overall behavior of the filter. The filter is tested using a suite of inexpensive sensors and then compared to a differential GPS position. The output of the filter is indeed a drivable solution that tracks the reference position remarkably well. This approach takes advantage of the short-term accuracy of odometry measurements and the long-term fix of a GPS unit. A maximum error of two meters of deviation from the reference is shown for a complex path over two minutes and 100 meters long. / Master of Science

A lane position system and enhancement techniques, for increasing the robustnessand availability of such a system, are investigated. The enhancements areperformed by using additional sensor sources like map data and GPS. The thesiscontains a description of the system, two models of the system and two implementedfilters for the system. The thesis also contains conclusions and results oftheoretical and experimental tests of the increased robustness and availability ofthe system. The system can be integrated with an existing system that investigatesdriver behavior, developed for fatigue. That system was developed in aproject named Drowsi, where among others Volvo Technology participated. / Ett filpositioneringssystem undersöks och förbättringstekniker för ökandet av robusthetoch tillgängligheten av ett sådant system genom att använda ytterligaresensorkällor som kartdata och GPS. Detta examensarbete presenterar beskrivningenav ett system, två modeller och två implementerade filter. Examensarbetetinnehåller också slutsatser och resultat av teoretiska och experimentella testersom plottar och grafer av ökad robusthet och tillgängligheten av systemet. Dettasystem kan bli integrerat med ett framtaget system som tittar på körrelaterat beteendevid trötthet. Systemet är utvecklat i ett projekt kallat Drowsi, där blandandra Volvo Technology deltog.

<p>A lane departure warning system relying exclusively on a camera has several shortcomings and tends to be sensitive to, e.g., bad weather and abrupt manoeuvres. To handle these situations, the system proposed in this thesis uses a dynamic model of the vehicle and integration of relative motion sensors to estimate the vehicle’s position on the road. The relative motion is measured using vision, inertial, and vehicle sensors. All these sensors types are affected by errors such as offset, drift and quantization. However the different sensors are sensitive to different types of errors, e.g., the camera system is rather poor at detecting rapid lateral movements, a type of situation which an inertial sensor practically never fails to detect. These kinds of complementary properties make sensor fusion interesting. The approach of this Master’s thesis is to use an already existing lane departure warning system as vision sensor in combination with an inertial measurement unit to produce a system that is robust and can achieve good warnings if an unintentional lane departure is about to occur. For the combination of sensor data, different sensor fusion models have been proposed and evaluated on experimental data. The models are based on a nonlinear model that is linearized so that a Kalman filter can be applied. Experiments show that the proposed solutions succeed at handling situations where a system relying solely on a camera would have problems. The results from the testing show that the original lane departure warning system, which is a single camera system, is outperformed by the suggested system.</p>

No description available.

A lane departure warning system relying exclusively on a camera has several shortcomings and tends to be sensitive to, e.g., bad weather and abrupt manoeuvres. To handle these situations, the system proposed in this thesis uses a dynamic model of the vehicle and integration of relative motion sensors to estimate the vehicle’s position on the road. The relative motion is measured using vision, inertial, and vehicle sensors. All these sensors types are affected by errors such as offset, drift and quantization. However the different sensors are sensitive to different types of errors, e.g., the camera system is rather poor at detecting rapid lateral movements, a type of situation which an inertial sensor practically never fails to detect. These kinds of complementary properties make sensor fusion interesting. The approach of this Master’s thesis is to use an already existing lane departure warning system as vision sensor in combination with an inertial measurement unit to produce a system that is robust and can achieve good warnings if an unintentional lane departure is about to occur. For the combination of sensor data, different sensor fusion models have been proposed and evaluated on experimental data. The models are based on a nonlinear model that is linearized so that a Kalman filter can be applied. Experiments show that the proposed solutions succeed at handling situations where a system relying solely on a camera would have problems. The results from the testing show that the original lane departure warning system, which is a single camera system, is outperformed by the suggested system.

