Key information

Title: Systems for Optimal Offshore Mineral Mining
Application number: 20250085
Case type: National
Status: 23.01.2025 Andre og senere realitetsskriv foreligger
Filed: 23.01.2025
Priority: 06.05.2024 NO 20240441
Effective date: 23.01.2025
Publicly available: 07.11.2025
Next annual fee due: 31.01.2027
Applicant: Modi Vivendi AS (NO)
Inventor: Gunnar Myhr (NO)

Abstract and drawing

Disclaimer: This text has been machine-scanned and may contain errors – please refer to "Publications" for legally binding content.

The objective technical problem to be solved by the invention, is the development of intelligent algorithms and systems (vehicles) to locate and extract economic sound concentrations of any combinations of nodules, manganese crusts and/or sulphide deposit, and separate uneconomic matter from valuable minerals. Emphases are made on applying differences in electric and/or acoustic properties to differentiate economically valuable minerals from cost bearing unprofitable other matter (e.g. mud, gravel, rocks, organic matter), thus providing added profitability compared to existing mining machines. Both independent systems and add-on vehicles to existing mining machines have been developed. Any environmental impacts are minimized by the nature of the invented technical solutions. MS and/or AI methods and algorithms can be incorporated.

Publications

Latest published version

Documents

Expand

Download files (39)

Date

Date	Doc. No.	Process	Case number	In/Out	Journal description	To/from
04.10.2026	18-04	Saksbehandling	20250085	INN	Track Changed Corrected Claims NO20250085	MODI VIVENDI AS
04.10.2026	18-03	Saksbehandling	20250085	INN	Reply_to_OA_30_08_26_NO20250085 - Oct 4 2026	MODI VIVENDI AS
04.10.2026	18-02	Saksbehandling	20250085	INN	Clean Corrected Claims NO20250085	MODI VIVENDI AS
04.10.2026	18-01	Saksbehandling	20250085	INN	Korrespondanse (Hovedbrev inn)	MODI VIVENDI AS
30.08.2026	17-01	Saksbehandling	20250085	UT	Substantive examination	Modi Vivendi AS
13.05.2026	15-01	Saksbehandling	20250085	UT	Generelt brev uten frist ENG	Modi Vivendi AS
08.05.2026	16-02	Saksbehandling	20250085	INN	Tilsvar OA 22_08_2025 - NO20250085 - Dato 26_08_2025	Modi Vivendi AS
08.05.2026	16-01	Saksbehandling	20250085	INN	Korrespondanse (Hovedbrev inn)	Modi Vivendi AS
15.10.2025	14-02	Saksbehandling	20250085	INN	REV 15_10_2025 - NO20250085 - SYSTEMS FOR OPTIMAL OPTIMAL OFFSHORE MINERAL MINING	Modi Vivendi AS
15.10.2025	14-01	Saksbehandling	20250085	INN	Korrespondanse (Hovedbrev inn)	Modi Vivendi AS
15.10.2025	13-02	Saksbehandling	20250085	INN	REV 15_10_2025 - NO20250085 - SYSTEMS FOR OPTIMAL OPTIMAL OFFSHORE MINERAL MINING	Gunnar Myhr
15.10.2025	13-01	Saksbehandling	20250085	INN	Korrespondanse (Hovedbrev inn)	Gunnar Myhr
15.10.2025	12-01	Saksbehandling	20250085	UT	Generelt Brev Patent	Modi Vivendi AS
17.09.2025	11-03	Saksbehandling	20250085	INN	Tilsvar OA 22_08_2025 - NO20250085 - Dato 26_08_2025	Modi Vivendi AS
17.09.2025	11-02	Saksbehandling	20250085	INN	01-Substantive examination (29) - NO20250085	Modi Vivendi AS
17.09.2025	11-01	Saksbehandling	20250085	INN	Korrespondanse (Hovedbrev inn)	Modi Vivendi AS
26.08.2025	10-02	Saksbehandling	20250085	INN	Tilsvar OA 22_08_2025 - NO20250085 - Dato 26_08_2025	Gunnar Myhr
26.08.2025	10-01	Saksbehandling	20250085	INN	Korrespondanse (Hovedbrev inn)	Gunnar Myhr
22.08.2025	09-01	Saksbehandling	20250085	UT	Substantive examination	Modi Vivendi AS
04.05.2025	08-03	Saksbehandling	20250085	INN	Corrected Claims with Annotations - NO20250085 - May 4 2025 pdf	Gunnar Myhr
04.05.2025	08-02	Saksbehandling	20250085	INN	Corrected Claims - Clean - NO20250085 - May 4 2025 pdf	Gunnar Myhr
04.05.2025	08-01	Saksbehandling	20250085	INN	Korrespondanse (Hovedbrev inn)	Gunnar Myhr
16.04.2025	07-01	Saksbehandling	20250085	INN	SV ANMODNING OM MØTE MED PATENTSTYRET 24. APRIL EVT. 25. APRIL 2025
03.04.2025	06-04	Saksbehandling	20250085	INN	Revised Claims as of April 3 2025 with annotations - NO20250085	Modi Vivendi AS
03.04.2025	06-03	Saksbehandling	20250085	INN	Revised Claims as of April 3 2025 - Clean - NO20250085	Modi Vivendi AS
03.04.2025	06-02	Saksbehandling	20250085	INN	Reply to Office Action as of 31_03_2025 - NO20250085 - 03_04_2025	Modi Vivendi AS
03.04.2025	06-01	Saksbehandling	20250085	INN	Korrespondanse (Hovedbrev inn)	Modi Vivendi AS
31.03.2025	05-02	Saksbehandling	20250085	UT	PT report 02:01:28
31.03.2025	05-01	Saksbehandling	20250085	UT	Substantive Examination	Modi Vivendi AS
01.02.2025	04-03	Saksbehandling	20250085	INN	NO20250085 - New Claims 1_13_14 with annotations	Gunnar Myhr
01.02.2025	04-02	Saksbehandling	20250085	INN	NO20250085 - New Claims 1_13_14 Clean	Gunnar Myhr
01.02.2025	04-01	Saksbehandling	20250085	INN	Korrespondanse (Hovedbrev inn)	Gunnar Myhr
27.01.2025	02-01	Saksbehandling	20250085	UT	Søknadskvittering Patent	Modi Vivendi AS
23.01.2025	03-01	Saksbehandling	20250085	INN	Korrespondanse (Hovedbrev inn)	Gunnar Myhr
23.01.2025	01-05	Saksbehandling	20250085	INN	Drawings	Modi Vivendi AS
23.01.2025	01-04	Saksbehandling	20250085	INN	Claims	Modi Vivendi AS
23.01.2025	01-03	Saksbehandling	20250085	INN	Body	Modi Vivendi AS
23.01.2025	01-02	Saksbehandling	20250085	INN	Abstract	Modi Vivendi AS
23.01.2025	01-01	Saksbehandling	20250085	INN	Søknadsskjema Patent	Modi Vivendi AS

Description, claims and drawings

Disclaimer: This text has been machine-scanned and may contain errors – please refer to "Publications" for legally binding content.

Description

This invention is related to intelligent systems or vehicles for optimal and economically sound offshore mineral mining operations related to, but not limited to, manganese nodules, manganese crusts and sulphide deposits, based on differences in acoustic and/or electric properties among valuable minerals and non-valuable substances. Both independent vehicles and add-on systems to existing mining machines have been developed to selectively harvest valuable minerals. The developed systems will, by nature, provide added profitability and minimize any environmental impacts. MS and/or AI methods and algorithms can be incorporated as added features.

1. Introduction

All “references”, terms, definitions, and phrases related to all aspects mentioned in the (scientific) references, "Introduction", "Prior Art", various “Minerals” and “Invention” sections and in the figures, appendix, also apply to, form the basis of, and are incorporated into the invention represented by this document.

Some general elements like comments and discussions related to e. g. definitions, real time control systems etc. follow, to some degree, what is stated in our previous documents NO345271, NO347777 and NO20231270 (“self-reference”).

The statement “at least one of” means, from a set of variables, activities or processes (synonymous terms) [a1, a2, …an] either a1, a2, …or an isolated (single activity) or any combinations among the variables or thereof. The terms, “a”, “one”, “at least one of” and “any combinations of” are synonymous.

“System”, “vehicle”, “vessel” and “machine” are synonymous terms.

“Economic sound concentrations” and “profit level(s)” are synonymous terms.

Machine Leaning (ML) and Artificial Intelligence (AI) are synonymous terms.

“Collector”, separator” and “separation (and collection) mechanism” are synonymous terms. Other definitions are stated in the text to follow.

All electric and/or acoustic data listed in this document for minerals and other matter are stated as examples. Actual numerical data, based on, but not limited to, type and location, can both be higher and lower than the stated examples.

Examples of components, features and/or characteristics, constituting the systems to be developed, are stated throughout in the paragraphs to follow.

2. Offshore Minerals in Deep Oceans – General Remarks

Due to general scarcity, increasing prices and politically unfavorable locations of valuable minerals (e.g. Russia or China), constitute to the interest and focus for offshore mining of manganese crusts, nodules and/or sulphide deposits located on the seabed. In general manganese or ferromanganese nodules and/or crusts, typically consist of various minerals, including manganese oxides, iron oxides, and other trace elements. The mineral composition can vary depending on factors such as e.g. the location, depth, and geological conditions. E.g. manganese crusts are primarily composed of manganese oxides, which can range from 20% to ranging from 10% to 30%. Other trace elements, manganese crusts may also contain smaller percentages of other elements such as nickel, copper, cobalt, and rare earth elements (REEs), among others. These trace elements can collectively make up around 1% to 10% of the crust's or nodule’s composition or sulphide deposits.

For general scientific references, some reputable sources could include International Marine Minerals Society (IMMS), Geological Society of America (GSA), American Geophysical Union (AGU), Deep Sea Minerals (DSM) Observatory. See relevant websites for further details.

2.1 Mineral Deposits – Norwegian Offshore Deep Seas – General Remarks

The Norwegian Offshore Directorate’s expedition to the Mohns Ridge in the Norwegian Sea has identified a large area of e.g. sulphide minerals. The minerals are precipitated through “black smokers” on the seabed, where seawater heated by underlying magma chambers dissolves many minerals and then flushes them back out to the seabed.

