Underwater Localization in Complex Environments

• Autor(a) principal: Maria Sara Delgadinho Noronha
• Data de Publicação: 2021
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: https://hdl.handle.net/10216/135435
• topic: Engenharia electrotécnica, electrónica e informática; Electrical engineering, Electronic engineering, Information engineering

The ability of an autonomous underwater vehicle (AUV) to locate itself in a complex environment as well as to detect relevant environmental features is of crucial importance for successful navigation. However, it's particularly challenging in underwater environments due to the rapid attenuation suffered by signals from global positioning systems or other radio frequency signals, dispersion and reflection thus needing a filtering process. Complex environment is defined here as a scenario with objects detached from the walls, for example the object can have a certain orientation variability therefore its position is not always known. Examples of scenarios can be a harbour, a tank or even a dam reservoir, where there are walls and within those walls an AUV may have the need to localize itself according to the other vehicles in the area and position itself relative to one to observe, analyse or scan it. Autonomous vehicles employ many different types of sensors for localization and perceiving their environments and they depend on the on-board computers to perform autonomous driving tasks. For this dissertation there is a concrete problem to solve, which is to locate a suspended cable in a water column in a known region in the sea and navigate according to it. Although the cable position in the world is well known, the cable dynamics does not allow knowing where it is exactly. So, in order to the vehicle localize itself according to it so it can be inspected, the localization has to be based on optical and acoustic sensors. This study explores the processing and analysis of optical and acoustic images, through the data acquired through a camera and by a mechanical scanning sonar (MSIS), respectively, in order to extract relevant environmental characteristics that allow the estimation of the location of the vehicle. The points of interest extracted from each of the sensors are used to feed a position estimator, by implementing an Extended Kalman Filter (EKF), in order to estimate the position of the cable and through the feedback of the filter improve the extraction processes of points of interest used.