In this thesis, we propose two novel approaches for hybrid Ensemble Kalman Filter (EnKF) to overcome limitations of the traditional EnKF. The first approach is to swap the ensemble mean for the ensemble mode estimation to improve the covariance calculation in EnKF. The second approach is a coarse scale permeability constraint while updating in EnKF. Both hybrid EnKF approaches are coupled with the streamline based Generalized Travel Time Inversion (GTTI) algorithm for periodic updating of the mean of the ensemble and to sequentially update the ensemble in a hybrid fashion. Through the development of the hybrid EnKF algorithm, the characteristics of the EnKF are also investigated. We found that the limits of the updated values constrain the assimilation results significantly and it is important to assess the measurement error variance to have a proper balance between preserving the prior information and the observation data misfit. Overshooting problems can be mitigated with the streamline based covariance localizations and normal score transformation of the parameters to support the Gaussian error statistics. The swapping mean and mode estimation approach can give us a better matching of the data as long as the mode solution of the inversion process is satisfactory in terms of matching the observation trajectory. The coarse scale permeability constrained hybrid approach gives us better parameter estimation in terms of capturing the main trend of the permeability field and each ensemble member is driven to the posterior mode solution from the inversion process. However the WWCT responses and pressure responses need to be captured through the inversion process to generate physically plausible coarse scale permeability data to constrain hybrid EnKF updating. Uncertainty quantification methods for EnKF were developed to verify the performance of the proposed hybrid EnKF compared to the traditional EnKF. The results show better assimilation quality through a sequence of updating and a stable solution is demonstrated. The potential of the proposed hybrid approaches are promising through the synthetic examples and a field scale application.

<p>In today’s heavy industry there are applications demanding high power supply in certain periods of a working cycle. A typical case might be startup of heavy machinery or just keeping a certain point in a distribution network at a certain energy level. To deal with this different techniques might be used, one way is to introduce a battery as an energy reserve in the system. One battery studied at ABB for this purpose is the so called High Temperature Sodium Nickel Chloride battery and a model of this battery has been developed at ABB. When operating a battery of the mentioned type in an application it is important to keep track of the energy stored in the battery. Earlier tests has shown that this is difficult in a noisy environment.</p><p>This master thesis investigates if a Kalman filter may be used to estimate the energy stored in the battery. The investigation is performed in steps, starting with a simplified model of the battery and then expanding to a more complete model. Evaluation of the methods and algorithms used is made by simulations and based on the assumption that there is a good model available. The model is special in such a way that it has a varying number of states despite that the number of outputs remains the same.</p><p>Some comparisons with actual measurements are also made and an analysis of the parameters in the model along with an introduction to the system identification problem is discussed, assuming that the structure of the model is correct.</p> / <p>I dagens tunga industri finns applikationer som kräver höga effektuttag under vissa perioder av en arbetscykel. Ett typiskt fall kan vara uppstart av tunga maskiner eller att hålla en given spänningsnivå i en belastningspunkt i ett distributionsnät. För att hantera detta finns olika metoder, en möjlighet är att använda ett batteri som en energireserv. Ett högtemperaturbatteri har studerats på ABB för detta ändamål och en model av detta batteri har tagits fram. När ett sådant batteri används är det viktigt att kontinuerligt veta hur mycket energi som finns till förfogande i batteriet. Tidigare tester har visat att detta är svårt i en brusig miljö.</p><p>I detta examensarbete kommer det undersökas om ett Kalman filter kan användas för att skatta energin i detta batteri. Undersökningen sker i steg och startar med en förenklad modell som sedan utvecklas till en mer komplett modell. Utvärdering av de metoder och algoritmer som används sker via simuleringar och baseras på antagandet att modellen är komplett och riktig. Denna modell är speciell på det sätt att den har ett variabelt antal tillstånd trots att antalet utsignaler är konstant.</p><p>Viss jämförelse med de mätningar som finns tillgängliga görs och en inledande analys av de ingående modellparametrarna presenteras. Även en introduktion till det omfattande systemidentifieringsproblemet diskuteras, med antagandet att modellens struktur är korrekt.</p>