The need to develop highly sensitive, selective, and cost-effective biosensors spans the areas of medicine, the environment, food safety, and homeland security. Accurate and reliable detection of trace quantities of small molecules is particularly challenging for existing sensor technology. Porous silicon is an excellent material for small molecule biosensing due to its large surface area and its capability of size selective sensing. In this work, a label-free porous silicon optical waveguide is demonstrated for the detection of short DNA oligonucleotides. The waveguide design enables the optical field to be localized exactly where biomolecules are attached to the sensor, leading to enhanced detection sensitivity. A prism is used to couple light into the waveguide at a particular angle where a sharp resonance dip is observed. The resonance angle is very sensitive to small changes in the porous silicon waveguide refractive index that occur, for example, when biomolecules bind inside the pores of the waveguide. Two different porous silicon waveguide structures are presented and compared: a p-type porous silicon waveguide with 20-30 nm pores in the Otto configuration and an n-type porous silicon waveguide with 100 nm pores in the Kretschmann configuration. Field profiles of the porous silicon waveguides are analyzed and their sensitivity enhancement over traditional sensor technology is shown for small molecule detection. In addition to experimental demonstration of quantitative and selective detection of DNA molecules using the porous silicon waveguide biosensor, the effects of biomolecular size on the sensitivity of porous silicon waveguide biosensors is discussed. With smaller pores, p-type porous silicon waveguides are more sensitive detecting shorter DNA molecules, while larger pore n-type porous silicon waveguides are capable of detecting longer DNA molecules that cannot easily infiltrate into p-type waveguides. Finally, further research is suggested to improve the sensor performance through modified design, alternate biomolecules to detect, and systems integration.

Recent advances in robotics and intelligent systems are expected to usher in a new era where the need for machines to understand humans becomes increasingly important. It should permit more meaningful and natural human-machine interaction (HMI) when a robot/computer can detect the affective cues of the person it is working with. The objective of this work is to investigate the following hypotheses for achieving an affect-sensitive HMI: (i) It is possible to detect the affective states of interest by using multiple indices derived from physiological signals in real-time; (ii) Such affective cues can be integrated within a machine's control architecture to make it capable of responding to them appropriately; and (iii) Such affect-sensitive systems are expected to improve the overall human-machine interaction experience. In this work, a systematic comparison of the strengths and weaknesses of machine learning methods was performed when they were employed for the physiology-based affect recognition. The impacts of the affect-sensitive closed-loop interaction were investigated in both human-robot interaction (HRI) and human-computer interaction (HCI) contexts. Furthermore, in response to the growing need for developing robot/computer assisted autism intervention systems for children with autism spectrum disorder (ASD), physiology-based affective modeling and adaptation methods were investigated for this specific population. Finally, physiology-based affective modeling using active learning for children with ASD was discussed.

This thesis develops techniques for accurately estimating the time-of-arrival (TOA) on resource constrained sensor units of a distributed radio geolocation system. Firstly, a multi-resolution approach for discriminating between signals of interest and spurious transmissions, which reduces computational costs involved in the discrimination operation by nearly two orders of magnitude, is presented. Secondly, Doppler frequency shifting is proposed as a solution for minimizing the effect of drifting sample rate clocks on different units. To avoid the computational costs associated with the Doppler frequency shifting solution, a related time-domain shifting technique is developed and analyzed. Finally, the problems of estimating a unitâs operating frequency and GPS jitter problem are posed together as a linear regression problem. It was shown that the proposed compensation techniques reduce the error contribution from sampling frequency disparity and GPS jitter from approximately 500 ns down to about 80 ns.

Silicon-germanium heterojunction bipolar transistor (SiGe HBT) BiCMOS technology is recognized by the space electronics community for its potential to transform high-speed microelectronic applications by monolithic incorporation of low-power complementary metal oxide semiconductor logic with high-speed SiGe HBT building blocks. However, SiGe HBTs suffer from a low single-event upset threshold and a large saturated cross section, two traits that make them liabilities for use in space-base applications. The deep trench isolation, n+ subcollector, and lightly-doped p-type substrate are the dominant SiGe HBT process features that influence single-event effects. These features control the single-event upset response as well as single-event current induction. This work presents a single-event rate prediction model for SiGe HBTs that takes these features into account as well as the first complete collection of measured wide bandwidth single-event current transients, including pulsed laser, heavy ion microbeam, and heavy ion broadbeam radiation sources. These transient data confirm important single-event upset mechanisms and provide a calibration baseline for future transient experiments and simulations.

DYNAMIC MODELING AND CONTROL OF NONHOLONOMIC WHEELED MOBILE ROBOT SUBJECTED TO WHEEL SLIP<P> SHAHRUL NAIM SIDEK<P> Improving navigation performance of autonomous wheeled mobile robot (WMR) in a dynamic unstructured environment requires improved maneuverability. In such cases, the dynamics of wheel slip may violate the ideal no-slip kinematic constraints generally used to model nonholonomic WMR. In this dissertation, a new model is proposed to tackle the modeling inadequacy that arises when slip is neglected by including both longitudinal and lateral slip dynamics into the overall dynamics of the WMR. The presented model of the WMR provides a realistic simulation scenario that can be utilized to develop model-based controllers to improve WMR navigation. In particular, to demonstrate how this model can be useful in developing model-based planning and control of WMR with wheel slip, a dynamic path-following controller is designed to allow the WMR to navigate efficiently by autonomously regulate its forward velocity based on the generated traction force at the wheel-surface contact point. Extensive simulation results show the importance of the proposed modeling technique to capture slip phenomenon and the efficacy of the presented control technique to exploit such slip for better navigation performance. In addition, through experiments, the dynamics of slip in the system model has been verified.

