The problem considered in this thesis is the development of an autonomous prototype robot capable of gathering sensory information from its environment allowing it to provide feedback on the condition of specific targets to aid in maintenance of hydro equipment. The context for the solution to this problem is based on the power grid environment operated by the local hydro utility. The intent is to monitor power line structures by travelling along skywire located at the top of towers, providing a view of everything beneath it including, for example, insulators, conductors, and towers. The contribution of this thesis is a novel robot design with the potential to prevent hazardous situations and the use of rough coverage feedback modified reinforcement learning algorithms to establish behaviours. / October 2007

This dissertation describes the design and implementation of a system that has been designed to extract 3D information from pairs of 2D images. System input consists of two images taken by an ordinary digital camera. System output is a full 3D model extracted from 2D images. There are no assumptions about the positions of the cameras during the time when the images are being taken, but the scene must not undergo any modifications. The process of extracting 3D information from 2D images consists of three basic steps. First, point matching is performed. The main contribution of this step is the introduction of an approach to matching image segments in the context of an approximation space. The second step copes with the problem of estimating external camera parameters. The proposed solution to this problem uses 3D geometry rather than the fundamental matrix widely used in 2D to 3D conversion. In the proposed approach (DirectGS), the distances between reprojected rays for all image points are minimised. The contribution of the approach considered in this step is a definition of an optimal search space for solving the 2D to 3D conversion problem and introduction of an efficient algorithm that minimises reprojection error. In the third step, the problem of dense matching is considered. The contribution of this step is the introduction of a proposed approach to dense matching of 3D object structures that utilises the presence of points on lines in 3D space. The theory and experiments developed for this dissertation demonstrate the usefulness of the proposed system in the process of digitizing 3D information. The main advantage of the proposed approach is its low cost, simplicity in use for an untrained user and the high precision of reconstructed objects. / October 2007

This thesis presents a rough set approach to reinforcement learning. This is made possible by considering behaviour patterns of learning agents in the context of approximation spaces. Rough set theory introduced by Zdzisław Pawlak in the early 1980s provides a ground for deriving pattern-based rewards within approximation spaces. Learning can be considered episodic. The framework provided by an approximation space makes it possible to derive pattern-based reference rewards at the end of each episode. Reference rewards provide a standard for reinforcement comparison as well as the actor-critic method of reinforcement learning. In addition, approximation spaces provide a basis for deriving episodic weights that provide a basis for a new form of off-policy Monte Carlo learning control method. A number of conventional and pattern-based reinforcement learning methods are investigated in this thesis. In addition, this thesis introduces two learning environments used to compare the algorithms. The first is a Monocular Vision System used to track a moving target. The second is an artificial ecosystem testbed that makes it possible to study swarm behaviour by collections of biologically-inspired bots. The simulated ecosystem has an ethological basis inspired by the work of Niko Tinbergen, who introduced in the 1960s methods of observing and explaining the behaviour of biological organisms that carry over into the study of the behaviour of interacting robotic devices that cooperate to survive and to carry out highly specialized tasks. Agent behaviour during each episode is recorded in a decision table called an ethogram, which records features such as states, proximate causes, responses (actions), action preferences, rewards and decisions (actions chosen and actions rejected). At all times an agent follows a policy that maps perceived states of the environment to actions. The goal of the learning algorithms is to find an optimal policy in a non-stationary environment. The results of the learning experiments with seven forms of reinforcement learning are given. The contribution of this thesis is a comprehensive introduction to a pattern-based evaluation of behaviour during reinforcement learning using approximation spaces. / May 2006

