Fast Visual Odometry for a Low-Cost Underwater Embedded Stereo System †.

Mohamad Nawaf,Mohamed Ben Ellefi,Pierre Drap,Jean-Philip Royer,Jean-Marc Boï,Djamal Merad,Mauro Saccone

doi:10.3390/s18072313

Abstract

This paper provides details of hardware and software conception and realization of a stereo embedded system for underwater imaging. The system provides several functions that facilitate underwater surveys and run smoothly in real-time. A first post-image acquisition module provides direct visual feedback on the quality of the taken images which helps appropriate actions to be taken regarding movement speed and lighting conditions. Our main contribution is a light visual odometry method adapted to the underwater context. The proposed method uses the captured stereo image stream to provide real-time navigation and a site coverage map which is necessary to conduct a complete underwater survey. The visual odometry uses a stochastic pose representation and semi-global optimization approach to handle large sites and provides long-term autonomy, whereas a novel stereo matching approach adapted to underwater imaging and system attached lighting allows fast processing and suitability to low computational resource systems. The system is tested in a real context and shows its robustness and promising future potential.

Highlights

Mobile systems nowadays are undergoing a growing need for self-localization to determine their absolute/relative position over time accurately
We provide a complete hardware and software solution for the problem through the conception and realization of a stereo embedded system dedicated to underwater imaging
As the uncertainty varies, the selection of the closest frames based on the Euclidean distance is not suitable. We prove that it is the case for any visual odometry method

Summary

Introduction

Mobile systems nowadays are undergoing a growing need for self-localization to determine their absolute/relative position over time accurately. Solutions based on active acoustics, such as imaging sonars, water linked GPS or Doppler Velocity Log (DVL) devices remain expensive and require high technical skills for deployment and operation. Their size specifications prevent their integration within small mobile systems or even the ability to be handheld. Sensors 2018, 18, 2313 alternative is ongoing; notably, recent advances in embedded systems have led to relatively small, powerful and cheap devices This opens promising potential to adopt a light visual odometry approach that provides a relative trajectory in real-time using image sensors, and this describes our main research direction. The developed solution is integrated within an underwater archaeological site survey where it plays an important role in facilitating image acquisition

Objectives

Results

Conclusion