This paper presents a new tracking system for autonomous underwater vehicles (AUVs) navigating in a close formation, based on computer vision and the use of active light markers. While acoustic localization can be very effective from medium to long distances, it is not so advantageous in short distances when the safety of the vehicles requires higher accuracy and update rates. The proposed system allows the estimation of the pose of a target vehicle at short ranges, with high accuracy and execution speed. To extend the field of view, an omnidirectional camera is used. This camera provides a full coverage of the lower hemisphere and enables the concurrent tracking of multiple vehicles in different positions. The system was evaluated in real sea conditions by tracking vehicles in mapping missions, where it demonstrated robust operation during extended periods of time.

This paper presents the development of an underwater omnidirectional multi-camera system (OMS) based on a commercially available six-camera system, originally designed for land applications. A full calibration method is presented for the estimation of both the intrinsic and extrinsic parameters, which is able to cope with wide-angle lenses and non-overlapping cameras simultaneously. This method is valid for any OMS in both land or water applications. For underwater use, a customized housing is required, which often leads to strong image distortion due to refraction among the different media. This phenomena makes the basic pinhole camera model invalid for underwater cameras, especially when using wide-angle lenses, and requires the explicit modeling of the individual optical rays. To address this problem, a ray tracing approach has been adopted to create a field-of-view (FOV) simulator for underwater cameras. The simulator allows for the testing of different housing geometries and optics for the cameras to ensure a complete hemisphere coverage in underwater operation. This paper describes the design and testing of a compact custom housing for a commercial off-the-shelf OMS camera (Ladybug 3) and presents the first results of its use. A proposed three-stage calibration process allows for the estimation of all of the relevant camera parameters. Experimental results are presented, which illustrate the performance of the calibration method and validate the approach.

Omnidirectional vision has recently captured plenty of attention within the computer vision community. The popularity of cameras able to capture 360 $^{\circ }$ has increased in the last few years. A significant number of these cameras are composed of multiple individual cameras that capture images or videos, which are stitched together at a later postprocess stage. Stitching strategies have the complex objective of seamlessly joining the images, so that the viewer has the feeling the panorama was captured from a single location. Conventional approaches either assume that the world is a simple sphere around the camera, which leads to visible misalignments on the final panoramas, or use feature-based stitching techniques that do not exploit the rigidity of multicamera systems. In this paper, we propose a new stitching pipeline based on state-of-the-art techniques for both online and offline applications. The goal is to stitch the images taking profit of the available information on the multicamera system and the environment. Exploiting the spatial information of the scene helps to achieve significantly better results. While for the online case, sparse data can be obtained from a simultaneous localization and mapping process, for the offline case, it is estimated from a 3-D reconstruction of the scene. The information available is represented in depth maps, which provide all information in a condensed form and allow easy representation of complex shapes. The new pipelines proposed for both online and offline cases are compared, visually and numerically, against conventional approaches, using a real data set. The data set was collected in a challenging underwater scene with a custom-designed multicamera system. The results obtained surpass those of conventional approaches.

Persistent deployment of underwater assets holds the key to achieve consistent, long-term undersea monitoring. To advance in that direction, the LOON-DOCK project aims to demonstrate remote AUV operation and survey data transmission through the Internet using an underwater docking station. We present a funnel-shaped docking station equipped with a contactless high bandwidth link and the necessary equipment to enable autonomous homing and docking by combining acoustic and optical sensing. The proposed combination, using a range-only localization at far distances and a light beacon localization at short ranges ensures a reliable strategy enabled with low-cost equipment and minimal requirements on both the vehicle and the dock sides. Moreover, remote operation is demonstrated by integrating the system in a unified web interface to control underwater assets, thus allowing a user to operate the AUV remotely through Internet. The full system has been successfully validated with tests conducted in a harbor environment.