It was demonstrated in recent studies that Complementary Limb Motion Estimation (CLME) is robust approach for controlling active knee prostheses. A wearable sensor system is then needed to provide inputs to the controller in a real-time platform. In the present work, a wearable sensor system based on magnetic and inertial measurement units (MIMU) together with a simple calibration procedure were proposed. This sensor system was intended to substitute and extend the capabilities of a previous device based on potentiometers and gyroscopes. The proposed sensor system and calibration were validated with an Optical Tracking System (OTS) in a standard gait lab and first results showed that the proposed solution had a performance comparable to similar studies in the literature.

This thesis develops an algorithm for tracking the boundary of an airbag throughout an image sequence. The algorithm is designed to work even if various problematic features, e.g. objects in the background, are present in the image. The work is built on an existing commercially available image processing and analysis suite targeted at the automotive industry. The software suite runs on standard PC hardware. Firstly, improvements to the airbag tracking algorithm already available in the suite are considered. Testing reveals that these measures are not sufficient to overcome the problems posed by the problematic image sequences. A new tracking algorithmis then proposed. It consists of a Canny edge detector, optional steps to enhance feature detection by removing edges in the background and edges interior to the boundary of the airbag and finally an active contour. The role of the active contour is to produce a closed curve while imposing smoothness constraints on the detected boundary. The active contour is in each frame initialized by linearly extrapolating the contour from previous frames. The algorithm works very well and it is fast enough to run on slower machines than was initially targeted.

Image segmentation is an important step in image processing, with many applications such as pattern recognition, object detection, and medical image analysis. It is a technique that separates objects of interests from the background in an image. Geometric active contour is a recent image segmentation method that overcomes previous problems with snakes. It is an attractive method for medical image segmentation as it is able to capture the object of interest in one continuous curve. The theory and implementation details of geometric active contours are discussed in this work. The robustness of the algorithm is tested through a series of tests, involving both synthetic images and medical images. Curve leaking past boundaries is a common problem in cases of non-ideal edges. Noise is also problematic for the advancement of the curve. Smoothing and parameters selection are discussed as ways to help solve these problems. This work also explores the incorporation of the multi-resolution method of Gaussian pyramids into the algorithm. Multi-resolution methods, used extensively in the areas of denoising and edge-selection, can help capture the spatial structure of an image. Results show that similar to the multi-resolution methods applied to parametric active contours, the multi-resolution can greatly increase the computation without sacrificing performance. In fact, results show that with successive smoothing and sub-sampling, performance often improves. Although smoothing and parameter adjustment help improve the performance of geometric active contours, the edge-based approach is still localized and the improvement is limited. Region-based approaches are recommended for further work on active contours.

A significant problem in many machine learning tasks is that it is time consuming and costly to gather the necessary labeled data for training the learning algorithm to a reasonable level of performance. In reality, it is often the case that a small amount of labeled data is available and that more unlabeled data could be labeled on demand at a cost. If the labeled data is obtained by a process outside of the control of the learner, then the learner is passive. If the learner picks the data to be labeled, then this becomes active learning. This has the advantage that the learner can pick data to gain specific information that will speed up the learning process. Support Vector Machines (SVMs) have many properties that make them attractive to use as a learning algorithm for many real world applications including classification tasks. Some researchers have proposed algorithms for active learning with SVMs, i.e. algorithms for choosing the next unlabeled instance to get label for. Their approach is supervised in nature since they do not consider all unlabeled instances while looking for the next instance. In this thesis, we propose three new algorithms for applying active learning for SVMs in a semi-supervised setting which takes advantage of the presence of all unlabeled points. The suggested approaches might, by reducing the number of experiments needed, yield considerable savings in costly classification problems in the cases when finding the training data for a classifier is expensive.