This thesis presents the design of a tactile device that can be used to display varying magnitudes of roughness. The device is designed to be attached to an existing force feedback device in order to create a package that is able to display both macro-level (force feedback) and micro-level (tactile feedback) information to the users. This device allows the users to feel a simulated texture by placing an index finger on an aperture. The stimulus is created with a spiral brush made of nylon bristles. The brush is attached to a DC motor and the speed and direction of rotation of the brush are used to generate textures at the fingertip through the aperture. Three psychophysical experiments are conducted to study the effects of speed and direction on the roughness perception. The first experiment is designed to investigate the sensitivity to a change in the speed of the brush. This experiment is conducted for two levels of base speed and it is found that as the base speed increases, the just noticeable difference (JND) with respect to speed decreases. In the second experiment, it is found that this tactile device is able to represent textures of rough nature, such as sandpaper. It is also found that the human roughness perception cannot be described in a unique manner. Two opposite definitions of rough textures are identified in this experiment. While some users relate an increase in the speed of the brush to increasing roughness, others relate it to decreasing roughness. Further, the results show that the effects of direction are insignificant on the roughness perception for both groups of users. In the third experiment, the effects of direction are studied more closely by presenting the two directions successively with a time gap of $0.5s$. It is found that with this small time gap, the users are able to discriminate between directions, unlike in the previous experiment. The roughness perception is affected by the change in direction when the time gap is small. These findings open further areas that need to be investigated before a robust tactile device can be designed.

This thesis presents the design of a tactile device that can be used to display varying magnitudes of roughness. The device is designed to be attached to an existing force feedback device in order to create a package that is able to display both macro-level (force feedback) and micro-level (tactile feedback) information to the users. This device allows the users to feel a simulated texture by placing an index finger on an aperture. The stimulus is created with a spiral brush made of nylon bristles. The brush is attached to a DC motor and the speed and direction of rotation of the brush are used to generate textures at the fingertip through the aperture. Three psychophysical experiments are conducted to study the effects of speed and direction on the roughness perception. The first experiment is designed to investigate the sensitivity to a change in the speed of the brush. This experiment is conducted for two levels of base speed and it is found that as the base speed increases, the just noticeable difference (JND) with respect to speed decreases. In the second experiment, it is found that this tactile device is able to represent textures of rough nature, such as sandpaper. It is also found that the human roughness perception cannot be described in a unique manner. Two opposite definitions of rough textures are identified in this experiment. While some users relate an increase in the speed of the brush to increasing roughness, others relate it to decreasing roughness. Further, the results show that the effects of direction are insignificant on the roughness perception for both groups of users. In the third experiment, the effects of direction are studied more closely by presenting the two directions successively with a time gap of $0.5s$. It is found that with this small time gap, the users are able to discriminate between directions, unlike in the previous experiment. The roughness perception is affected by the change in direction when the time gap is small. These findings open further areas that need to be investigated before a robust tactile device can be designed.

A 4-year study was conducted on the Missouri Coteau in the Mixed Grassland Ecoregion of Saskatchewan to determine the effects of mowing to a 7.5 cm stubble height on the growth of <i>Festuca hallii</i> (Vasey) Piper and <i>Stipa curtiseta</i> (A.S. Hitch.) Barkworth. Green standing crop (GSC), dead standing crop (DSC) and above ground net primary production (ANPP) were compared to an unmowed control after a single mowing in April, May, June, July, August, September, October or November on 5 landforms including north aspect-concave-slope, north aspect-convex-slope, south aspect-concave-slope, south aspect-convex-slope and level upland. Mowing reduced GSC, DSC and ANPP with reductions varying among months of mowing and among landforms. Green standing crop, DSC and ANPP of <i>F. hallii</i> were greatest on the north aspects and least on south aspect-convex slope and ranged from 2 to 122 g m-2, 3 to 121 g m-2, and 8 to 122 g m-2, respectively. Mowing reduced GSC of <i>F. hallii</i> for 1 to 11 growing season months and DSC for 1 to >11 growing season months. Mowing in May or November reduced ANPP of <i>F. hallii</i> for 1 growing season, while mowing in other months reduced ANPP for 2 growing seasons. Green standing crop of <i>S. curtiseta</i>, ranging from 3 g m-2 to 55 g m-2, was least on the north aspects and greatest on the south aspect-convex slope and was reduced 1 to 5 growing season months following mowing. Mowing after June reduced DSC of <i>S. curtiseta</i>(5 to 58 g m-2) for 1 to 10 growing season months, and ANPP (6 to 64 g m-2) for 1 growing season. Generally, mowing reduced GSC, DSC and ANPP of <i>F. hallii</i> longer than <i>S. curtiseta</i>. Production of <i>F. hallii</i>- and <i>S. curtiseta</i>-dominated plant communities in the Northern Mixed Prairie will be maintained by providing rest periods between defoliation events based on the number of growing season months for <i>F. hallii</i> to recover production.