Marine geoscience research indicates that the exploration and extraction of these seafloor massive sulphide (SMS) deposits could potentially yield a commercial resource base of millions of tons of copper, zinc, and cobalt, with copper being the most significant part of the findings. The economic potential of these deposits is substantial.

The typical mineral composition of manganese crusts in the Norwegian offshore deep seas primarily consists of manganese and iron, with smaller amounts of other metals such as cobalt, nickel, copper, lithium, scandium, and various REEs. These crusts were formed by precipitation from seawater and accumulate on underwater rock formations in the deep sea.

A detailed breakdown of the mineral compositions and their respective percentages for manganese crusts in the Norwegian offshore deep seas, as reported by the Norwegian Offshore Directorate and related marine geoscience research:

• Manganese (Mn): Predominant element,

• Iron (Fe): Second most abundant element,

• Cobalt (Co): Smaller amounts, significant for economic value,

• Nickel (Ni) and Copper (Cu): Present in trace amounts,

• Lithium (Li), Scandium (Sc), and REEs occur in minor concentrations.

The exact percentages can vary, but typically, manganese crusts are composed of approximately 15-30% manganese and 5-15% iron. Cobalt, nickel, and copper usually constitute less than 1% each, while lithium, scandium, and REEs are found in even smaller quantities.

The typical mineral compositions for sulphide deposits in the Norwegian offshore deep seas, particularly in areas like the mid-Atlantic Ridge between Jan Mayen and Svalbard, are as follows:

• Copper (Cu): Up to 14% in some samples,

• Zinc (Zn): Approximately 3%,

• Cobalt (Co): Less than 1%.

These deposits may also contain significant amounts of other minerals, including manganese, magnesium, vanadium, titanium, and various REEs minerals. The presence of these minerals is crucial for technologies such as batteries, wind turbines, PCs, and mobile phones.

General documentation for mineral compositions in Norwegian offshore deep seas can be obtained in e.g.: https://www.sodir.no/en/facts/seabed-minerals/manganese-crusts/ and https://maritime-executive.com/article/norway-discovers-seabed-mineral-deposits-innorwegian-sea.

3. Prior Art

WO2018068362A1 Describes a deep-sea mining system that includes a transport ship, an undersea self-propelled floating body, a seabed mobile delivery station, and a seabed mineral collection machine. The system uses flexible hoses to connect these components, allowing for the collection and transportation of minerals from the seabed to the ship. The design aims to simplify the structure and enable large-scale use.

US20230074267A1 Outlines a deep-sea mining vehicle designed to collect mineral deposits from the seabed and transport them to a floating device. The vehicle features a support frame with movement capabilities, a storage for collected minerals, and a suction head with a pressure chamber to facilitate the collection process. The vehicle is particularly suited for collecting polymetallic nodules like manganese nodules.

TechnipFMC, Wilhelmsen and NorSea are investing in deep-sea mining.

The stated article discusses the investment by TechnipFMC, Wilhelmsen, and NorSea in Loke Marine Minerals. They are developing a patent-pending autonomous subsea production system designed to presumably have minimal environmental impact. The system aims to enable efficient deep-sea mining operations.

https://www.offshore-energy.biz/technipfmc-wilhelmsen-norsea-investing-in-deep-seamining/.

In addition to the above stated prior art, the below listed documents 1-5 may have some relevance. Given the objective technical problem to be solved; how to optimize the collection of valuable seabed minerals by distinguishing economically viable deposits from uneconomic material using real-time decision-making, based on acoustic and/or electric properties (sensor data), the question would be how would a skilled person with knowledge of 1-5, might attempt to solve the problem?

1 (CN112282761A):

• 1 focuses on detecting manganese nodules on the seabed using sensors. The skilled person might start by refining the detection process, improving the accuracy of identifying mineral deposits using these sensors. However, 1 does not suggest or teach any method for making economic decisions about the profitability of the deposits.

• Likely approach: The skilled person would likely attempt to optimize the sensor system to better distinguish nodules from non-nodules (mud, gravel, etc.). However, without guidance on economic analysis or focus on intelligent optimization, the skilled person would rely solely on detection, not on evaluating profitability.

2(KR20170127836A):

• 2 discusses the use of a robotic mining system for collecting manganese nodules, but the focus is on the mechanical aspects of the mining process. There is no mention of real-time analysis or applying any economic thresholds.

• Likely approach: The skilled person might combine the mechanical systems of 2 with the detection systems of 1 to collect detected mineral deposits. However, the skilled person would not address the economic viability of the deposits or use any intelligent system to differentiate between profitable and unprofitable deposits.

3(CN115879648B):

• 3 focuses on detecting manganese nodules using sensors, similar to 1, and does not address economic analysis or intelligent systems for decision-making.

• Likely approach: The skilled person might look at 3 as reinforcing the detection methods described in 1. The emphasis would still be on detecting and locating mineral deposits, with no further optimization based on profitability.

4(Feng Hong et al.):

• 4 discusses using deep learning techniques to classify seabed materials using acoustic probes. While this document introduces machine learning, it does not apply this technique to evaluate the economic viability of deposits or guide the collection of economically valuable material.

• Likely approach: The skilled person might be inspired to use deep learning techniques to improve the classification of detected deposits. However, 4 does not teach using this classification for real-time economic decision-making, nor does it suggest how to optimize the collection process based on profitability.

5(Umesh Neettiyath et al.):

• 5 covers a robotic survey system that uses machine learning to estimate the distribution of manganese crusts. While it touches on machine learning, it does not suggest applying this technology for economic optimization or real-time selective extraction.

• Likely approach: The skilled person might apply machine learning to improve the spatial accuracy of detecting minerals, but they would not be led to incorporate economic thresholds or adaptive learning for profitability assessment.

The EPO Guidelines for Examination (G-VII, 5.1) explicitly state that an inventive step assessment should involve only the “most relevant” prior art (i.e., the closest prior art) and further documents may be combined if the skilled person would be motivated to do so.

However, combining more than two documents is generally discouraged unless there is a clear, simple, and obvious motivation for such a combination. Combining too many documents would leads to an unjustifiable complexity and would conflict the principles of the problem-solution approach. The EPO guidelines (G-VII, 8) strongly caution against the use of hindsight when combining documents to assess inventive step. Hindsight means that a person skilled in the art is able to combine prior art references only because they already know the claimed invention. In the problem-solution approach, the documents should naturally combine without knowledge of the invention, and the solution should be obvious at the time of filing.

Using the problem-solution approach, it is clear that a skilled person in the art, with knowledge of 1-5 but not of NO20240441 and this document, would likely focus on improving detection and mechanical collection techniques. However, they would fail to solve the objective technical problem as defined in NO20240441 and this document, therefore, the solutions provided represent non-obvious inventive step over the prior art.

4. Invention

Challenges related to offshore mining are to locate economic sound concentrations, with relevant economic valuable mineral compositions, based on any combinations, but not limited to, of nodules, manganese crusts and/or sulphide deposits. Also, current mining techniques add e.g. the absorption or extraction of mud, gravel, rock and/or other uneconomic matter when extracting presumable valuable minerals from the seabed. These non-economic types of matter need pumping power, can degrade equipment, can be elevated several 1000 m from the seabed to the surface, require top-side separation, thus adding significant costs to offshore mining operations.

The overall objective technical problem to be solved by the invention, is the development of intelligent systems to, but not limited to, locate and partially or fully extract economic sound concentrations of any combinations of nodules, manganese crusts and/or sulphide deposit, thus select and separate uneconomic matter from valuable minerals. Emphases are made on applying differences in electric and/or acoustic properties to differentiate economically valuable minerals or concentrations of minerals from cost bearing unprofitable other matter (e.g. mud, gravel, rocks, organic matter). Some debris (non-valuable matter) is expected to be extracted, thus the term “fully” (100 %) is a target.

Both stand-alone vehicles and add-on systems to existing harvesting machines (see e.g. paragraph 3) are to be developed.

MS and/or AI methods and algorithms can be added to the systems or vehicles.

4.1 The Concept of Economic Sound Concentrations

a. The density of any combinations of nodules, manganese crusts and/or sulphide deposit can vary on the seabed. It can be measured as numbers and/or weight (kg) per area (m2).

b. The mixture or composition of valuable minerals can vary as a function of area.

c. The concentration of various valuable minerals within e.g. nodules, manganese crusts and/or sulphide deposit vs un-valuable matter can vary (%).

d. Location can be a cost issue (e.g. distance to shore and general weather conditions can represent added cost elements).

e. Depth can be a cost issue.

f. Seabed topography can complicate operations and can represent an added cost element.

g. Commodity prices can vary.

h. Politics. Changing or different political governments and/or attitudes towards offshore mining can change.

i. Definition of profits and/or rate of return(s) and/or interest rates can vary.

j. Real extraction rate (%).

The listing a – j are examples of considerations or variables which will define the concept of “economic sound concentrations”. The concept “economic sound concentrations” will vary with location and time. A certain level of “economic sound concentrations” or “profit level(s)” will define a “threshold” or a targeted profitable value (or number). A “number” can be gross profits/week in USD, based on type of mineral (tons) multiplied by the commodity price for the specific mineral, and summarize for all the minerals in question. Such a “number” can (also) be represented by a probability, with or without any combinations of e.g., but not limited to, any combinations of the elements represented by the listing a – i or a - j. The real extraction rate (j) is particularly dependent on the density (a), composition of valuable minerals (b) and concentrations of various valuable minerals (c), related to the degree of selective harvesting. “Fully” extraction or 100 % selectivity, in the meaning that no nonvaluable matter is to be harvested, is a target. In practical offshore mining operations, some debris (non-valuable matter) is (also) expected to be extracted. In some circumstances the real extraction rate (j) can be purposefully defined or set by the user as a number other than (below) 100 %, depending on factors a – i. This if a lower extraction rate can maximize profits, based on local conditions. The concept “threshold” is a variable. Certain offshore mining operations which fulfill a threshold at one point in time can be terminated, temporarily or permanently, if the threshold is believed not to be reached.

