Evaluating a 3D node graphing algorithm : Developing an algorithm for improving 3D map data and comparing resulting node graphs used for underground mines

Abstract: Mining companies are rapidly modernizing, and part of this modernization requires the tracking of equipment and personnel within mines. Mobilaris uses 3D maps that represent mine paths using two lines, one representing the left wall and one representing the right. These lines are often discontinuous and mixed with other lines that represent symbols, old lines, erroneous lines, etc. To better track the positions of items within a mine, a node graph that maps the possible cave paths is used. This node graph is partly generated and partly constructed manually. The manual corrections currently require a great deal of time. The purpose of this degree is to determine the viability of the current node graphing algorithm by developing a program that improves the map data. This improvement will be carried out to give the node graphing algorithm the best possible map data to determine where it is usable and where it is not. Multiple quick and problem-specific algorithms are developed to remove lines not part of the two path lines; these do not cover all found cases due to time constraints. An algorithm that merges path lines and an algorithm that connects disconnected path lines are developed. These algorithms greatly improve the quality of the map data. After processing the maps, the improvements to the generated node graphs are small but not insignificant, with a 1% to rare 20% reduction in breaks. The greatest improvements are made by postprocessing the generated node graphs; here, the differences in improvement between the old node graphs and the new are significant, with a 75-80% reduction in breaks. Many of the weaknesses of the current node graphing algorithm are also determined; some of these can be corrected by postprocessing the node graph, while some need either new or additional algorithms to be solved.