A Probabilistic 3D Map Representation for Forward-Looking SONAR Reconstructions
Matteo Franchi, Alessandro Bucci, Leonardo Zacchini, Edoardo Topini, Alessandro Ridolfi, Benedetto Allotta
Abstract
The ability of acoustic waves to penetrate through waters in all visibility conditions and for long ranges has made SOund NAvigation Ranging (SONAR) devices increasingly widespread for underwater sensing. In particular, imaging sonars, such as Forward-Looking SONARs (FLSs), have become ubiquitous nowadays both for Autonomous Underwater Vehicles (AUVs) and Remotely Operated Vehicles (ROVs). However, the loss of elevation angle information during the 3D to 2D image projection lessens the ability to understand the underwater environment. This paper presents a probabilistic 3D occupancy mapping framework that employs FLS images, and it is specifically tailored to AUVs. The proposed method is tested through the use of data recorded during sea trials performed in 2018 with FeelHippo AUV, a vehicle developed by the Department of Industrial Engineering of the University of Florence (UNIFI DIEF), at the NATO Science and Technology Organization Centre for Maritime Research and Experimentation (CMRE), La Spezia (Italy).