Relative location of autonomous underwater vehicles using visual markers

Abstract

The use of autonomous underwater vehicles (AUV) has been increasing in recent years to carry out tasks to replace human intervention, namely extracting information from the rivers and oceans, including the collection of biodiversity data in the environment ormapping them. Its activities have become increasingly autonomous and the underwatermissions are being longer, in some cases. When the missions are long and considering that the AUVs have processing and storage limitations, they need to pass on the collected information in order to continue operating. Within the scope of the \emph{GROW} project developed by \emph{INESC TEC}, there is a need that an AUV mule tracks an AUV on a mission to exchange data. To do it, AUVs must be communicating close to each other, so that data exchange is efficient and as quickly as possible. Thus, the objective of this dissertation is to make the detection of the AUV on a mission to support a tracking application. The detection of the AUV is performed by an optical camera that intends to detect artificial markers that are attached to the AUV body to be detected. Aruco markers are used and also, the aruco library provided by \emph{OpenCv}. In this way, the detection quality of the aruco markers will be characterised. With this, decide and build a marker configuration that satisfies detection parameters, to be placed in the AUV body, in order to increase the number of angles at which the AUV can be detected. After analysing the marker configurations created, it was understood that the detection capacity on the part of the aruco library is weak from a certain pitch angle of inclination of the AUV. In this way, a method was proposed to improve this detection and extract more information about the AUV in question. Finally, the proposed marker configurations were tested in a water tank with controlled conditions.