Underwater place recognition using forward‐looking sonar images: A topological approach

Abstract

Autonomous underwater vehicles are a prominent tool for underwater exploration because they can access dangerous places avoiding the risks for the human beings.However, the autonomous navigation still a challenge due to the characteristics of the environment that decrease the performance of the sensor and the robot perception.In this context, this paper proposes a loop closure detector addressed to the simultaneous localization and mapping problem at semistructured environments using acoustic images acquired by forward-looking sonars. The images are segmented by an adaptative approach based on the acoustic beams analysis. A pose-invariant topological graph is build to represent the relationship between image features. The loop closure detection is achieved using a graph comparison. The approach is evaluated in a real environment at a marina. The results reveal all loop closures of the data set are detected with a high precision and present an invariant to image rotation.

Cross-view and Cross-domain Underwater Localization based on Optical Aerial and Acoustic Underwater Images

“…(FLS) BlueView P900 [31]. The vehicle was attached below a floating board such as it stays underwater while a Differential Global Position System DGPS stays in the top of the board outside water as shown in Fig.…”

Underwater navigation with 2D forward looking SONAR: An adaptive unscented Kalman filter‐based strategy for AUVs

“…When applied to real FLS images acquired during an underwater mission, the coupling between rotations and translations, and erroneous peak detection in Equation (19), must be handled. The latter is due to the noise in FLS images that may come from (to name a few) low signal‐to‐noise ratio (SNR), low resolution, inhomogeneous insonification, and image alterations related to FLS's viewpoint modifications (see Hurtós et al, 2015 and Santos et al, 2019).…”

A Survey of Underwater Acoustic SLAM System

“…Compared to optical images, acoustic images suffer problems to their construction and the physical effects of the acoustic waves such as: inhomogeneous resolution, acoustic distortion, loss of three-dimensional information, nonuniform pixel intensities, acoustic reverberation, acoustic shadow, and noise increasing the difficulty of registration [40]. Due to those inherent characteristics of sonar data, pixel-level features, for example Harris corner [20,31], extracted in sonar images suffer from low repeatability rates and instability.…”

“…For instance, in a partially structured environment, like harbor area, composed of piers, boats and ships, the topological relationship between them should remain constant and can be adopted as a landmark which can be identified on the next robot visit. Santos et al [40] proposed to describe the sonar image as a topological graph by establishing the relationships between each Gaussian function which was used to fit the image segments. This description is translation and rotation invariant and graph comparison can be carried out to detect loop closures.…”