Urea is an important nitrogen fertilizer for plant nutrition, but is very susceptible to moisture sorption and caking even at low moisture contents, e.g. 0.25% w/w. When urea particles adsorb moisture followed by drying, crystal bridges form between urea particles. For particles in a bed, this process is called caking. Cakes in stored urea cause a degradation of its quality and value. Investigations of the moisture absorption in beds of manufactured urea particles and adsorption on the external and internal surfaces of urea particles are a necessary step if engineers are to recommend procedures to reduce caking and control inventories. Research on moisture adsorption and cake strength of urea fertilizer has not been sufficiently explored. Only recently have researchers started to devise tests to investigate the crystal bonding between two urea particles. Prior to this research, investigations of the moisture interactions in beds of urea were nearly non-existent. This thesis presents experimental, theoretical and numerical methods to investigate the coupled heat and moisture transfer processes in a bed of urea particles while the bed is exposed to ambient air with changing temperature and humidity.<p> Urea particles are nearly spherical with uniform particle size distribution. The particle size, its internal pore structure and rough crystalline external surface depend on the manufacturing process. In this thesis, two types of urea products are investigated, i.e. prill Georgia urea and granular Terico urea. The rough external surface and internal pore structure of each particle makes the total surface area exposed to water much larger than similar smooth and solid spherical particles. Although Georgia urea has higher external surface area than Terico urea, the latter type has larger total surface area and internal pore volume. For both Terico urea and Georgia, the internal surface area dominates the water sorption process but the external moisture sorption of Georgia urea is more important than that of Terico urea.<p> All the water vapor interaction experiments were carried out with air flow through a test bed because it shortens the duration of each experiment to a few hours in most cases. A series of experiments with step changes in inlet air temperature and humidity for air flow through a urea bed indicated that the measured outlet air temperature and humidity responses, each at a specific air flow rate, reveals a typical exponential or transient time change that can be characterized by a time constant. After formulating the theoretical problem for step changes in the inlet properties, the analytical solutions showed that the time constants of outlet response to whether a temperature step change or a humidity step change are functions of the convection coefficient and air velocity. The predicted outlet air temperature is determined by only one time constant for a temperature step change while it is determined by these two time constants for a humidity step change.<p> A new test cell with sampling test ports was developed to measure the transient moisture uptake of a urea particle bed and its distribution at any time without any interruption of the experiment. A novel particle sampling device, modified from a syringe and pistons, was designed to minimize the particle exposure to ambient air during the moisture content determination using a Karl Fischer titrator. Data from two continuous cyclic step changes in the inlet flow with relative humidities between 4% and 70% at room temperature showed a hysteresis in the isothermal moisture content for only the first cycle. After the second sorption- desorption cycle, the hysteresis disappeared. This implies that the internal pore and particle surface geometry changes are very slow after the first cycle.<p> A new theoretical porous media model was developed for a coupled heat and moisture transport process when humid air flowed uniformly through a large test bed in two coupled computational domains: internal domain (i.e., the particle phase) and the external domain (i.e., the interstitial air space). The moisture migration in two computational domains included: water vapor diffusion inside each particle, and water vapor convection and diffusion in the interstitial air space in the urea particle bed. For energy transport, the temperature was assumed to be uniform inside each particle, but heat convection and conduction between the urea particles and the interstitial air outside particles occurred throughout the bed. Both heat transfer and mass transfer in internal domain and external domain were coupled by the heat and mass convection at the gas-particle interface. The numerical simulation was compared with the data of moisture uptake and showed good agreement implying that the internal moisture diffusion that dominates the moisture uptake process is a very slow process.<p> These above experimental, theoretical and numerical research studies provide a set of information on how urea particles adsorb or desorb moisture from or to ambient air on the external and internal pore surface, which offers a useful suggestion for urea caking prevention and is also a first and necessary step to the study of further caking formation and strength.

