SURF3DSLAM

Probabilistic 3D surface matching for bathymetry based Simultaneous Localization and Mapping of underwater vehicles

 Partecipanti

Mappa

 Word cloud

Esplora la "nuvola delle parole (Word Cloud) per avere un'idea di massima del progetto.

 Obiettivo del progetto (Objective)

'The development of Autonomous Underwater Vehicles (AUVs) has offered numerous advantages, but has also presented new challenges. Free from the burden of an umbilical cable connecting them to the support ship, they are a quantum leap from widely used Remotely Operated Vehicles (ROVs) providing a stable platform for the new high resolution scientific sensors and reducing the operational costs. One of the most significant challenges is the problem of underwater navigation or, in other words, how to determine the vehicle's position within the environment, at least as accurate as needed, according to the standards of each scientific survey. Our research objective is to accurate localize the AUV in the three dimensional environment (3D) using its own sensors without any external infrastructure under the concept of Simultaneous Localization and Mapping (SLAM). The environment will be sensed by state of the art multibeam echosounders that they are becoming broadly available, providing high precision of the surrounding bathymetry and they are not disturbed by turbidity or lighting conditions. Moreover, attitude sensors that are commonly found in AUVs will aid on the results precision. Under probabilistic 3D raw data registration and SLAM framework, most of the AUVs navigation constraints will be removed and will allow safe and precise surveys in the challenging underwater environment.'

Introduzione (Teaser)

Autonomous underwater vehicles (AUVs) are useful but difficult to localise underwater. New EU-developed methods help to estimate AUVs trajectories, using a special multi-beam sonar and image-compositing algorithms to monitor the vehicles' 3D surroundings.

Descrizione progetto (Article)

AUVs are small robotic submarines not connected by cable to other vehicles or ships. They have many scientific applications, especially as stable sensor platforms. However, their use can be challenging because of the difficulty in determining their position.

The EU funded SURF3DSLAM, a two-year project aimed at developing a technique for locating AUVs in three dimensions using their own sensors. The concept, called Simultaneous Localisation and Mapping (SLAM), uses precise multi-beam echosounders to sense the vehicle's depth and surroundings. The system is unaffected by visibility or lighting conditions. The objective was to adapt to three dimensions a 2D technique for probabilistic, sonar-based, range-image registration. SURF3DSLAM concluded in March 2013.

Project members described a method for pose-based SLAM, using probabilistic surface matching and a multi-beam sonar profiler. The algorithm combines scanned strips of the seafloor, and uses position calculations. The method was successfully demonstrated using real data from a submarine survey. The project also investigated a technique for surface adaptation.

Published results total 12 peer-reviewed publications, 1 book chapter and 2 posters presented at international conferences.

The multidisciplinary work of SURF3DSLAM significantly advanced both robotics and oceanographic research.