5. Scientific Basis for the Invention

We emphasize throughout this document that local concentrations of valuable minerals and non-valuable substances have to be analyzed with respect to acoustic and/or electric properties. The outlined (general) remarks to follow, have to be considered as guidelines in the search for optimized solutions or profit levels or the obtaining of threshold level(s).

5.1 Electric Properties for Minerals

The typical electrical parameters for differentiating minerals include, but are not limited to, electrical conductivity, resistivity, and dielectric properties. These parameters can vary widely depending on, but not limited to, the type of mineral, its purity, and environmental conditions such as salinity, temperature and pressure.

For instance, the electrical conductivity of minerals is influenced by the presence of free ions and the mobility of these ions within the mineral structure in queston. Minerals with high ionic content, like halite (NaCl), typically have higher conductivity. In contrast, minerals with covalent or metallic bonding, like quartz or gold, exhibit different electrical behaviors.

Resistivity is another crucial parameter, which is the inverse of conductivity. It measures how strongly a mineral opposes the flow of electric current. This property is particularly important in geophysical exploration for deposits. Additionally, dielectric properties determine how a mineral responds to an electric field and are related to the mineral’s ability to store electrical energy.

Electrical Conductivity: This can vary widely from 10-2 to 107 S/m. For example, metallic minerals like native copper can have very high conductivity, while minerals with covalent bonds like quartz have much lower values.

Resistivity: Common rock types can have resistivity values ranging from 102 to 108 ohm-m. Sulfide minerals, which often have partially metallic bonds, tend to have lower resistivity.

Dielectric Constant: This is typically in the range of 2 to 200 for most minerals. It’s a measure of a mineral’s insulating ability.

These are general ranges and have to be treated as examples. Specific minerals can have values that fall outside these ranges under certain conditions, e.g. dependent on location, depth, etc.

5.1.2 Electric Properties for Sulphide Deposits

The typical electrical parameters for offshore sulphide deposits, such as SMS (Seafloor Massive Sulphide), are characterized by their distinct electrical properties compared to unmineralized host rocks. These properties are crucial for exploration and resource evaluation, as they also can be used to detect SMS deposits during drilling operations, potentially reducing costs and accelerating offshore activities.

Some examples of electrical properties based on scientific research:

Frequency-Dependent Complex Electrical Resistivity: SMS deposits exhibit frequencydependent complex electrical resistivity in the range between 0.002 and 100 Hz. This property can potentially discriminate between different types of fresh rocks, alterations, and mineralizations.

Resistivity Magnitude: The resistivity magnitude alone, with appropriate electrical sensors, can distinguish between mineralized and non-mineralized samples. Exploiting the frequencydependent behavior of resistivity might amplify the differences and further improve rock characterization.

Self-Potential and Magnetic Susceptibility: These parameters differ distinctly between sulfides and the host rocks and sediments, which is beneficial for prospecting SMS deposits.

5.2 Acoustic Parameters for Minerals

Acoustic parameters for minerals are essential for understanding their behavior under various conditions. Some key parameters and references, but not limited to;

Centroid frequency (Fcf) and transmission coefficient (Tcf): These parameters are generally positively correlated with grain uniformity in rocks. The mineral grain size impacts these parameters differently in igneous and sedimentary rocks.

Ultrasonic characteristics (Vp, Fcf, Tcf) and damage characteristics (D3): These are significantly influenced by the main mineral compositions such as quartz, plagioclase, and kfeldspar.

Longitudinal velocity (Vp): This correlates with the density and mean atomic weight of rocks. It can be estimated from modified Hashin-Shtrikman bounds for rocks with similar mean atomic weight.

Sound velocity: For minerals, the longitudinal sound velocity typically ranges from 4,000 to 7,000 m/s. This parameter is crucial for understanding the elastic properties of minerals and can vary depending on the mineral composition and structure.

Acoustic impedance (Z): This parameter is defined as the product of the density of the material and the sound velocity through it. For most minerals, acoustic impedance can range from 15 to 30 MRayl (mega ray-leighs).

Porosity: In porous minerals, porosity can significantly affect acoustic properties. The porosity percentage can vary widely, from less than 1% to over 30%, depending on the type of mineral and its formation process.

These parameters and values are examples. Other parameters can be applied depending on local conditions. In this context “sound” and “velocity” are synonymous terms.

5.2.1 Acoustic Parameters for Offshore Sulphide Deposits

Acoustic parameters for offshore sulphide deposits are crucial for exploration and resource evaluation. Here are some typical parameters;

Sound Velocity (Vp): The sound velocity in sulphide deposits can vary significantly, but it generally ranges from 1,500 to 6,000 m/s depending on the composition and porosity of the deposit.

Density (ρ): The density of sulphide deposits is typically higher than that of the surrounding water and sediments, ranging from 3,000 to 5,000 kg/m³.

Acoustic Impedance (Z): This is calculated by multiplying the density (ρ) by the sound velocity (Vp) and can range from 5 MRayl to 30 MRayl.

-Attenuation (α): Attenuation rates can be high in sulphide deposits due to the complex structure and can range from 0.1 to 2 dB/m.

These parameters and values are examples. Other parameters can also be applied dependent on local conditions.

5.2.2 Acoustic Properties of Offshore Manganese Crusts and/or Nodules

Offshore manganese crusts, known for their cobalt-rich composition, have distinct acoustic properties that are key for exploration and resource assessment. Some typical acoustic parameters for crusts:

Sound velocity (Vp): The velocity of sound through manganese crusts can vary, but it typically ranges from 1,500 to 6,000 m/s.

Density (ρ): Manganese crusts have a higher density than seawater, generally ranging from 2,000 to 3,000 kg/m³.

Acoustic Impedance (Z): Typical values for manganese crusts ranging from 10 MRayl to 18 MRayl.

Attenuation (α): The attenuation of sound in manganese crusts can be significant due to their structure, with values ranging, but not limited to, from 0.1 to 1 dB/m.

These parameters and values are examples. Other parameters can also be applied depending on local conditions.

Understanding the electrical and acoustic properties of minerals and geological deposits is essential for geophysical exploration and resource assessment related to economically sound, optimal and intelligent offshore mineral mining. Below is a summary of some of these properties, including numerical values where available, along with detailed references specifying the source documents and page numbers.

5.3 Further Comments with Specific Scientific References

5.3.1 Electrical Properties

Sulfide Minerals: Sulfide ores or deposits, such as those containing nickel (Ni), copper (Cu), and Platinum Group Elements (PGE), exhibit high electrical conductivity due to the metallic nature of their bonding. These minerals are strongly anomalous in electrical conductivity compared to their host rocks.

Source: King, A. (2007). Review of Geophysical Technology for Ni-Cu-PGE deposits.

Proceedings of Exploration 07: Fifth Decennial International Conference on Mineral Exploration, 647–665.

Seafloor Massive Sulfides (SMS): SMS deposits are characterized by distinct electrical properties compared to unmineralized host rocks. The presence of conductive sulfide minerals significantly influences the electrical conductivity of these deposits.

Source: Spagnoli, G., Hannington, M., Bairlein, K., Hördt, A., Jegen, M., Petersen, S., & Laurila, T. (2016). Electrical properties of seafloor massive sulfides. Geo-Marine Letters, 36, 235–245.

Offshore Manganese Crusts and Nodules: Manganese nodules and crusts are known to have specific electrical properties that can be detected using geophysical methods. These properties are influenced by their mineral composition and the surrounding environment.

Source: Hein, J. R., Mizell, K., Koschinsky, A., & Conrad, T. A. (2013). Deep-Ocean Mineral Deposits as a Source of Critical Metals for High- and Green-Technology Applications:

Comparison with Land-Based Resources. Ore Geology Reviews, 51, 1–14. Available at Academia.edu.

5.3.2 Acoustic Properties

Sulfide Minerals: Sulfide ores exhibit distinct acoustic velocities, which can be utilized in geophysical surveys to detect and characterize these deposits.

Source: King, A. (2007). Review of Geophysical Technology for Ni-Cu-PGE deposits.

Proceedings of Exploration 07: Fifth Decennial International Conference on Mineral Exploration, 647–665.

Offshore Sulphide Deposits:The acoustic properties of seafloor massive sulfide deposits are influenced by their mineral composition and porosity. Understanding these properties is crucial for interpreting sub-seafloor images from geophysical surveys.

Source: Ohta, Y., Goto, T., Koike, K., Kashiwaya, K., Lin, W., Tadai, O., Kasaya, T., Kanamatsu, T., & Machiyama, H. (2024). Correlation between induced polarization and sulfide content of rock samples obtained from seafloor hydrothermal mounds in the Okinawa Trough, Japan. Earth, Planets and Space, 76, Article number: 54.

Offshore Manganese Crusts and Nodules: The acoustic properties of manganese nodules and crusts are essential for their detection and assessment. Recent studies have demonstrated the use of parametric acoustic probes in recognizing manganese crusts.

Source: Li, J., Li, X., & Zhang, W. (2022). First Demonstration of Recognition of Manganese Crust by Deep-Sea Parametric Acoustic Probe. Minerals, 12(2), 249.

Hein, J. R., Mizell, K., Koschinsky, A., & Conrad, T. A. (2013). Deep-Ocean Mineral Deposits as a Source of Critical Metals for High- and Green-Technology Applications:

Comparison with Land-Based Resources. Ore Geology Reviews, 51, 1–14. Available at Academia.edu.

6. Primary Components and Features Constituting the Invention

This section follows to some degree the (self) references stated in the “Introduction” chapter related to e.g. sensors and control and guidance system. Paragraph 6 represents primary components, features and characteristics of the invention, to be developed in paragraphs 7 and 8.

A system is a group of interacting or interrelated elements or components (synonymous) that act or are interconnected according to a set of rules, features (synonymous terms) or mechanisms to form a unified whole. A system is surrounded by its boundaries.

6.1 Sensors

Besides the at least one electronic and/or acoustic sensor, the at least one sensor incorporated in any of the systems and/or elements within a system constituting this invention can, but are not limited to, detect (quality and quantity) pressure, temperature, heat (infrared), velocities, frequencies (sound, light), stress, strain, movements, motion, represent cameras, low light cameras, infrared cameras, GPS trackers, telephones, Internet of Things (IoT), satellite communications. Such sensors can be, but are not limited to, analog or digital electronic, electro - mechanical, optical or of any sound or ultrasound types. The at least one sensor can be connected by at least one wire or wireless communication unit or device, but not necessarily.

