No description available.

This thesis describes the research carried out in overcoming the problems encountered during the development of an autonomous mobile robot (AMR) which uses a single television camera for navigation in environments with visible edges, such as corridors and hallways. The objective was to determine the minimal sensing and signal processing requirements for a real AMR that could achieve self-steering, navigation and obstacle avoidance in real unmodified environments. A goal was to design algorithms that could meet the objective while being able to run on a laptop personal computer (PC). This constraint confined the research to computationally efficient algorithms and memory management techniques. The methods by which the objective was successfully achieved are described. A number of noise reduction and feature extraction algorithms have been tested to determine their suitability in this type of environment, and where possible these have been modified to improve their performance. The current methods of locating lines of perspective and vanishing points in images are described, and a novel algorithm has been devised for this application which is more efficient in both its memory usage and execution time. A novel obstacle avoidance mechanism is described which is shown to provide the low level piloting capability necessary to deal with unexpected situations. The difficulties of using a single camera are described, and it is shown that a second camera is required in order to provide robust performance. A prototype AMR was built and used to demonstrate reliable navigation and obstacle avoidance in real time in real corridors. Test results indicate that the prototype could be developed into a competitively priced commercial service robot.

No description available.

Coherent optical data processing is recognised to be a natural solution to certain information processing problems. Attempts to exploit the benefits of optical processing are limited by the quality of available Spatial Light Modulators. Spatial Light Modulators are devices which controllably impress information onto the amplitude or phase of an optical wavefront. They are used both to input data into an opticaI system and as modulating elements within the system (often in the Fourier plane of a Fourier transform processor). This thesis describes the successful development of an electronically addressed spatial light modulator using liquid crystal as the light modulating material and a silicon integrated circuit as the addressing medium. It is a pixelated binary spatial light modulator operating in reflection. Each pixel contains a memory element which stores the programmed logical state of the pixel. The addressing and pixel circuits were fabricated in a 1.5m nMOS technology on a 10 mm square chip. Th e pixels are arranged on a square array containing 50 x 50 elements. The liquid crystal was configured to modulate the light amplitude using the hybrid field effect in a nematic liquid crystal. The spatial light modulator is used as a Fourier plane filter in a coherent optical processing system. Its performance is assessed and the direction of future research into this type of spatial light modulator is discussed.

This thesis brings together ideas from psychology (particularly the work of Sever and Townsend 1979) and from linguistics (particularly the work of Partee, 1983 and Moens and Steedman 1986) about the nature of temporal representation, especially with regard to the effect of different temporal connectives on language processing. Experiments in the second chapter looked at memory for temporal information and order and results indicated that information about temporal order is less well remembered that information about spatial order. The third chapter examines the role of before as introducing sentences that can be either factive or non-facrive and concluded mat there is no clear divide between these two types of sentence, the difference depends on knowledge of regularities in the world, and it is only with clearly non-factive before sentences that readers have definite expectations . about factivity. This theme is continued in the fourth chapter which looks at SALIENCE, by examining continuations from sentences with temporal connectives and notes that a tendency to continue text from the main clause is modified by an effect of continuing from the las: occurring event, irrespective of order in the text. The fifth chapter examines the effect of context for sentence processing and concludes that context consistent with the main clause of a sentence is preferred. It also shows that similar processes are involved in building up a model containing temporal information to those involved in building a model of a spatial array. The last experiment demonstrates that lack of a clear temporal referent disrupts language processing in the same way as lack of a clear antecedent for a pronoun does. The results are discussed in terms of a possible model for representation which would include events being represented in a form similar to a "nucleus".

At the heart of many power system computations lies the solution of a large sparse set of linear equations. These equations arise from the modelling of the network and are the cause of a computational bottleneck in power system analysis applications. Efficient sequential techniques have been developed to solve these equations but the solution is still too slow for applications such as real-time dynamic simulation and on-line security analysis. Parallel computing techniques have been explored in the attempt to find faster solutions but the methods developed to date have not efficiently exploited the full power of parallel processing. This thesis considers the solution of the linear network equations encountered in power system computations. Based on the insight provided by the elimination tree, it is proposed that a novel matrix structure is adopted to allow the exploitation of parallelism which exists within the cutset of a typical parallel solution. Using this matrix structure it is possible to reduce the size of the sequential part of the problem and to increase the speed and efficiency of typical LU-based parallel solution. A method for transforming the admittance matrix into the required form is presented along with network partitioning and load balancing techniques. Sequential solution techniques are considered and existing parallel methods are surveyed to determine their strengths and weaknesses. Combining the benefits of existing solutions with the new matrix structure allows an improved LU-based parallel solution to be derived. A simulation of the improved LU solution is used to show the improvements in performance over a standard LU-based solution that result from the adoption of the new techniques. The results of a multiprocessor implementation of the method are presented and the new method is shown to have a better performance than existing methods for distributed memory multiprocessors.

No description available.

Micro-affordance effects have been reported for several different components of the reachto- grasp action during on-line visual processing (Tucker and Ellis, 1998; Ellis and Tucker, 2000; and Tucker and Ellis, 2001). One property of these effects is that they have been shown to terminate once an object is removed from view (Tucker and Ellis, 2001). This thesis describes eight experiments that examine the presence of micro-affordance effects during off-line visual processing. All eight experiments employ a stimulus-response compatibility paradigm. Three different experimental designs were employed to examine the presence of micro-affordance effects arising from the relationship between: (a) the power and precision component of the reach-to-grasp action and the compatibility of an object for grasping with either a power or precision grasp, and (b) the orientation of an object for grasping and hand of response. The results of the experimentss uggestt hat: (a) the representationsu tilised during off-line visual processing can potentiate actions arising from the two components of the reach-tograsp action investigated;( b) the representationsu tilised during off-line visual processing can also inhibit micro-affordance effects; (c) main effects of object orientation (faster response times to either left or right-oriented objects) in those experiments examining the relationship between the orientation of an object for grasping and hand of response can be used to support a theory for the existenceo f prototype object representationsh, eld in long term memory, for the process of object recognition, and (d) due to differences in the object properties thought to give rise to micro-affordance effects, and the existence of different off-line visual processes,d ifferent experimentald esignsa re required to elicit microaffordance effects arising from the two types of micro-affordance effects investigated in this thesis.

Neural network analysis is proposed and evaluated as a method of analysis of marine biological data, specifically images of plankton specimens. The quantification of the various plankton species is of great scientific importance, from modelling global climatic change to predicting the economic effects of toxic red tides. A preliminary evaluation of the neural network technique is made by the development of a back-propagation system that successfully learns to distinguish between two co-occurring morphologically similar species from the North Atlantic Ocean, namely Ceratium arcticum and C. longipes. Various techniques are developed to handle the indeterminately labelled source data, pre-process the images and successfully train the networks. An analysis of the network solutions is made, and some consideration given to how the system might be extended.

No description available.