Embedded avian recognition and underwater bio-acoustic scene

Marion Poupard (Jan. 2018 – Dec. 2020)


Funding : LPO (bird protection league) and University of Toulon

Supervision : Hervé GLOTIN (LIS), Thierry SORIANO
and Thierry LANGAGNE (CNRS)


H. Glotin’s team at LSIS Laboratory has developed a sensor with a high level of physical integration, high bandwidth and high dynamics with which animal sound recognition studies and some classifications have already been successfully conducted. This thesis first proposes to interrogate data described by ontologies to improve bioacoustic recognition in acoustic landscapes. These ecosystem metadata will be processed and integrated into acoustic recognition by the PhD student on the RDF/Metadata platform dedicated to environmental monitoring. This platform, simple to use, will allow to infer metadata and acoustic indexing data by web engines and resources (RDF data billiards, weather, ecosystem, migration, human activities…).
In a second phase, Marion Poupard’s research work in conjunction with the Cosmer laboratory will focus on the physical and software integration of the specific sensor developed by the LSIS laboratory for the analysis of certain underwater animal populations on an AUV (Autonomous Underwater Vehicle) type submarine robot, and on the definition of environmental monitoring tasks for future measurement campaigns permitted by this new sensor. For the physical and software integration part, we will explore current methods related to system engineering applied to mecatronic systems. There are many works around the MBSE (Model Based System Engineering) to be analysed through a specific state of the art. The AUV software environment is of the Opensource type around the operating system for ROS robotics. After a period of increasing competence on this open system, the doctoral student will propose algorithms and software extensions allowing the control of the hydrophone embedded on the robot. The sensor and its electronics will have to be the subject of a physical design study for a specific enclosure under the constraints of the marine environment, in particular to include the SMIoT JASON system already developed by the DYNI team. For the part concerning the definition and management of tasks related to measurement campaigns, it will be necessary to determine which type of operating mode will be best suited to the needs, between the pure remote operation mode with remote control through the AUV cable link or an autonomous mode which can also be supported by the AUV-embedded architecture based on Raspberry and JASON boards.

There will then be a phase of development of control and measurement parameters first in the basin and then on the marine site. Finally, several series of validation tests will have to follow according to the objectives of the industrial partner BIOSONG, potentially in Aquitaine.