6.2 Real-Time Control and Guidance Systems (RTCGS)

The at least one data processing unit, within the invention or systems to be developed, will include real-time control, which requires controllers to capture all the significant target activities and to deliver their responses as swiftly as possible so that system performance is never degraded.

In Advanced Industrial Control Technology, Peng Zhang, 2010, ISBN: 978-1-4377-7807-6, https://www.sciencedirect.com/book/9781437778076/advanced-industrial-control technology#book-info, structure and requirements to real-time control systems are outlined, excerpts (chapter 1.2.1, line 7):

“A control operation is a series of events or actions occurring within system hardware and software to give a specific result. A real-time control system is a system in which the correctness of a result depends not only on its logical correctness but also on the time interval in which the result is made available. The following three standards give the definition of a real-time control operation, and an industrial control system in which all the

control operations occur in real-time qualifies as a real-time control system:

(1) Reliable operation execution - the operation execution must be stable, and it must be repeatable.

(2) Determined operation deadline - any control operation needs time to execute.

(3) Predictable operation result - the result for any control operation must be predictable.”

Based on e.g. combinations of, but not limited to, Model Predictive Control, (MPC), Proportional Integral Derivative (PID), Deep Reinforcement Learning (DRL), three characteristics of ML/AI-based controllers can be utilized, if ML/AI capabilities are to be applied as added features;

1. Learning: DRL- based controllers learn by methodically and continuously practicing ML.

2. Delayed gratification: DRL- based controllers can learn to recognize sub-optimal behavior in the short term, which enables the optimization of gains in the long term.

3. Non-traditional input data: DRL- based controllers manage the intake and are able evaluate sensor information that automated systems cannot.

The enablement of e.g. DRL- based control systems to a process facility, require, but are not limited to, the following steps in delivering a DRL- based controls:

1. Preparation of a companion simulation model for the (AI) brain,

2. Design and training of the (AI) brain,

3. Assessment of the trained (AI) brain,

4. Deployment.

It is to be emphasized that applications of any combinations of MPC, PID or DRL do not require ML/AI models.

6.2.1 ML/AI can be incorporated as Added Features

In this context the ML/AI brain(s), as features within the at least one data processing unit

is/are, but not limited to, to be trained to, at least one of; foresee, forecast, predict, simulate,

optimize or otherwise.

Aid in offshore mining for e.g. manganese crusts, nodules, and sulphide deposits by enhancing exploration, extraction, and processing activities can be advantageous. Some ways ML/AI can contribute to this field can be, but are not limited to:

• Exploration: AI can analyze geological data to predict the location of mineral deposits with greater accuracy. ML/AI learning algorithms can process vast amounts of seismic and subsea data to identify patterns, but not limited to, that may indicate the presence of e.g. manganese crusts or nodules or sulphide deposits.

• Extraction: ML/AI-driven autonomous underwater vehicles (AUVs) and remotely operated vehicles (ROVs) can be used for the precise and efficient collection of mineral deposits. These vehicles can navigate challenging underwater terrain and perform tasks with minimal human intervention.

• Processing: ML/AI can optimize the processing of collected minerals by predicting the quality of mineral deposits and suggesting the most efficient processing methods. This can lead to a reduction in energy consumption and an increase in yield.

• Environmental Monitoring: ML/AI can help monitor the environmental impact of mining activities by analyzing data from various sensors in real-time. This allows for immediate adjustments to minimize ecological disruption.

• Safety: ML/AI can enhance the safety of offshore mining operations by predicting equipment failures and potential hazards.

Overall ML/AI has the potential to make offshore mining for manganese crusts, nodules, and sulphide deposits more efficient, safer, and environmentally friendly.

6.2.1.1 Development of ML/AI Components to Fulfill the Requirements for “Systems for Optimal Offshore Mineral Mining” – if Applied

To effectively control the provided optimal mineral mining system, with priority from NO20240441, multiple integrated AI systems work together to enable intelligent decisionmaking, efficient resource collection, and environmental protection. At the core, sensory systems process acoustic and electrical data to identify valuable mineral deposits, while a Deep Q-Learning agent makes high-level navigation and collection strategy decisions. These systems are supported by a comprehensive mapping and data collection framework that maintains historical information and helps optimize future operations.

The AI components are initially trained in simulation environments that model seabed physics and sensor responses, allowing for safe development and testing before deployment. As the system operates in real conditions, it continuously learns and improves through self-correction mechanisms, adapting to new scenarios and optimizing its performance based on actual operational data.

Each component is designed with fallback systems and safety protocols, ensuring reliable operation even in challenging deep-sea conditions. The integration between these systems is carefully orchestrated to maintain real-time processing capabilities while meeting the economic and environmental requirements specified in the patent.

This specification details each component's functionality, their interactions, and how they collectively achieve the provided document's objectives of efficient, environmentally conscious offshore mineral mining.

The provided components represent examples. The template is fully in compliance with EPO decisions T0161/18 and T1069/21. Other components and sub-systems and/or models can be applied.

A thorough assessment of primary components and steps within a suggested ML/AI system is provided in the appendix, which represents an integral part of this document.

Excerpts;

1. Simulation System

A comprehensive simulation environment is required to develop and validate the AI systems before real-world deployment. This environment serves multiple purposes: training the DQL navigation agent, testing sensor processing algorithms, and generating synthetic training data for the classification systems. The simulation must accurately model the complex dynamics of deep-sea operations while remaining computationally efficient enough for rapid training iterations. It could further comprise:

1.1 Physics Modeling

The simulation incorporates key physical aspects of the deep-sea environment:

1.2 Sensor Response Simulation

Accurate modeling of sensor responses is crucial for developing robust detection algorithms:

1.3 Environmental Variations

The simulation includes dynamic environmental conditions to ensure system robustness:

1.4 Mineral Deposit Scenarios

A variety of test scenarios are implemented to validate system performance:

1.5 Training Data Generation

The simulation environment serves as a primary source of synthetic training data, generating extensive datasets for training both the sensory prediction systems and the DQL agent: 1.5.1 Large-scale Synthetic Dataset Creation

1.5.2 Data Augmentation and Variation

1.5.3 Ground Truth Labeling

1.5.4 Validation Framework

This comprehensive data generation system ensures that the AI components have sufficient training data to develop robust and accurate prediction capabilities before real-world deployment, while the validation framework provides quantitative metrics for system performance assessment.

2. Mapping and Data Collection System

The mapping and data collection system serves as the foundational layer for all other AI components, maintaining a comprehensive understanding of the operational environment and historical data. This system continuously updates a 3D representation of the seabed while storing and organizing all sensor readings and operational outcomes. It could further comprise:

2.1 Real-time 3D Mapping

2.2 Data Storage and Organization

2.3 Mineral Deposit Tracking

2.4 Environmental Monitoring

2.5 Integration with Decision Systems

This system not only maintains the operational knowledge base but also provides crucial input for the DQL agent's decision-making process and enables continuous improvement of the prediction systems through comprehensive performance tracking.

3. Sensor Data Processing System

The sensor data processing system handles the raw acoustic and electrical sensor inputs, processes them to extract meaningful features, and evolves over time through the incorporation of deep learning models trained on operational data. This hybrid approach ensures reliable baseline performance while enabling continuous improvement through learning. It could further comprise:

3.1 Traditional Signal Processing Pipeline

3.2 Deep Learning Enhancement

3.3 Sensor Fusion

3.4 Performance Monitoring

3.5 Integration Points

4. DQL Navigation and Decision Agent

The Deep Q-Learning Navigation and Decision Agent serves as the primary decision-maker for the system, optimizing exploration, scanning, and collection strategies to maximize economic return while minimizing environmental impact. This agent operates continuously in a complex state space, making both strategic and tactical decisions about system operation. It could further comprise:

4.1 State Space Definition

4.2 Action Space

4.3 Reward Function Design

4.4 Training and Optimization

4.5 Performance Monitoring

5. Classification and Prediction System

The Classification & Prediction System processes the sensor data to identify mineral deposits, predict their quantities, and estimate their economic value. This system employs a combination of machine learning models working together to provide accurate predictions with associated confidence metrics. It could further comprise:

5.1 Mineral Classification

5.2 Quantity Prediction

5.3 Economic Value Estimation

5.4 Uncertainty Quantification

5.5 Confidence Scoring

5.6 Integration and Updates

6. System Integration Architecture

The integration architecture ensures seamless operation between all AI components and physical systems, providing robust, real-time performance while maintaining safety and reliability. It could further comprise:

6.1 Data Flow Architecture

6.2 Hardware Integration

6.3 Operational Control Flow

7. Self-Correction and Learning System

The self-correction and learning systems enable continuous improvement of all AI components through operational experience, ensuring optimal performance and reliability over time. It could further comprise:

7.1 Continuous Learning Framework

7.2 Error Detection and Correction

7.3 Knowledge Integration

7.4 Safety and Reliability

7.5 Long-term Optimization

8. Performance and Safety System

This system provides comprehensive monitoring, evaluation, and safety mechanisms, ensuring reliable operation while maintaining clear performance metrics and robust fallback procedures. It could further comprise:

8.1 Performance Metrics Framework

8.2 Fallback Systems

8.3 Emergency Response System

8.4 Fault Tolerance Architecture

8.5 Safety Enforcement

8.6 System Validation

9. Economic Optimization System

This section details the economic decision-making capabilities of the AI system, providing comprehensive specifications for market adaptation, cost optimization, and value maximization strategies. It could further comprise:

9.1 Market Intelligence System

9.1.1 Real-time Market Data Integration

9.1.2 Predictive Market Modeling

9.2 Operational Cost Optimization

9.2.1 Dynamic Cost Modeling

9.2.2 Resource Allocation Optimization

9.3 Value Maximization Framework

9.3.1 Grade-Based Optimization

9.3.2 Collection Strategy Optimization

9.4 DQL Agent Economic Integration

9.4.1 Economic State Space

9.4.2 Economic Action Space

9.5 Adaptive Economic Strategy System

9.5.1 Short-term Optimization

9.5.2 Long-term Strategy

9.6 Performance Metrics and Reporting

9.6.1 Economic Performance Metrics

9.6.2 Reporting and Analysis

This comprehensive economic framework ensures optimal financial performance through sophisticated market intelligence, cost optimization, and value maximization strategies. The system's adaptive nature allows for dynamic response to changing market conditions while maintaining operational efficiency.

