Generation of Augmented Bathymetry to Aid Development of Terrain-Aided Autonomous Underwater Vehicle Localization and Navigation Approaches

Abstract

Perlin noise is used in terrain generation when the available terrain is not sufficiently detailed (for example, asteroid surface modelling for navigation algorithms). Perlin noise is pseudo-random coherent noise. The noise function is smooth up to and including its second order derivative. The frequency of the Perlin noise function can be adjusted. Consequently, successive noise functions can be generated with differing frequency content (usually differing by an octave) and amplitudes. These noise functions can be superimposed to create variable terrain.The research and development reported builds on previous work towards extended range autonomous underwater vehicle (AUV) localization and navigation aided by gravity anomalies and bathymetry. As part of the previous work, a high-fidelity AUV navigation testbed (ANT) which uses the open-source robotics middleware ROS, and interfacing with an open-source AUV vehicle simulator and physics-based engine, was developed. With the earlier contributions there was no way to validate the constant density assumption when using a priori bathymetry derived gravity anomaly maps for localization. Secondly, there was no realization or exploration of the impact of spatial details in the bathymetry, therefore, the derived gravity anomaly field, on the AUV localization performance.As part of this work the functionality of the ANT has been improved to include generating synthetic terrain and augmenting existing bathymetry. The main contribution of this paper is documenting the augmentation of existing lower-resolution bathymetric measurements with terrain variations using Perlin noise. The augmented terrain maintains the integrity of the lower-resolution bathymetric measurements while allowing terrain variation between measurements that are realistic. The level of detail and amplitude of variations is completely subject to design.