Vehicle suspension systems have been extensively explored in the past decades, contributing to ride comfort, handling and safety improvements. The new generation of powertrain and propulsion systems, as a new trend in modern vehicles, poses significant challenges to suspension system design. Consequently, novel suspension concepts are required, not only to improve the vehicle’s dynamic performance, but also to enhance the fuel economy by utilizing regeneration functions. However, the development of new-generation suspension systems necessitates advanced suspension components, such as springs and dampers. This Ph.D. thesis, on the development of hybrid electromagnetic dampers is an Ontario Centres of Excellence (OCE) collaborative project sponsored by Mechworks Systems Inc. The ultimate goal of this project is to conduct feasibility study of the development of electromagnetic dampers for automotive suspension system applications. With new improvements in power electronics and magnetic materials, electromagnetic dampers are forging the way as a new technology in vibration isolation systems such as vehicle suspension systems. The use of electromagnetic dampers in active vehicle suspension systems has drawn considerable attention in the recent years, attributed to the fact that active suspension systems have superior performance in terms of ride comfort and road-handling performances compared to their passive and semi-active counterparts in automotive applications. As a response to the expanding demand for superior vehicle suspension systems, this thesis describes the design and development of a new electromagnetic damper as a customized linear permanent magnet actuator to be used in active suspension systems. The proposed electromagnetic damper has energy harvesting capability. Unlike commercial passive/semi-active dampers that convert the vibration kinetic energy into heat, the dissipated energy in electromagnetic dampers can be regenerated as useful electrical energy. Electromagnetic dampers are used in active suspension systems, where the damping coefficient is controlled rapidly and reliably through electrical manipulations. Although demonstrating superb performance, active suspensions still have some issues that must be overcome. They have high energy consumption, weight, and cost, and are not fail-safe in case of a power break-down. Since the introduction of the electromagnetic dampers, the challenge was to address these drawbacks. Hybrid electromagnetic dampers, which are proposed in this Ph.D. thesis, are potential solutions to high weight, high cost, and fail-safety issues of an active suspension system. The hybrid electromagnetic damper utilizes the high performance of an active electromagnetic damper with the reliability of passive dampers in a single package, offering a fail-safe damper while decreasing weight and cost. Two hybrid damper designs are proposed in this thesis. The first one operates based on hydraulic damping as a source of passive damping, while the second design employs the eddy current damping effect to provide the passive damping part of the system. It is demonstrated that the introduction of the passive damping can reduce power consumption and weight in an active automotive suspension system. The ultimate objective of this thesis is to employ existing suspension system and damper design knowledge together with new ideas from electromagnetic theories to develop new electromagnetic dampers. At the same time, the development of eddy current dampers, as a potential source for passive damping element in the final hybrid design, is considered and thoroughly studied. For the very first time, the eddy current damping effect is introduced for the automotive suspension applications. The eddy current passive damper, as a stand-alone unit, is designed, modeled, fabricated and successfully tested. The feasibility of using passive eddy current dampers for automotive suspension applications is also studied. The structure of new passive eddy current dampers is straightforward, requiring no external power supply or any other electronic devices. Proposed novel eddy current dampers are oil-free and non-contact, offering high reliability and durability with their simplified design. To achieve the defined goals, analytical modeling, numerical simulations, and lab-based experiments are conducted. A number of experimental test-beds are prepared for various experimental analyses on the fabricated prototypes as well as off-the-shelf dampers. Various prototypes, such as eddy current and electromagnetic dampers, are manufactured, and tested in frequency/time domains for verification of the derived analytical and numerical models, and for proof of concept. In addition, fluid and heat transfer analyses are done during the process of the feasibility study to ensure the durability and practical viability of the proposed hybrid electromagnetic dampers. The presented study is only a small portion of the growing research in this area, and it is hoped that the results obtained here will lead to the realization of a safer and more superior automotive suspension system.

Rapid progress over the last decade on large area thin film transistor (TFT) arrays led to the emergence of high-performance, low-power, low-cost active matrix flat panel imagers. Despite the shortcomings associated with the instability and low mobility of TFTs, the amorphous silicon TFT technology still remains the primary solution for the backplane of flat panel imagers. The use of a-Si:H TFTs as the building block of the large area integrated circuit becomes challenging particularly when the role of the TFT is extended from traditional switching applications to on-pixel signal amplifier for large area digital imaging. This is the idea behind active pixel sensor (APS) architectures in which under each pixel an amplifier circuit consisting of one or two switching TFTs integrated with one amplifying TFT is fabricated. To take advantage of the full potential of these amplifiers, it is crucial to develop APS architectures to compensate for the limitations of the TFTs. In this thesis several APS architectures are designed, simulated, fabricated, and tested addressing these challenges using the mask sets presented in Appendix A. The proposed APS architectures can compensate for inherent stabilities of the comprising TFTs. Therefore, the sensitivity of their output data to the transistor variations is significantly suppressed. This is achieved by using a well defined external current source instead of the traditional voltage source to reset the APS architectures during the reset cycle of their periodic operation. The performance of these circuits is analyzed in terms of their stability, settling time, noise, and temperature-dependence. For appropriate readout of the current mode APS architectures, high gain transresistance amplifiers with correlated double sampling capability is designed, simulated and fabricated in CMOS technology. Measurement and measurement based calculation results reveal that the proposed APS architectures can meet even the stringent requirements of low noise, real-time digital fluoroscopy.