10. Environmental Protection System

This section details the environmental protection mechanisms of the AI system, providing comprehensive specifications for impact assessment, prevention, monitoring, and ecosystem preservation strategies. It could further comprise:

10.1 Environmental Impact Assessment

10.1.1 Physical Impact Monitoring

10.1.2 Ecosystem Impact Assessment

10.2 Preventive Protection Framework

10.2.1 Pre-operation Assessment

10.2.2 Operational Controls

10.3 Real-time Monitoring Systems

10.3.1 Continuous Environmental Monitoring

10.3.2 Adaptive Response System

10.4 AI-driven Protection Mechanisms

10.4.1 Neural Network Environmental Analysis

10.4.2 Protective Decision Making

10.5 Recovery Management System

10.5.1 Post-operation Monitoring

10.5.2 Adaptive Management

10.6 Integration and Reporting

10.6.1 Environmental Performance Metrics

10.6.2 Documentation and Reporting

This comprehensive environmental protection framework ensures responsible resource extraction while maintaining ecosystem integrity through sophisticated monitoring, protection, and recovery systems. The integration of AI-driven decision making enables proactive environmental protection and effective impact mitigation.

This is an example of primary components and steps within a suggested ML/AI system and/or algorithm. Actual system can vary dependent on actual design of the developed vehicle (fig.1, fig. 3) and/or adaptations to local conditions, at the area of applications.

6.2.2 Selv-Correction

With or without AI/ML algorithms, the at least one RTCGS (as of paragraph 6.2) can be designed to correct their own predictions over time and improve their accuracy. This can be achieved through a process of continuous learning and adjustment based on the comparison of predicted and actual outcomes. E.g. in Q-learning, a form of reinforcement learning, a selfcorrecting algorithm has been proposed to balance the overestimation of the single estimator used in conventional Q-learning and the underestimation of the double estimator used in Double Q-learning. As the algorithm(s) receive(s) feedback about its performance, it could adjust its parameters to improve future predictions and recommendations. As an example, “Advances in Neural Information Processing Systems”, 33 (NeurIPS 2020), Edited by: H. Larochelle and M. Ranzato and R. Hadsell and M.F. Balcan and H. Lin (ISBN:

9781713829546) could provide further details related to Q-learning in “Selv-Correction” contexts.

6.3 Example of High-Level Algorithm(s) or Pseudocode(s)

An integral part of the at least one RTCGS (as of paragraph 6.2), would require at least one pseudocode. Programmers need at least a pseudocode or a readable description providing a detailed template for writing a code in a specific language to provide commands to the at least one selection, separation and collection mechanism. Paragraph 6.3.1 is such an example, written in Python.

To separate, but not limited to, manganese crusts, nodules or sulphide deposits from other matter based on electric and/or acoustic parameters, need signal processing.

Such a high-level algorithm could, but not limited to, incorporate the following steps:

1. Data Collection: Gather acoustic and/or electric parameter data from the area of interest. This could include e.g., but not limited to, sound speed or velocity, density, acoustic impedance, and attenuation for acoustic parameters, and resistivity or conductivity for electric parameters.

2. Data Preprocessing:

- Filter the acoustic data to remove noise and enhance signal quality.

- Normalize electric sensor data to account for variations in sensor performance and environmental factors.

3. Feature Extraction: Process the collected data to extract meaningful features that can, but not limited to, differentiate between e.g. manganese crusts, nodules and/or sulphide deposits, and other matter. Features could, but not limited to, include statistical descriptors of the signal(s), spectral content, and derived quantities like reflection coefficients.

4. Data Labeling: If not already labeled, the data needs to be classified into categories (e.g., manganese crusts, nodules, sulphide deposits, other matter) based on ground truth information obtained from e.g. samples or previous surveys.

5. Signal Processing: Use the labeled dataset to train a supervised model such as e.g. a Convolutional Neural Network (CNN) model, which can recognize patterns in the feature set that are indicative of different materials.

6. Real Time Classification: Apply the trained model to new data to classify it into categories. Train the model e.g. on labeled datasets where the ground truth (manganese crusts, nodules, sulphide deposits, or other matter) is known. The model will output probabilities for each category, and the highest probability can determine the classification.

7. Separation Decision/Threshold: Incorporate at least e.g. a set of threshold probabilities for the model's output to decide whether the material is a manganese crust, nodules or sulphide deposit of economic value or meets the at least one threshold value (see e.g. paragraph 4.1).

8. Post-Processing: Apply additional rules or filters based on domain knowledge to refine the classification results. For example, geological knowledge about the distribution of these deposits can help in validating the model’s output.

9. Verification and Self-Correction: Periodically collect samples of the separated material and analyze them to verify the accuracy of the separation process. Adjust the at least one model and threshold settings based on verification results to improve accuracy.

ML and/or AI techniques could be incorporated as added features in this context.

This example of algorithm(s) leverages both acoustic and/or electric parameters to identify and separate target mineral deposits. For implementation, a dataset of labeled examples to train the classification model and a system capable of executing the algorithm in the deep-sea environment would be most advantageous. Note - This is a high-level overview, and developing fully functional systems would require detailed engineering.

None of the above stated features are limited to the stated characteristics. The stated algorithm(s) have to be viewed as example(s). Real implemented algorithms can include more or less statements and/or more or less features.

6.3.1 A Python Program

To creating a conceptual outline of a Python program for separating e.g. manganese crusts, nodules and/ or sulphide deposits, based on the algorithm(s) outlined in paragraph 6.3, but not limited to, would most likely (also) have to be based on more specific input materials. The algorithms would likely involve signal processing techniques to classify the data accurately. In this example we have included a ML classifier.

import numpy as np

import scipy.signal

from sklearn.ensemble import RandomForestClassifier

def preprocess_data(data):

# Implement preprocessing steps such as normalization, filtering, etc. pass

def extract_features(data):

# Extract relevant features from the data, which could include statistical,

# spectral, or temporal features depending on the nature of the data. pass

def train_classifier(features, labels):

# Train a machine learning classifier, such as a random forest,

# using the extracted features and provided labels.

classifier = RandomForestClassifier()

classifier.fit(features, labels)

return classifier

def classify_data(classifier, data):

# Use the trained classifier to predict the class of new data samples. features = extract_features(data)

return classifier.predict(features)

# Example usage:

# Load your dataset containing electric or acoustic parameters

data = np.load('data.npy')

preprocessed_data = preprocess_data(data)

# Assume labels are provided for supervised learning

labels = np.load('labels.npy')

# Feature extraction

features = extract_features(preprocessed_data)

# Train the classifier

classifier = train_classifier(features, labels)

# Classify new data

new_data = np.load('new_data.npy')

predictions = classify_data(classifier, new_data)

This is a high-level template pseudocode and would need to be adapted to the specific characteristics of the data and the problem at hand, and as such it must be view as an example. The actual implementation would require detailed knowledge of local geography, of the data acquisition process, the detailed electric and/or acoustic parameters involved, and the specific mechanical extraction solution or arrangement(s) applied for separating, but not limited to, manganese crusts, nodules and/or sulphide deposits from other matter.

7. Examples of (Independent) Systems or Vehicles to Distinguish Manganese Crusts, Nodules and/or Sulphide Deposits from Un-Valuable Matter

A system capable of executing the outlined or developed algorithms for separating manganese crusts, nodules and/or sulphide deposits using electric and/or acoustic input and/or generated parameters, would likely be an advanced underwater vehicle, but not limited to, with the following components and/or features. Also see fig.1 and fig.2 for visual presentations, where the following numbered items represent key components:

1.Sensing Unit:

The system would need sensors capable of measuring the electric and/or acoustic parameters of the deposits. These could include combinations of electric field sensors and acoustic transducers.

a. Acoustic Sensors: Could also be applied to collect high-resolution echo data from the seabed, which could help in identifying the location, density and composition of the various materials.

b. Electric Sensors: To measure e.g. the electrical conductivity and/or resistivity of the seabed material, as different minerals have distinct electrical properties, as previously discussed.

2. Data Processing Unit:

Equipped with at least one powerful onboard computer to process the collected acoustic and/or electric data obtained from (1), in real-time.

Could also utilize learning and/or self-correcting algorithms to analyze the acoustic and electric data, classify materials, and determine the presence of targeted (high value) minerals. In addition to functionalities, algorithms and/or characteristics outlined and discussed in paragraphs 6.2 and/or 6.3, the following features would be strongly advantageous;

c) Processing Unit with Key Functionalities: This could be the (one) computer in question or an added microcontroller that runs the at least one separation algorithm. It would take the sensor data as input and output commands to the separation and collection mechanism (4).

d) Data Acquisition Sub-System: This sub-system would collect the data from the sensors and convert them into a format that can be processed. This often involves analogue-to-digital conversion, but not necessary.

e) Wavelet-Based Algorithm: Could Improve signal quality by reducing system interference and ambient noise. It may include a wavelet-based envelope extraction method, and an adaptive estimation strategy based on dual-channel information.

f) Wavelet Regression: Can be applied to reduce measuring noise, assuming that the crusts have local thickness invariability. This is particularly useful for crusts with a structure of multilayers at the top surface and a single layer at the bottom surface.

g) User Interface: This would allow the user (synonymous to “master control station”) to start and stop the process, adjust parameters, and monitor the system’s status. The user can be located top-side (manned or un-manned surface vessel) and/or be a submerged manned or unmanned (independent) machine and/or a vehicle (10).

3. Navigation and Mobility System:

i) A propulsion system that allows the vehicle to maneuver over the seabed with precision. This could be any combination of propellers and/or impellers. Any possible intake(s) for water jet is not shown. This would enable free movements in water and/or on the seabed.

j) And/or a propulsion system that can include any combination of wheels and/or belts. This would enable movements on the seabed.

Autonomous navigation capabilities guided by sonar and/or GPS to move to specific coordinates, could be used.

