Abstract
Geolocation of wild species is a critical source of information for ecological studies and for managers to make decisions related to the habitat and distribution of these species. Several devices (e.g. electronic tags) have been developed to track animals. It is however very difficult to have accurate trajectories at fine temporal and spatial scale for marine species. Indeed, sea water and wild animals complicate the acquisition of accurate positions, especially for long distance.The subject of the thesis is the development of a solution for underwater trajectory estimation and transmission, which can be embedded within a bio-telemeter. It uses the fusion of data from different sensors. These sources can be particularly heterogeneous (inertial data, depth and velocity estimation, geolocated position).To estimate trajectories and accuracy of this estimation, reference data are collected using platforms developed over this PhD, i.e. an Autonomous Surface Vehicle (ASV) with underwater acoustic geolocation and a differential GPS. First, we test different algorithms and present the trade-off of these different solutions in terms of accuracy, consumption, and portability. The trajectory estimation techniques developed and embedded are then improved through the inclusion of the animal behavior to speed estimation. This has another advantage, which is to complete the trajectory information with an ethogram calculated in-board. A final step, the developed algorithms are embedded within a bio-telemeter. Several tests are performed in a controlled environment for different contexts of analysis. The proposed solutions allow novice or experienced users in programming to use the tools for different degrees of analysis. Finally, post-processing of the transmitted data is used to show analysis potential and possible improvements that could be performed. The biological model used for this PhD is the sea turtles.