Real-time Monocular Visual Odometry for Turbid and Dynamic Underwater Environments

In the context of robotic underwater operations, the visual degradations by the medium properties make difficult the exclusive use of cameras localization purpose. Hence, most localization methods are based on navigational sensors associated with acoustic positioning. On the hand, visual odometry and visual SLAM have been exhaustively studied for or terrestrial applications, but state-of-the-art algorithms fail. In this paper we tackle the problem of using a simple low-cost for underwater localization and propose a new monocular visual odometry dedicated to the underwater environment. We evaluate different tracking and show that optical flow based tracking is more suited to underwater than classical approaches based on descriptors. We also propose a-based visual odometry approach highly relying on nonlinear. The proposed algorithm has been assessed on both simulated and underwater datasets and outperforms state-of-the-art visual SLAM methods many of the most challenging conditions. The main application of this is the localization of Remotely Operated Vehicles (ROVs) used for archaeological missions but the developed system can be used in any applications as long as visual information is available.