4.Separation and Collection Mechanism:

Based on output from the at least one RTCGS (as of paragraphs 6.2 and 6.3) and a processor (2), a separation and collection arrangement, functionality and/or sub-system, will be applied. This could be an electro-mechanical system that physically selects and/or separates the deposits based on the commands from the processing unit (2). The exact nature of this mechanism would or could depend on the physical properties of the deposits and the environment in which they are found.

As an example of such a mechanism could be at least one degrees of freedom robotic arm, a trunk and/or proboscis arrangement (4, m), which could combine or represent any combinations of, but not limited to, a suction device, a claw and/or a cutting, crushing or grinding (synonymous terms) device, to collect the identified mineral deposits of value. Examples of such (independent) mechanisms, are;

Example of a suction robotic arm with multiple degrees of freedom.

Source: https://superakbov.best/product„tag/85648945„.html

Example of a robotic claw arm. Source: https://thinkrobotics.com/products/highprecision-industrial-robot-arm-with-planetary-stepper-motors

Example of a cutter assembly. Jackson, E. and Clarke, D. “Excavation of Seafloor Massive Sulphides. ’Ocean 2007. Source:

www.researchgate.net/pubhcation/224305659 Subsea Excavation of Seafloor Massive Sul phides.

Such a robotic arm and/or trunk (4, m) would operate with at least one degree of freedom (4, 1 - letter 1) (x, y, z) on the outside of the vehicle (fig. 1). This movement and/or rotational component (4, 1 - letter 1) will receive output signals from the at least one processor (2).

The collector or separator (separation and collection mechanism) (4) can include additional sensors (ref. paragraph 6.1). These can be, but not necessarily, of any acoustic and/or electronic type.

The collector component or a robotic arm and/or a trunk (4, m) can be replaceable or interchangeable with a different collector component or a robotic arm and/or a trunk (4, m) with other arrangements (e.g. any combination of, but not limited to, a suction device, a claw, cutting features, sensors) depending on local conditions. This to maximize profit level(s). The connecting mechanism (4, k) of such an interchangeable collector component (4, m) can be, but is not limited to, an electro-mechanical bayonet type of arrangement. The collector (4) and the collector component (4, m) can be identical arrangements, and/or (4) can represent additional features, like e.g. sensors, located between (4, m) and the at least one local storage unit (8). This or these intermediate elements are not shown.

The connecting mechanism (4, k) and the potential rotational component (4, l – letter l) can be integrated into one unit.

The collector (4) can be built-in within the vehicle (4, n), (fig.1), meaning it does not stick out, providing the belly of the vehicle or system (fig. 1) to be smooth.

Further engineering examples of a dynamic electro-mechanical robotic arm (4), a trunk (4), (4, m), a movement and/or a rotational component (4, l – letter l), can be represented by;

Esposito, D. et al, “Design and Evaluation of a Low-Cost Electromechanical System to Test Dynamic Performance of Force Sensors at Low Frequencies,” Machines 2022, 10(11), 1017; https://doi.org/10.3390/machines10111017.

Lu, Y., Lu, X., Ye, G. et al. «Study of electromechanical coupling dynamical characteristics of the grinding robot joint transmission system”, Eur. Phys. J. Plus 139, 767 (2024). https://doi.org/10.1140/epjp/s13360-024-05577-2.

Palpacelli, M, et al, “Details on the Design of a Lockable Spherical Joint for Robotic Applications,” Journal of Intelligent & Robotic Systems 81(2), April 2015, DOI:10.1007/s10846-015-0230-2.

5. Communications and Navigation System:

Represents a robust communication device or link enabling to transmit data and receive instructions from the surface and/or submerged master control station (user) and/or at least one external vessel (10). May include acoustic modems for underwater communication and satellite links when at the surface. Will most likely include a GPS tracking device.

6. Power Supply:

A combination of e.g. batteries and possibly fuel cells to provide long-duration power supply for extended missions, or any other wireless and/or wired power supply.

7. Environmental Monitoring and/or Sonar:

Sensors to monitor environmental parameters to ensure compliance with environmental regulations and minimize the ecological impact.

Could and/or represent a sonar, most probably an acoustic device (1, a). It could provide added navigational data and/or a global overview of the area or seabed in question.

8. Local Storage Unit:

A storage unit to hold the collected minerals until they can be transported to the surface, or it could be connected to the surface by a (flexible) hose (not shown).

9. Hatch or Locking Device:

Arrangement to connect an external storage unit (not shown), (e.g. a vessel and/or a tank) to the local storage unit (8). The means of connection could be any combinations of a pipeline and/or a flexible hose (not shown on any of the figures).

The total system is represented by omponents 1 – 9.

Additional sensors (ref. paragraph 6.1) can be located on and/or within a vehicle (fig. 1) (not shown).

The total system (1 – 9) could be deployed from both at least one surface vessel (not shown) and/or at least one existing external underwater production vehicle or device (10) and/or be remotely operated and/or be programmed to perform autonomous missions. It would execute the algorithms by collecting data, processing then to identify target valuable minerals, and then physically selecting and/or separating and storing the valuable minerals for retrieval. The degree of separation can vary, ref. discussion related to listing (j) in paragraph 4.1. The design would need to be robust to withstand the harsh conditions of the deep-sea environment. This is an example. Actual systems could include both fewer or additional added features and/or additional algorithms, and would require multidisciplinary engineering and computer science to be commercially designed.

The developed systems facilitate added profitability and reduce environmental impacts.

The at least one component, feature and/or system(s) as outlined in paragraph 7 and fig.1 and fig. 2 could (also) be totally or partially integrated into existing deep-sea mining vehicles (10), to execute algorithms for the selection and/or separation, but not limited to, of manganese crusts, nodules or sulphide deposits based on electric or acoustic parameters. These systems (or harvesting machines) (10) could then be capable of distinguishing these materials from other matter, thus facilitating more profitable, efficient and targeted mining operations.

Examples of such mining machines (10) are discussed in paragraph 3 (Prior Art).

8. An Add-On System or a Pathfinder to an Existing Offshore Mineral Extraction Vehicles

A high-level description of an add-on system (“path finder”), capable of executing algorithms for, but not limited to, separating manganese crusts and/or sulphide deposits using electric and/or acoustic parameters, for at least one external offshore mining vessel (10), has been provided. Examples of such existing external offshore mining vessels are described or discussed in the Prior Art section (paragraph 3).

Such an add-on system(s) is/are described as outlined in fig. 3 and fig.4:

It would include, but not necessarily, the components, features and/or characteristics of paragraphs 6 and 7. However, the components 4, 8 and 9 could be excluded. Added acoustic and/or electronic components (1*, a, b) could be advantageous. The at least one communication unit (5) could be wireless or wired to the at least one external vessel or harvesting machine (10), which can collect minerals and/or objects from the seafloor. If the at least one add-on system is wired to the at least one external vessel (10), it could be powered by the at least one external vessel (10) and the internal power unit (6) could be abandoned. The at least one add-on system would guide or direct the at least one external vessel (10) by input data representing economic sound concentrations of valuable minerals, ref. paragraph 4.1, and provide added profitability and reduced environmental impacts.

Additional sensors (ref. paragraph 6.1) can be located on and/or within a vehicle (fig. 3) (not shown).

The path finder (fig.3, fig 4) can be remotely operated and/or be programmed to perform autonomous missions.

This is an example of a high-level conceptual design, and the actual implementation(s) would require detailed engineering design and testing. It’s also important to consider safety, environmental impact, and legal regulations when designing such a system.

The at least one component, feature and/or system(s) as outlined in paragraph 8 and fig. 3 and fig. 4, based on discussions in paragraphs 6 and 7, could (also) be totally or partially integrated into existing deep-sea mining vehicles (10) to execute algorithms for the selection and/or separation, but not limited to, of manganese crusts, nodules or sulphide deposits based on electric or acoustic parameters. These systems (or harvesting machines) (10) could then be capable of distinguishing these materials, partially or fully, ref. discussion under listing (j) paragraph 4.1, from other matter, thus facilitating more profitable, efficient, environmentally friendly and targeted mining operations.

Examples of such mining machines (10) are discussed in paragraph 3 (Prior Art).

9. A Favorable Total Solutions and/or Systems

A favorable total, independent and/or add-on system for offshore mineral mining could comprise at least one of the following components and/or features:

At least one electric and/or acoustic sensor (fig. 1; 1, a, b, fig.2; 1, a, b) that senses parameters of an area of interest on a seabed to identify mineral deposits and generates data signals based on the sensed parameters;

At least one processor (fig.1; 2, fig.2; 2) that processes the signals and determines locations on the seabed where economically sound concentrations, determined using quantifiable thresholds of mineral deposits, are identified, wherein the processor utilizes signal processing to identify the economic sound concentrations of the mineral deposits based on the data signals; and

a collector (fig.1; 4, fig.2; 4) that selectively collects the identified mineral deposits based on the identified locations determined by the processor. The collector can, but not necessarily, comprise an at least one degree of freedom of any combination of, but not limited to, a robotic arm, a trunk, a claw, a cutting device or a suction-based system and a storage unit (fig.1; 8, fig.2; 8) to hold the collected minerals,

and/or

a communication device (fig.3; 5, fig.4; 5) which is in communication with at least one external vessel (fig.3; 10, fig.4; 10).

The collector (fig.1; 4, fig. 2; 4) can be built-in within the total system (fig.1).

Wherein the at least one electric and/or acoustic sensor (fig.1; 1, a, b, fig.2; 1, a, b) can measure at least one of, but are not limited to, electric conductivity, resistivity, dielectric properties, centroid frequency, ultrasonic characteristics, sounds, velocities, density, acoustic impedance, attenuation, of seabed material to identify mineral deposits.

Wherein the at least one sensor (fig. 1; 7, fig. 2; 7 and/or fig.3;7, fig.4;7) includes an acoustic sensor that senses echo data from the seabed to identify mineral deposits.

Wherein the mineral deposits to be identified and can be separated (fig. 1; 4), but not limited to, manganese crusts, nodules or sulphide deposits from other matter based on electric and/or acoustic parameters (fig.1; 1, 2, 4, fig.2; 1, 2, 4).

The system further comprising a vehicle that is maneuverable (fig.1; 3) in water and/or over the seabed or a vehicle that is maneuverable (fig.3; 3) in water and/or over the seabed.