The bidirectional reflectance distribution function (BRDF) that describes the scattered energy distribution is the most fundamental radiative property to calculate other properties. Although recent progress in surface metrology allows topography measurement in an atomic level, most studies still assume statistical distributions of roughness because of difficulty in roughness modeling. If the BRDF of rough silicon wafers is modeled with assumptions, predicted radiative properties may be inaccurate because non-Gaussian and anisotropic roughness of some wafers cannot be approximated with known statistics. Therefore, this thesis focuses on development of BRDF modeling that accounts for anisotropic roughness to accurately predict radiative properties of rough silicon surfaces with thin-film coatings. Monte Carlo ray-tracing methods are developed to consider multiple scattering and the change of polarization states and to satisfy physical laws such as the reciprocity principle. Silicon surface topographic data measured with an atomic force microscope are incorporated into the ray-tracing algorithms to model anisotropic roughness statistics. For validation, BRDF and emittance predictions are compared with measurements using an optical scatterometer and an integrating sphere. Good agreement between prediction and measurement demonstrates that the incorporation of topography measurement into BRDF modeling is essential for accurate property prediction. Roughness effects on the BRDF are so strong that BRDFs also reveal anisotropic features regardless of the presence of coating. Anisotropic roughness increases multiple scattering although first-order scattering is dominant, and thus enhances emittance noticeably. Silicon dioxide coating changes the magnitude of BRDF and emittance and reduces the anisotropic roughness effect on emittance enhancement. The research in this thesis advances the method to predict radiative properties by incorporating anisotropic rough statistics into BRDF modeling.

Reverberation is the phenomenon when the sound source transmits and causes scattering in active sonar system. This kind of effects often produced in the waveguide, resulting signal interference and signal mask issues. Reverberation can affect the signal to noise ratio, thus understanding the characteristic of environmental reverberation is important. In recent years, there were many studies for the Sizihwan Bay Marine Test Field (MTF), including environmental surveys, Harbor defense and acoustic inversion, however the issue related to reverberation has not been explored. The purpose of this study is to research reverberation in the MTF and focusing on volume reverberation and surface reverberation. In the past, the experiment of Underwater Intruder Detection with active sonar system demonstrated the reverberation in Kaohsiung second harbor. Therefore, this research is about using the experimental data to questions about volume reverberation. Results show, volume reverberation in the port area caused by ship, boundaries, current, impurities in water and biota. On the other hand surface reverberation, this study focusing on numerical simulation, match the results of experimental. Numerical results of RMS height, correlation length and frequency affect the reverberation intensity, but it does not identify the phenomenon of reverberation in experimental results, mainly is the intensity of the source is not enough. This study combined with simulation and experiment, and overviewed the reverberation properties in MTF. Also provided suggestions for following studies.

Omni-directional acoustic wave source interactions with a rough seabed with a continuously varying density and sound speed in a fluid-like sediment layer. The acoustic properties in the sediment layer possess an exponential type of variation in density and one of the three classes of sound speed profiles, which are constant,~$k^2$-linear, or inverse-square variations. Analytical solution of mean field. The mean field reflection coefficients corresponding to the aforementioned density and sound speed profiles for various frequencies, roughness parameters, are numberically generated and analyzed. Physical interpretations are provided for various results. This simple model characterizes two important features of sea floor, including seabed roughness, sediment inhomogenieties, therefore, provide a canonical analysis in seabed acoustics.