Survey on advances on terrain based navigation for autonomous underwater vehicles

Abstract

The autonomy of robotic underwater vehicles is dependent on the ability to perform long-term and long-range missions without need of human intervention. While current state-of-the-art underwater navigation techniques are able to provide sufficient levels of precision in positioning, they require the use of support vessels or acoustic beacons. This can pose limitations on the size of the survey area, but also on the whole cost of the operations. Terrain Based Navigation is a sensor-based navigation technique that bounds the error growth of deadreckoning using a map with terrain information, provided that there is enough terrain variability. An obvious advantage of Terrain Based Navigation is the fact that no external aiding signals or devices are required. Because of this unique feature, terrain navigation has the potential to dramatically improve the autonomy of Autonomous Underwater Vehicles (AUVs). This paper consists on a comprehensive survey on the recent developments for Terrain Based Navigation methods proposed for AUVs. The survey includes a brief introduction to the original Terrain Based Navigation formulations, as well as a description of the algorithms, and a list of the different implementation alternatives found in the literature. Additionally, and due to the relevance, Bathymetric SLAM techniques will also be discussed.

Inspection of an underwater structure using point‐cloud SLAM with an AUV and a laser scanner

“…To overcome these issues, the vehicle should use maps of its environment and correlate its sensor measurements to the map to discover its position. The most widely used strategy for this purpose is the use of terrain-based navigation methods where a map of the environment exists a priori (Melo and Matos, 2017). Even more challenging and interesting is the use of SLAM methods, where the robot incrementally builds a model of the environment and simultaneously uses it to estimate its position within it.…”

Terrain‐aided navigation for long‐endurance and deep‐rated autonomous underwater vehicles

“…However, these conditions are strong and restrictive since most systems are nonlinear and non‐Gaussian. Although linearisation can be performed to extend the KF applicability to some TAN scenarios, the extended KF (EKF) is unlikely to perform well in areas with steep terrain morphology (Melo & Matos, ).…”

“…Although TAN has drawn significant attention over the last two decades in underwater domain, real‐world implementations are limited in number and typically focus on short‐ to very‐short‐range applications (1–5 km; see Carreno, Wilson, Ridao, and Petillot () and Melo and Matos () for an overview of the current status). In contrast, this study develops a navigation method for extended time periods and deep‐water operations.…”

Impact of terrain factors on the matching performance of terrain‐aided navigation

“…As a good alternative to GNSS (global navigation satellite system) in a GNSS-denied environment, TAN (terrainaided navigation) has been studied for many years. [1][2][3][4][5][6][7][8] Although many efforts have been devoted to the matching algorithms for TAN, the matching performance depends heavily on the variation of a region. [8][9][10] If the region is too flat or similar to others in shape, the matching algorithms are vulnerable to divergence.…”