CFM 2019

Design and control of a low-cost autonomous profiling float
Thomas Le Mézo  1, *@  , Gilles Le Maillot  1@  , Thierry Ropert  2@  , Luc Jaulin  1@  , Aurelien Ponte  3@  , Benoît Zerr  1@  
1 : Lab-STICC  (Lab-STICC)
ENSTA Bretagne
2 : Institut de Recherche Dupuy de Lôme  (IRDL)
ENSTA Bretagne
3 : Laboratoire de physique des océans  (LPO)  -  Site web
Université de Bretagne Occidentale (UBO), INSU, Institut Français de Recherche pour l'Exploitation de la Mer (IFREMER), CNRS : UMR6523, Institut de recherche pour le développement [IRD], Institut Universitaire Européen de la Mer (IUEM)
Z.I. Pointe du Diable B.P. 70 29280 Plouzané -  France
* : Auteur correspondant

Profiling floats which are autonomous underwater vehicle (AUV) that can only regulate there depth are widely used in oceanography. They are equipped with instruments such as temperature, pressure or conductivity sensors. By carrying out profiles of the ocean up to 6000 meters, they help to a better understanding of the ocean and provide crucial data for models through several years missions.
Designing low-cost profiling floats for shallow depth is a key challenge to be able to use them in swarms. Indeed, increasing the data density in certain areas could improve the knowledge and understanding of small scale oceanographical dynamics. Moreover, implementing an efficient regulation law that minimize the energy consumption while maintaining a low error relative to the depth set point is challenging in the context of low-cost actuators and sensors.
In this talk, we will present the SeaBot robot and the first trials at sea carried out in Brest. We will focus on the technological choices that have contributed to reduce the global cost of the float. Several iterative development cycles have been necessary to achieve a proper integration of the electronic and mechanical parts. The choice of the command law based on a full state feedback coupled with an extended kalman filter (EKF) will be presented. This allow to reduce drastically the time dedicated to calibration than with classical PID based controllers, and improve the robustness of the system.

Personnes connectées : 62