Soft errors in integrated circuits (ICs) are a critical problem facing state-of-the-art technologies. In both the terrestrial and space environment, the source of soft errors is charged particle interaction with ICs. This work examines the effects of multiple soft errors from a single particle interaction. In memory devices, clusters of physically adjacent soft errors are referred to as multiple cell upsets (MCUs). In this work, the impact of device orientation on the MCU response from accelerated heavy ion and neutron testing is analyzed. The size, shape, and probability of MCU are shown to depend on orientation for both particle types. The worst case MCU events occur at large angles of incidence. Additionally, heavy ions also exhibit a strong dependence on the ion's trajectory with respect to the SRAM layout for the size and shape of MCU events.

Single event latchup (SEL) has been observed on a range of different devices over the past three decades and can result in large currents on metal interconnects, resulting in long-term device reliability issues or catastrophic failure. Because of this, it is important to thoroughly screen parts for SEL vulnerability. In this dissertation, both technology computer aided design (TCAD) simulations and experimental data provide support for a variation in device vulnerability based on the lateral orientation for grazing angle ion strikes. These results show industry standard tests for single event effects (SEEs) may be inadequate for device validation and characterization.

Radiation-induced soft errors have become a key reliability issue for advanced semiconductor integrated circuits. With technology scaling, a large fraction of the observed soft failures are estimated to be related to latched single-event transients (SETs). Precise knowledge of the particle-induced transient pulse widths is important for determining error rates and for the design of hardening techniques to mitigate the effect of these transients. <p> This work presents a novel, autonomous pulse characterization circuit technique to measure the distribution of SET pulse widths for different radiation environments. The pulse characterization technique has been implemented in a range of CMOS technologies and test chips have been used to measure the distribution of SET pulse widths for heavy-ions, neutrons and alpha particles. The dissertation focuses on test results from IBM 130-nm and 90-nm bulk CMOS processes. <p> Heavy-ion SET measurements show a reduction in the threshold for a measurable SET from about 7 MeV-cm^2/mg for 130-nm to less than 2 MeV-cm^2/mg for 90-nm. SET pulse widths ranging from about hundred ps to over 1 ns were measured with heavy-ions in 130-nm and 90-nm processes and the pulse widths were found to increase when scaling from 130-nm to 90-nm. Technology scaling trends in SET pulse widths are explained based on experimental measurements and with the use of mixed-mode 3D-TCAD simulations. Reasons for long transients measured at low LETs in the 90-nm process, including the pulse broadening phenomenon, are examined. Results indicate that lower drive currents and reduced contact size to the well region are factors that lead to an increase in SET pulse widths with scaling. The presence of a parasitic bipolar charge collection in these processes, triggered by a well potential collapse effect, leads to wider transients. Simulation and experimental results illustrate that the use of larger contacts to the well region mitigates the well potential collapse and hence limits the SET pulse width. <p> The SET measurements reported in this work are the first-ever for neutrons and alpha particles and these results are important for predicting error rates for commercial terrestrial applications. Most neutron and alpha particle-induced SETs were found to be of the order of hundreds of picoseconds. Neutron and alpha failures-in-time (FIT) rates were found to be about 10^-5 FIT/inverter.

Using Monte Carlo based simulations, the proton-induced energy deposition in a silicon PIN focal plane array was analyzed, and the importance of considering the materials surrounding a device was shown by comparing the results with experiment. This includes materials around all sides of the device, even those a centimeter away. During ground testing, caution must be used when setting up the experiment because dewar windows and mountings can also affect the amount of energy deposition observed in the device. Failure to include materials below a structure during simulation will result in an underprediction of the response for devices with high sensitivity to single events, such as detector arrays. This work shows that a high fidelity simulation is needed to estimate the energy deposited. MRED (Monte Carlo Radiative Energy Deposition) simulations on a silicon imaging device show that direct ionization is the dominant mechanism for energy deposition below 350 keV in the focal plane detector considered here, while nuclear reactions dominate at higher energies. Even though direct ionization is the dominant mechanism, a constant LET and path length calculation does not address the fluctuations in dE/dx, only the variation in path length, and therefore does not capture the shape of the differential distribution. Modeling codes that use only a single value LET will fail to predict the response accurately. MRED simulations show that, when an isotropic beam is considered and an on-orbit event rate calculated, a simple structure including only one pixel and excluding surrounding materials to obtain error rate calculations is sufficient, with significantly improved accuracy over CREME96 calculations.

Abstract <p> This thesis evaluates the performance of two non-rigid image registration techniques. The Moving Least Squares (MLS) technique is compared to the widely used Thin-plate Spline (TPS) method. Both methods interpolate a set of fiducial points in registering two images. An attractive feature of the MLS method is that it seeks to minimize local scaling and shearing, producing a global transformation that is as rigid as possible. The MLS and TPS techniques are applied to two- and three-dimensional medical images. Both qualitative and quantitative comparisons are presented. The two techniques are quantitatively evaluated by computing target registration errors (TREs) at selected points of interest. Our results indicate that the MLS algorithm performs better than the TPS method with lower TRE values and visually better registered images, indicating that MLS may be a better candidate for registration tasks when rigid registration is insufficient but the deformation field is sought to be minimal.