A network of n transmitters and n receivers is considered. We assume that transmitter i aims to send data to its designated destination, receiver i. Communications occur in a single-hop fashion and destination nodes are simple linear receivers without multi-user detection. Therefore, in each time slot every source node can only talk to one other destination node. Thus, there is a total of n communication links. An important question now arises. How many links can be active in such a network so that each of them supports a minimum rate Rmin? This dissertation is devoted to this problem and tries to solve it in two di erent settings, dense and extended networks. In both settings our approach is asymptotic, meaning, we only examine the behaviour of the network when the number of nodes tends to in nity. We are also interested in the events that occur asymptotically almost surely (a.a.s.), i.e., events that have probabilities approaching one as the size of the networks gets large. In the rst part of the thesis, we consider a dense network where fading is the dominant factor a ecting the quality of transmissions. Rayliegh channels are used to model the impact of fading. It is shown that a.a.s. log(n)^2 links can simultaneously maintain Rmin and thus be active. In the second part, an extended network is considered where nodes are distant from each other and thus, a more complete model must take internode distances into account. We will show that in this case, almost all of the links can be active while maintaining the minimum rate.

An investor can generate higher returns on the German stock market if he is using an active portfolio management strategy rather than its passive counterpart. This is possible because the market is not efficient and the DAX, namely the market portfolio, can be outperformed in regard to the average annual return and its variance. Therefore, the CAPM does not hold for the German stock market. The investor has to use the 10 weeks old changes of the ifo business climate index to forecast the DAX movement in the upcoming month. Even though this forecasting method only gave the correct trading signal for 56% of the months between 1991 and 2011, it outperformed the Buy and Hold strategy by 324 basis points. The main reason for this is that the business index was able to warn the investor of months in which the DAX lost over 10% of its value. The superiority of the active strategy was still valid when transaction costs were taken into account and was even stronger when call money was the alternative investment to the DAX rather than cash.

A superimposed steering system is the combination of a conventional steering system with an electric motor which is used to alter the steering angle imposed by the driver. The potential benefits are increased agility, automatic compensation for lateral wind forces and decreased braking distance (in combination with an electronic stability program). In this thesis we implement a model and a controller for a superimposed steering system thus achieving the first of these potential benefits. The vehicle model is based on the single-track or bicycle model. Unlike most other publications, the motor model in this thesis goes down to the level of the electrical dynamics of the motor. The model is divided into three main modules (vehicle module, steering module and friction module) as well as several submodules to ensure easy adaptability. The overall control objective consists of increasing vehicle agility by achieving a variable ratio between the steering wheel angle and the actual road wheel angle as a function vehicle velocity. We divide the controller into a torque and a current controller. The actual controller is derived in three steps starting from an analog torque controller as well as an analog current controller then moving to a digital torque controller. In doing so we use the model matching, feedback linearization and state feedback control techniques. The model and the controller are evaluated using the parameters of a small truck and different road scenarios. Finally, the Validation Square technique is applied to assess the validity of the results.

Thanks to development of power semiconductor devices and integrated circuits, active front-end converters with controllability of bidirectional power flow have become popular and viable in industrial applications. This thesis proposes an improved resonant current control for the active front-end converter with LCL filter. The proposed control consists of a band-pass filter tuned at fundamental frequency and various band-rejected filters resonant at harmonic frequencies to provide fundamental current tracking capability as well as enhance harmonic current rejection. Based on this algorithm, the active front-end converter can control dc voltage with unity power factor by sensing converter output current, LCL filter voltage and dc voltage. This approach also conducts harmonic current rejection under distorted line voltage with no phase-locked-loop used, which is the significant advantage in terms of phase lag of frame transformation and computing effort of digital signal processing. Current tracking performance and harmonic rejection capability of the proposed method are verified based on frequency-domain analysis. Computer simulations and experimental results are also implemented to validate effectiveness.