The system wherein a vehicle (fig. 1) includes a storage unit (fig. 1; 8) that holds the mineral deposits collected by the collector (fig 1; 4, fig 2; 4).

The system wherein a vehicle (fig. 1) include(s) a communication device (fig. 1; 5, fig. 2; 5) and/or the vehicle (fig.3) include(s) a communication device (fig.3; 5, fig. 4; 5) configured to transmit and receive information with a master control station and/or at least one external vessel (fig. 3; 10, fig.4; 10).

The system wherein a vehicle (fig. 1) include(s) at least one environmental monitoring sensor (fig. 1; 7) to ensure compliance with regulations and to minimize environmental impact by the collector (fig 1; 4, fig 2; 4).

The system of wherein a separator (fig. 1; 4, fig. 2; 4), based on output data from the at least one processor (2), that separates and/or collects the identified mineral deposits from other matter that is identified on the seabed, to differentiate and/or collect the identified mineral deposits from the other matter, wherein the separator is operated by the processor (fig.1; 2, fig.

2; 2) to differentiate and/or collect the identified economically sound concentrations of mineral deposits from other matter.

Additional sensors (ref. paragraph 6.1) can be located on and/or within a vehicle (fig.1, fig.3) (not shown).

A method for offshore mineral mining, comprising at least one of the following components and/or features:

At least one processor (fig.1; 2, fig.2; 2) that processes the signals and determines locations on the seabed where economically sound concentrations, determined using quantifiable thresholds of mineral deposits are identified, wherein the processor utilizes signal processing to identify the economic sound concentrations of the mineral deposits based on the data signals; and

a collector (fig.1; 4, fig.1; 4) that selectively collects the identified mineral deposits based on the identified locations determined by the processor. The collector can, but not necessarily, comprise an at least one degree of freedom of any combinations of, but not limited to, a robotic arm, a trunk, a claw, a cutting device or a suction-based system and a storage unit (fig.1; 8, fig.2; 8) to hold the collected minerals,

and/or a communication device (fig.3; 5, fig.4; 5) which is in communication with at least one external vessel (fig.3, 10, fig. 4, 10).

The method further comprising a vehicle that is maneuverable (fig.1; 3) in water and/or over the seabed or a vehicle that is maneuverable (fig.3; 3) in water and/or over the seabed.

The method wherein the mineral deposits to be identified and can be separated, but not limited to, manganese crusts, nodules or sulphide deposits from other matter based on electric and/or acoustic parameters (fig.1; 1, 2, 4, 7, fig.2; 1, 2, 4, 7).

A non-transitory memory storing a computer (fig.1; 2, fig.2; 2 and/or fig.3; 2, fig.4; 2) providing an executable program for supporting offshore mineral mining, the program causing at least one processor (fig.1; 2, fig. 2; 2) and/or (fig.3; 2, fig.4; 2) to perform the steps of:

receiving data (fig. 1; 1, fig.2; 1) and/or data (fig.3; 1, fig.4; 1) related to parameters of an area of interest on a seabed identifying mineral deposits;

and/or

receiving data (fig. 1; 7, fig.2; 7 and/or data fig.3; 7, fig.4; 7) related to parameters of an area of interest on a seabed identifying mineral deposits;

processing the data (fig.1; 2, fig.2; 2 and/or fig.3; 2, fig. 4; 2) to determine locations on the seabed where economic sound concentrations of mineral deposits are identified,

the processing utilizing signal processing, to identify the economic sound concentrations of mineral deposits based on the data signals; and

instructing a collector (fig.1; 4, fig.1; 4) that selectively collects the identified mineral deposits based on the identified locations determined by the processor. The collector can, but not necessarily, comprise at least one degree of freedom of any combinations of, but not limited to, a robotic arm, a trunk, a claw, a cutting device or a suction-based system and a storage unit (fig.1; 8, fig.2; 8) to hold the collected minerals,

and/or a communication device (fig.3; 5, fig.4; 5) which is in communication with at least one user and/or external vessel (fig. 3, 10, fig.4, 10).

The system and/or the method wherein the at least one processor (fig.1; 2, fig.2; 2) and/or (fig. 3; 2, fig.4; 2) can include at least one self-correcting algorithm.

The system and/or method wherein the at least one processor (fig.1; 2, fig.2; 2 and/or fig.3; 2, fig. 4; 2) includes at least one Learning (ML) and/or AI algorithm to make high-level navigation and collection strategy decisions and constitute a mapping and data collection framework to maintain historical information and optimize future operations, characterized by at least one of:

- Simulation System

- Mapping and Data Collection System

- Sensor Data Processing System

- DQL Navigation and Decision Agent

- Classification and Prediction System

- System Integration Architecture

- Self-Correction and Learning System

- Performance and Safety System

- Economic Optimization System

- Environmental Protection System.

The system and/or the method wherein the at least one vessel (fig.1, fig.2 and/or fig. 3, fig.4) can include a power unit (6) and/or a locking device (9).

A method:

- sensing, using at least one electric and/or acoustic sensor (fig.1; 1, a, b, fig. 2; 1, a, b), parameters of an area of interest on a seabed to identify mineral deposits and generating data signals based on the sensed parameters;

- processing, using at least one processor (fig.1; 2, fig.2; 2) the data signals and determining locations on the seabed where economically sound concentrations, determined using quantifiable thresholds of mineral deposits are identified, wherein the processor utilizes signal processing to identify the economic sound concentrations of the mineral deposits based on the data signals; and

selectively collecting, using a collector (fig.1; 4, fig. 1; 4), the identified mineral deposits based on the identified locations determined by the processor.”

A non-transitory memory storing a computer (fig.1; 2, fig.2; 2 and/or fig.3; 2, fig. 4; 2) providing an executable program for supporting offshore mineral mining, the program causing at least one processor (fig.1; 2, fig. 2; 2 and/or fig.3; 2, fig.4; 2) to perform the steps of, but not limited to, execute and/or control at least one of the following tasks, components or functionalities;

- Data Collection: Gather acoustic and/or electric parameter data from an area of interest. This could include e.g. sound speed or velocity, density, acoustic impedance, and attenuation for acoustic parameters, and resistivity or conductivity for electric parameters.

- Data Preprocessing:

- Filter the acoustic data to remove noise and enhance signal quality.

- Normalize electric sensor data to account for variations in sensor performance and environmental factors.

- Feature Extraction: Process the collected data to extract meaningful features that can, but not limited to, differentiate between manganese crusts, nodules and/or sulphide deposits, and other matter. Features could include statistical descriptors of the signal(s), spectral content, and derived quantities like reflection coefficients.

- Data Labeling: Data needs to be classified into categories (e.g., manganese crusts, nodules, sulphide deposits, other matter) based on ground truth information obtained from e.g. samples or previous surveys.

- Signal Processing: Use the labeled dataset to train a supervised model such as e.g. a Convolutional Neural Network (CNN) model, which can recognize patterns in the feature set that are indicative of different materials.

- Real Time Classification: Apply the trained model to new data to classify it into categories. Train the model e.g. on labeled datasets where the ground truth (manganese crusts, nodules, sulphide deposits, or other matter) is known. The model can output probabilities for each category, and the highest probability can determine the classification.

- Separation Decision/Threshold: Incorporate at least a set of threshold probabilities for the model's output to decide whether the material is, but not limited to, a manganese crust, nodules or sulphide deposit of economic value or meets the at least one threshold.

- Post-Processing: Apply additional rules or filters based on domain knowledge to refine the classification results. For example, geological knowledge about the distribution of these deposits can help in validating the model’s output.

- Verification and Self-Correction: Periodically collect samples of the separated material and analyze them to verify the accuracy of the separation process. Adjust the at least one model and threshold settings based on verification results to improve accuracy.

a. Acoustic Sensors: Could also be applied to collect high-resolution echo data from the seabed, which could also help in identifying the density and composition of the various materials.

b. Electric Sensors: To measure e.g. the electrical conductivity and/or resistivity of the seabed material, as different minerals have distinct electrical properties.

c) Processing Unit with Key Functionalities: This could be the one computer in question and/or added microcontroller that runs the at least one separation algorithm. It would take the sensor data as input and output commands to the separation mechanism (4).

d) Data Acquisition Sub-System: This sub-system would collect the data from the sensors (1, 7) and convert it into a format that can be processed. This often involves analogue-to-digital conversion, but not necessary.

e) Wavelet-Based Algorithm: Improves signal quality by reducing system interference and ambient noise. It may include a wavelet-based envelope extraction method, and an adaptive estimation strategy based on dual-channel information.

g) User Interface: This would allow the user to start and stop the process, adjust parameters, and monitor the system’s status. The user can be located top-side (manned or un-manned surface vessel) and/or be a submerged manned or un-manned machine and/or a vehicle (10).

