AbstractHigh-precision navigation is vital to the production of all Synthetic Aperture Sonar (SAS) data products. The required sub-wavelength precision is typically achieved using the Redundant Phase Centre (RPC) micro-navigation algorithm, which exploits time delays measured between redundant signals from adjacent pings to estimate the motion of the vehicle. Today's SAS-equipped Autonomous Underwater Vehicles (AUVs) typically feature two arrays on one or both sides of the vehicle. These interferometric SAS systems are able to make high-resolution estimates of sea-floor bathymetry in addition to the standard SAS imagery. The addition of interferometric arrays also dramatically increases the number of redundant signal pairs between pings that contain useful information.
This thesis describes novel techniques which enable this information to be exploited in order to simultaneously make high-precision navigation and coarse sea-floor bathymetry estimates. Central to the new method is the concept of the triad of confounding factors; that is, time delays measured between redundant signals are affected by three quantities: 1) the motion of the vehicle, 2) the sea-floor bathymetry, and 3) the speed of sound. Consequently, inversion of a model of sufficient accuracy allows high-precision estimation of these three quantities. %In pursuit of higher precision navigation estimation, the models for propagation time used in this work do not exploit the phase centre approximation or stop-and-hop approximation.
Results from a novel algorithm that simultaneously estimates the vehicle path to sub-wavelength precision and makes a coarse bathymetry estimate are presented, using both simulated and experimental SAS data. The method yields improved SAS image quality when compared to conventional slant-plane RPC micro-navigation, which is demonstrated using experimental data collected by the 270-330 kHz SAS of the CMRE MUSCLE AUV. Since ground-truth is not available using experimental data, the precision of the navigation and coarse bathymetry estimates is demonstrated using simulated data.
The new algorithm has the potential to improve the navigation precision of SAS-equipped AUVs, improving SAS image geo-referencing precision and allowing AUVs to perform longer duration surveys without surfacing. The method is designed to be easily generalisable to repeat-pass SAS, and as such it has the potential to provide the sub-wavelength track co-registration that is required for repeat-pass interferometric bathymetry estimation.
|Date of Award||1 Apr 2020|
|Sponsors||James Dyson Foundation|
|Supervisor||Nigel Johnston (Supervisor), Philippe Blondel (Supervisor) & Alan Hunter (Supervisor)|
- Synthetic Aperture Sonar
- Autonomous Underwater Vehicles
- Acoustic Imaging