Navigation and real-time sensor fusion for automated observation and data assimilation

This thesis deals with the navigation of autonomous underwater vehicles (AUVs) based on real-time sensor measurements using automatic control methods. The use of AUVs for data collection has enormous potential to reduce costs and improve data quality. AUVs are cheaper to operate than towed or remotely operated systems, especially since the vessel can perform other measurements during this time. In science, AUVs are typically used to acquire data by navigating along so-called lawnmower patterns but the use of AUVs on completely pre-programmed trajectories is suboptimal. Since the spatial boundaries of the phenomenon to be observed cannot be precisely determined a priori, (near) real-time adaptation of the navigation is essential to achieve high data quality comparable to that of towed or remotely operated systems. In this thesis, three application examples of varying complexity are considered, namely the detection and tracking of a boundary layer (thermocline), the localization of a source using the example of a hydrothermal vent, and the estimation of the concentration of a sediment plume as it occurs in deep-sea mining.
In the first two application examples an approach based on extremum seeking control (ESC) is used. This is an optimization method that does not require an explicit model, which can be particularly advantageous in the marine context.
In the third use case, Bayesian optimization (BO) is used to estimate the concentration of a sediment plume. This optimization method is used to solve difficult-to-evaluate functions, also known as blackbox functions.

Rights

Use and reproduction:


CC BY-NC-ND 4.0

Please note that individual components of the publication may be subject to other licensing or copyright conditions.

Cite

Citation style:
Could not load citation form.