Underwater Mapping and Navigation: Applications of 3D Feature Extraction Algorithms to 3D Sonar Datasets

J.N. Markiel, Dorota Grejner-Brzezinska, Charles Toth, William Woodward, Jim Moore

Abstract:	Underwater environments present exceptional challenges for precision navigation; in addition to the engineering challenges required for the protection of electronics and systems in harsh conditions, the basic problem of determining ongoing position faces considerably greater challenges as compared to land based navigation. Bathymetric maps are generally limited in spatial resolution and subject to rapid (and undetected) changes on a temporal basis. Under the ocean surface, independent sources to establish a position fix (including GPS positioning) are typically unavailable, and navigation via an inertial navigation system (INS) is constrained by the general lack of sufficiently dense network data for gravity anomalies. Recent developments in sonar technology are expanding the ability to generate real-time maps in full 3D, enabling a complete reconstruction of the underwater environment within the acquisition area. The technology enables capture of both static and non-static features underwater; utilized in tandem with a GPS and INS on-board a hydrographic survey vessel, the features in the 3D environment can be mapped and geo-referenced. The necessity of a GPS signal lock means that geo-referencing can be performed if the GPS position is available. Georeferencing from a totally submerged system is therefore limited, if not impossible, unless known reference points are made available to the system on either a-priori or onthe- fly basis. Algorithmic developments in the separation of static and non-static features have been successfully demonstrated by the authors to enable indoor navigation utilizing 3D Flash LADAR (Laser Distance and Ranging) datasets. The algorithm does not require a-priori assumptions regarding the sensor environment beyond the need for features to be present. Extending these same techniques to 3D sonar data can enable the system to navigate from static features underwater; the challenge is to identify common static features from a time series of sonar “images” and to utilize tri-lateration to establish ongoing positioning. These position updates are then utilized in lieu of GPS within the schema of a traditional GPS/INS system allowing for the drift in the inertial system to be corrected in the Extended Kalman Filter solution. Utilizing actual 3D sonar datasets, we demonstrate the results of our studies related to the separation of static and non-static features in an underwater environment for the purposes of mapping and navigation. In the instance of mapping, the system initiates from a known location via GPS; and then utilizes the known position of static features to enable geo-referencing of new features despite loss of GPS availability. This extends the reach of the acquisition platform beyond the surface vessel. Additionally, we analyze the potential for underwater navigation based upon the synergy of the 3D sonar capabilities and the 3D feature-based navigation algorithm.
Published in:	Proceedings of the 2010 International Technical Meeting of The Institute of Navigation January 25 - 27, 2010 Catamaran Resort Hotel San Diego, CA
Pages:	448 - 458
Cite this article:	Markiel, J.N., Grejner-Brzezinska, Dorota, Toth, Charles, Woodward, William, Moore, Jim, "Underwater Mapping and Navigation: Applications of 3D Feature Extraction Algorithms to 3D Sonar Datasets," Proceedings of the 2010 International Technical Meeting of The Institute of Navigation, San Diego, CA, January 2010, pp. 448-458.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In