ScienceDirect IFAC-PapersOnLine 48-2 (2015) 206–211 Available online at www.sciencedirect.com 2405-8963 © 2015, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved. Peer review under responsibility of International Federation of Automatic Control. 10.1016/j.ifacol.2015.06.034 © 2015, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved. 1. INTRODUCTION Only a fraction of our cultural heritage deposited on the seabed has been located and investigated, with UNESCO (2015) estimating over 3 million undiscovered shipwrecks spread across the planets oceans. Submerged prehistoric landscapes and shipwrecks deposited on the seabed are non-renewable sources to understanding and knowledge of our past, and the scientific community and antiquarian authorities around the world are working against the clock to secure as much of this heritage as possible for posterity. Climate change and increased commercial activities in the marine environments put this cultural heritage under pressure, and accentuates the need for new and efficient methods for recording and managing underwater cultural heritage. Recent years have seen advances within underwater robotics that can remedy this situation by reducing the dependency on divers. Simultaneously, developments in sensor technologies and data processing have lead to rela- tively inexpensive commercially available off-the-shelf dig- ital underwater cameras and photogrammetry software that can be run on a powerful but ordinary PC. The inherent optical properties of the water (light re- flectance, scattering and attenuation) means algorithms intended for underwater computer vision generally need to be more robust than algorithms intended for use in air. While the Scale Invariant Feature Transform (SIFT) is one of the most efficient feature detectors in many applications, ⋆ email: stein.nornes@ntnu.no Meline et al. (2012) demonstrated how suspended particles causes the performance of SIFT to deteriorate severely relative to other approaches. Similarly, Campos et al. (2014) demonstrated how surface reconstruction methods with insufficient levels of noise tolerance can give wildly varying results for underwater datasets. McCarthy and Benjamin (2014) have shown how a diver- based photogrammetric approach can significantly im- prove the efficiency of the process of recording underwater cultural heritage. Taking the diver out of the equation by using a Remotely Operated Vehicle (ROV) as a platform for data capture removes two very limiting operational constraints (depth and bottom time). Removing the pilot as well by having an Autonomous Underwater Vehicle (AUV) perform this task autonomously would represent a significant breakthrough for marine archaeological record- ing. Some photogrammetric AUV surveys have been con- ducted, but are restricted to wrecks and seafloor that provide a relatively flat and benign environment for AUV robotic operations (Foley et al., 2009; Johnson-Roberson et al., 2010; Gracias et al., 2013; Demesticha et al., 2014). Drap et al. (2013) demonstrated an ROV based pho- togrammetric survey of parts of a more threedimensional wreck, relying on a combination of high resolution images and sonar data to construct a georeferenced 3D-model. The reliance on sonar data increases the number of dives required to collect the necessary data. In this paper we will present the process and results from an ROV based photogrammetric recording of a shipwreck Keywords: Unmanned Underwater Vehicles (UUV), Marine archaeology, Mapping, UUV vision systems, Photogrammetry * Centre for Autonomous Marine Operations and Systems (AMOS), Department of Marine Technology, Norwegian University of Science and Technology, NTNU, NO-7491 Trondheim, Norway ** Department of Archaeology and Cultural History, University Museum, Norwegian University of Science and Technology, NTNU, NO-7491 Trondheim, Norway Abstract: This paper presents the results of an underwater photogrammetric survey of an intact standing steel wreck with high vertical profiles from the seabed at 60 meters depth. The survey was conducted in Trondheim Harbour in August 2014 using a Remotely Operated Vehicle (ROV) equipped with a stereo camera rig. The paper demonstrates how the emergence of commercially available photogrammetric software has reduced the required resources for creating high-resolution 3D-models of archaeological sites from photographs. At the same time, the resources and ROV-pilot skills required for the survey itself still represent an obstacle for the end users. The results and experiences of this survey are therefore used as the basis for a discussion on the possible benefits, challenges and strategies for conducting such a survey autonomously. Underwater Photogrammetric Mapping of an Intact Standing Steel Wreck with ROV Stein M. Nornes *,⋆ Martin Ludvigsen * Øyvind Ødeg˚ ard *,** Asgeir J. Sørensen *