Seminar: Planning an interesting path
Abdullah Ali Faruq
Supervisor: Dr. Antonina Kolokolova
Planning an interesting path
Department of Computer Science
Monday, June 15, 2015, 2:00 pm, Room EN 2022
Localization when under water is a crucial problem in glider navigation, as there is no GPS signal and ded reckoning is far from precise. Recently, Ralf Bachmayer and Brian Claus from MUN Autonomous Ocean Systems Laboratory have been incorporating terrain-based navigation capabilities into their gliders, using single-beam sonar readings to estimate glider position. This approach works well if the ocean bottom under the glider is uneven,rich in unique features, rather than flat and uniform. This poses the question of designing a path for the glider so as to optimize the localization opportunities and minimize the positioning error.
Here, we present a dynamic programming algorithm addressing this problem. The main part of the algorithm consists of preprocessing the terrain map to compute pairwise shortest paths within an error margin; this can be done once, given the error parameters, a terrain map, and ocean current map. The second part of the algorithm is computing the actual "interesting" path between two locations, within a given error threshold.