| dc.description.abstract | The availability of minerals is essential for the production of goods and the development of sustainable technologies. The mineral demand is expected to increase in the coming years as a result of population growth and technological needs. Marine minerals are resources that may supplement traditional land-based sources to meet this new demand. Submerged massive sulphides (SMS) are one of the potential sources of marine minerals. Such deposits are continuously being formed at active hydrothermal vents along mid-ocean ridges at depths between 800m to 5000m. Samples have revealed these sites to be rich in copper (Cu), zinc (Zn), lead (Pb), together with smaller quantities of gold (Au), silver (Ag), and other trace minerals. Currently, there are no ongoing large-scale mining operations related to these deposits. Before excavation and extraction of these minerals can be even considered, the return of investment must reflect the financial risks involved. The economic value is difficult to estimate due to unknown metal tonnages and grades for discovered and undiscovered deposits alike. The number of deposits one can expect to find is also uncertain. The return on investment for a marine mineral venture may depend on availability of multiple deposits of economic character. Locating these deposits is time consuming and expensive, due to the great depths at which they reside and the large spatial scales involved. For this reason, the development of efficient exploration methodologies is cruicial for future deep sea mining endeavours.
Conventional large-scale exploration approaches are focused on either detection of fluid discharge from ongoing hydrothermal activity or detailed bathymetric and geophysical data (e.g. magnetic and electromagnetic). The former can be performed directly from a ship, using towed instrumentation, but is limited to the detection of active vent sites. The search is also imprecise, as adjacent vent sites can be difficult to discriminate and the plume direction can fluctuate in the ocean currents. The latter is not limited to detecting a plume, but can directly measure properties of the seabed and subsurface. Autonomous underwater vehicles (AUVs) can be used to collect these data efficiently. When used in conventional survey activities, these vehicles are programmed ahead of time to cover an area in full at a desired altitude to collect data at a target spatial resolution. Transmission of data over an acoustic link is constrained, and the interaction possible by the operators in response to the collected data in-situ is limited. The combination of automated data analysis, such as machine learning techniques and applied statistics, with advanced onboard planning can increase the autonomy of the underwater vehicle by making strategic choices in response to collected data. This can enable more efficient search methodologies by reducing the need for costly re-deployment of resources, with selective coverage or resolution depending on the favourability of the area.
This thesis outlines an exploration methodology based around autonomous vehicles, where minimal interaction from operators is necessary. The spatial distribution of the deposits may be significant, and covering the entirety of the target areas in full detail is prohibitive. For this reason, a hierarchical search structure is proposed, where target areas are investigated incrementally in greater detail. The areas of interest are typically located at great depths, and the deployment and recovery of an autonomous underwater vehicle or remotely operated vehicle leads to significant time delays due to the dive and ascent. A key aspect of the proposed strategy is that data processing and decision-making is performed on the vehicle itself. The prerequisites for a fully autonomous search is divided into four categories; (1) supporting systems required for aiding and monitoring an autonomous operation, (2) a high-altitude regional survey collecting initial data, (3) high-resolution geophysical survey of targets determined from the previously collected data, (4) closeproximity survey with optical instruments for target verification. The thesis is presented as a collection of publications, where each addresses aspects under one of these categories.
A path planning algorithm for an autonomous surface vessel (ASV) tracking one or more autonomous underwater vehicles simultaneously through acoustic measurements (category 1) has been developed. This is an important supporting role during autonomous surveying, as it can provide position updates for the onboard navigation system and act as an acoustic communication gateway, without requiring the presence of a crewed vessel. The application of underwater hyperspectral imaging for mineral exploration has been investigated, which can be used to discriminate between different materials on the seafloor (category 4). Autonomous underwater vehicles built for covering larger areas are generally not capable of collecting samples or interacting with the seabed physically. Underwater hyperspectral imaging can mitigate the need for collecting samples and thus reduce the number of time-consuming ROV deployments. An algorithm for co-registration of adjacent near-seabed transects based on multibeam echosounder data is presented, which can enable spatial interpretation of data across transects subjected to navigational drift (categories 3 and 4). Inertial sensors and Doppler velocity logs have inherent uncertainties and biases which causes navigational drift over time. Although these errors can be bounded by the use of acoustic positioning, the relative error between adjacent transects can nonetheless be large compared to the spatial resolution of the sensor (e.g. a camera). The presented algorithm utilizes the acoustic data, which has a wider swath than optical instruments, in order to improve the relative positioning of the transects. Contributions have also been presented with respect to in-situ autonomy based on backscatter measurements and automatic segmentation of corals in sonar data (category 2). While these contributions are not presented in the context of mineral exploration, the methodologies developed are transferrable. | en_US |
