Global ETD Search

1	Learning and generalization in feed-forward neural networks Smieja, Frank Jan January 1989 (has links) No description available. 003.5 Bionics
2	Environment sensing and modelling, path finding and position location for a mobile robot Seals, Richard Charles January 1985 (has links) The results of previous research into mobile robotics were studied and several fundamental properties deduced. These were then considered in detail and optimum techniques selected for implementation on a prototype mobile robot. Several constraints were placed .on the prototype mobile robot, such as the potential for untethered use (i.e. no umbilical connections) and minimum restrictions on the environment. The investigation was concentrated in four main areas: environment modelling, path selection and following, absolute position location and environment sensors. An initial simulation was implemented on a mainframe computer which used an x-Y grid square model of the environment with a simple scanning rangefinder, to investigate the usefulness of the Means-end path finding algorithm. Results were obtained for varying states of the environment (i.e. known, unknown and partially known) which indicated that the path finding algori thm was sui table for implementation on the prototype mobile robot. An improved environment model which used a quasi-continuous X-Y cartesian coordinate system was then constructed. This was designed to enable environment data from the simulated or prototype scanning rangefinders to be used as input to the control processes, wi th movement of the simulated or prototype mobile robot as the output. In this way the mobile robot was able to find paths using the model of the environment and then attempt to follow them in the physical envi ronment. Al ternati vely, the prototype mobile robot was able to find a path through an unknown environment using data from the rangefinder only. In addi tion, a posi tion location process was implemented which operated by identifying the position of known objects in the environment by matching the distribution of three of the detected objects agai nst the di stri bution of all known sets of three objects. Once the objects had been identified the posi tion of the prototype mobile robot was calculated. The preliminary results indicated that the position could be found using this technique but that more investigation was required to reduce the ambiguity of the results. 003.5 Bionics
3	A flexible bidirectional question-answering system Levine, John Michael January 1992 (has links) No description available. 003.5 Bionics
4	The use of explicit intermediate representations in intelligent tutoring systems Twidale, Michael Bernard January 1989 (has links) No description available. 003.5 Bionics
5	Machine learning and the acquisition of natural concepts based on the psychological theory of prototypes Musgrove, P. B. January 1989 (has links) No description available. 003.5 Bionics
6	Conceptual structures in experience bases and analogical reasoning Banerjee, S. January 1990 (has links) No description available. 003.5 Bionics
7	Conceptual geometric reasoning in artificial intelligence and engineering Brown, Kenneth N. January 1991 (has links) No description available. 003.5 Bionics
8	Conceptual graphs and fuzziness in artificial intelligence Morton, S. K. January 1987 (has links) No description available. 003.5 Bionics
9	The development of a new design concept in modular robot technology Loweth, W. J. January 1984 (has links) No description available. 003.5 Bionics
10	Problem solving with reinforcement learning Rummery, Gavin Adrian January 1995 (has links) No description available. 003.5 Bionics

Search results