APPENDIX

Detailed assessment of primary components and steps within a suggested ML/AI system related to:

6.2.1.1 Development of ML/AI Components to Fulfill the Requirements for “Systems for Optimal Offshore Mineral Mining” – if Applied

1. Simulation Environment

1.1 Physics Modeling

The simulation incorporates key physical aspects of the deep-sea environment:

● Seabed terrain generation using procedural methods with configurable parameters for different geological formations

● Water dynamics including currents, pressure variations, and turbulence

● Physical interactions between collection mechanisms and seabed materials

● Vehicle dynamics modeling accounting for buoyancy, drag, and thrust

● Energy consumption models for different operations (movement, scanning, collection) 1.2 Sensor Response Simulation

Accurate modeling of sensor responses is crucial for developing robust detection algorithms:

● Acoustic sensor simulation incorporating:

○ Sound propagation in water

○ Material-specific acoustic properties (as specified in paragraph 5.2)

○ Noise and interference patterns

○ Resolution degradation with distance

● Electrical sensor simulation including:

○ Conductivity and resistivity responses

○ Environmental interference

○ Sensor range limitations

○ Signal attenuation through different materials

1.3 Environmental Variations

The simulation includes dynamic environmental conditions to ensure system robustness:

● Water current patterns at different depths

● Visibility variations affecting sensor performance

● Temperature gradients and their effect on sensor readings

● Seabed composition variations

● Seasonal and weather-related changes

1.4 Mineral Deposit Scenarios

A variety of test scenarios are implemented to validate system performance:

● Different types of mineral deposits (manganese crusts, nodules, sulphide deposits)

● Varying concentrations and distributions

● Mixed deposits with different economic values

● Challenging scenarios with partial sensor occlusion

● Edge cases for system testing

1.5 Training Data Generation

The simulation environment serves as a primary source of synthetic training data, generating extensive datasets for training both the sensory prediction systems and the DQL agent:

1.5.1 Large-scale Synthetic Dataset Creation

● Automated generation of diverse scenarios:

○ Random placement of mineral deposits with varying compositions

○ Random sampling positions and orientations

○ Multiple sensor readings from different distances and angles

○ Generation of thousands of unique configurations

● Synthetic sensor response generation:

○ Simulated acoustic and electrical readings for each scenario

○ Multiple readings per position to account for noise

○ Varying environmental conditions for each reading

1.5.2 Data Augmentation and Variation

● Introduction of realistic noise patterns

● Simulation of sensor degradation and malfunction

● Environmental interference patterns

● Partial occlusion scenarios

● Different seabed topology configurations

1.5.3 Ground Truth Labeling

● Automatic annotation of:

○ Mineral type and composition

○ Deposit size and distribution

○ Economic value calculations

○ Environmental sensitivity metrics

○ Optimal collection strategies

1.5.4 Validation Framework

● Separate validation scenarios

● Performance benchmarks

● Edge case testing

● System stress testing

● Cross-validation datasets

2. Mapping & Data Collection System

2.1 Real-time 3D Mapping

● Dynamic map generation and updating:

○ High-resolution bathymetric data

○ Integration of multiple sensor inputs

○ Real-time topography updates

○ Obstacle and hazard identification

○ Confidence metrics for map accuracy

2.2 Data Storage and Organization

● Hierarchical data structure for efficient access:

○ Raw sensor readings (acoustic and electrical)

○ Processed sensor data

○ Mineral classification results

○ Collection operation outcomes

○ Environmental impact assessments

● Time-stamped historical records

● Spatial indexing for quick retrieval

● Compression strategies for long-term storage

2.3 Mineral Deposit Tracking

● Continuous updating of identified deposits:

○ Location and extent

○ Composition predictions

○ Confidence levels

○ Economic value estimates

○ Collection status

● Historical mining activity records

● Success/failure outcome logging

2.4 Environmental Monitoring

● Tracking of seabed disturbance

● Cumulative impact assessment

● Recovery monitoring of previously mined areas

● Identification of sensitive areas

● Long-term change detection

2.5 Integration with Decision Systems

● State space representation for DQL agent:

○ Local area characteristics

○ Global position context

○ Historical performance data

○ Resource availability

● Performance feedback loops:

○ Prediction accuracy tracking

○ Collection success rates

○ Economic outcome validation

○ Environmental impact verification

3. Sensor Data Processing

3.1 Traditional Signal Processing Pipeline

● Initial data cleaning:

○ Noise filtering

○ Signal normalization

○ Outlier detection

○ Missing data handling

● Feature extraction:

○ Wavelet-based decomposition

○ Frequency domain analysis

○ Statistical features

○ Signal correlation metrics

● Baseline classification rules:

○ Threshold-based detection

○ Pattern matching

○ Signal characteristic analysis

3.2 Deep Learning Enhancement

● Multi-modal neural network architecture:

○ Acoustic signal processing branch

○ Electrical signal processing branch

○ Feature fusion layers

○ Classification/prediction outputs

● Training methodology:

○ Initial training on simulation data

○ Fine-tuning with real operational data

○ Continuous learning from new observations

○ Validation against traditional methods

3.3 Sensor Fusion

● Integration of multiple sensor types:

○ Acoustic sensor arrays

○ Electrical conductivity sensors

○ Environmental sensor data

● Multi-scale analysis:

○ Broad area scanning

○ Detailed local inspection

○ Cross-validation between sensors

● Confidence scoring system

3.4 Performance Monitoring

● Real-time quality assessment:

○ Signal quality metrics

○ Prediction confidence scores

○ Cross-validation results

● Adaptive processing:

○ Environmental condition compensation

○ Sensor degradation correction

○ Dynamic noise filtering

● Performance optimization:

○ Processing speed vs accuracy trade-offs

○ Resource usage optimization

○ Real-time requirements compliance

3.5 Integration Points

● Input to Classification System:

○ Processed sensor features

○ Confidence metrics

○ Environmental context

● Input to DQL Agent:

○ Current sensor status

○ Processing results

○ Quality metrics

● Feedback mechanisms:

○ Performance logging

○ Error correction

○ Model updating

4. DQL Navigation & Decision Agent

4.1 State Space Definition

The agent operates in a rich state space encompassing:

● Position Information:

○ Current 3D coordinates and orientation

○ Velocity and acceleration

○ Distance to known deposits

○ Distance to unexplored areas

● Resource Status:

○ Current energy levels

○ Energy consumption rate

○ Storage capacity remaining

○ Expected operation time

● Environmental Data:

○ Local scan data from sensors

○ Seabed topology

○ Environmental sensitivity metrics

○ Current conditions (currents, visibility) ● Historical Context:

○ Map data of explored areas

○ Previous collection successes/failures ○ Known deposit distributions ○ Area economic performance history ● Economic Parameters:

○ Current mineral value thresholds

○ Operating cost metrics

○ Collection efficiency metrics ○ Risk assessment scores

4.2 Action Space

The agent can execute multiple types of actions:

● Navigation Actions:

○ 3D movement commands

○ Speed and direction control

○ Depth adjustments

○ Station-keeping

● Scanning Operations:

○ Scan resolution selection

○ Sensor type selection

○ Area coverage patterns

○ Detailed inspection triggers

● Collection Decisions:

○ Initiate/abort collection

○ Collection method selection

○ Collection intensity control

○ Storage management

4.3 Reward Function Design

The reward function balances multiple objectives:

● Economic Components:

○ Mineral value collected

○ Operating costs (energy, time)

○ Equipment wear

○ Transportation costs

● Environmental Factors:

○ Seabed disruption metrics

○ Ecosystem impact scores

○ Recovery time estimates

○ Cumulative impact assessment

● Exploration Value:

○ New area discovery bonus

○ Information gain metrics

○ Uncertainty reduction value

○ Pattern recognition rewards

4.4 Training and Optimization

● Multi-stage Training Process:

○ Initial simulation training

○ Supervised learning from expert demonstrations

○ Real-world fine-tuning

○ Continuous learning during operation

● Safety Constraints:

○ Action space limitations

○ Risk-aware exploration

○ Emergency protocol integration

○ Environmental protection rules

4.5 Performance Monitoring

● Real-time Metrics:

○ Decision quality assessment

○ Economic performance tracking

○ Environmental impact monitoring

○ Exploration efficiency metrics

● Adaptation Mechanisms:

○ Policy updates based on outcomes

○ Dynamic threshold adjustment

○ Learning rate optimization

○ Exploration strategy adaptation

5. Classification & Prediction System

5.1 Mineral Classification

● Multi-class Classification System:

○ Mineral type identification

○ Composition analysis

○ Purity assessment

○ Deposit structure classification

● Feature Processing:

○ Acoustic signature analysis

○ Electrical conductivity patterns

○ Multi-sensor data fusion

○ Contextual feature extraction

● Classification Models:

○ Primary deep learning classifier

○ Ensemble of specialized models

○ Traditional backup classifiers

○ Real-time model selection

5.2 Quantity Prediction

● Deposit Size Estimation:

○ Volume calculation

○ Density prediction

○ Distribution mapping

○ Continuity assessment

● Resource Modeling:

o 3D deposit modeling

o Concentration gradients

o Structural analysis

o Geological context integration • Prediction Updates:

o Real-time refinement

o Multi-scale analysis

o Historical pattern matching o Error correction

conomic Value Estimation

• Value Calculation:

o Current market prices

o Extraction difficulty factors o Processing requirements

o Transportation costs

• Risk Assessment:

o Quality uncertainty

o Extraction challenges

o Environmental considerations o Operational risks

• ROI Prediction:

o Cost-benefit analysis

o Operation time estimation

o Resource requirement calculation o Value optimization strategies

Uncertainty Quantification

• Probabilistic Modeling:

o Confidence intervals

o Error bounds

o Variance estimation

o Risk quantification

• Multi-source Validation:

o Cross-sensor verification

o Historical data comparison

o Model ensemble agreement o Environmental factor impact

Confidence Scoring

• Comprehensive Scoring System:

o Classification confidence

o Quantity prediction reliability o Value estimation accuracy

o Environmental impact certainty • Decision Support:

o Confidence thresholds

o Risk-aware decision making o Uncertainty-based planning o Resource allocation optimization • Performance Tracking:

o Prediction accuracy monitoring ○ Error pattern analysis

○ Model calibration metrics

○ Continuous improvement feedback

5.6 Integration and Updates

● System Integration:

○ Real-time data pipeline

○ DQL agent feedback

○ Mapping system updates

○ Operation planning input

● Model Updates:

○ Online learning implementation

○ Performance-based adaptation

○ New pattern incorporation

○ Concept drift handling

6. System Integration Architecture

The integration architecture ensures seamless operation between all AI components and physical systems, providing robust, real-time performance while maintaining safety and reliability.

6.1 Data Flow Architecture

● Real-time Processing Pipeline:

○ Sensor data ingestion and distribution

○ Priority-based processing queues

○ Real-time performance monitoring

○ Load balancing and scaling

● Inter-system Communication:

○ Message passing protocols

○ Data serialization standards

○ State synchronization

○ Event handling system

● Storage Architecture:

○ Real-time operational data

○ Historical data management

○ Model storage and versioning

○ Performance metrics logging

6.2 Hardware Integration

● Sensor Systems Interface:

○ Raw data acquisition

○ Calibration management

○ Error detection

○ Health monitoring

● Collection Mechanism Control:

○ Command translation

○ Feedback processing

○ Safety constraint enforcement

○ Emergency override systems

● Physical System Monitoring:

○ Component status tracking

○ Performance metrics

○ Maintenance predictions

○ Failure detection

6.3 Operational Control Flow

● Decision Execution Pipeline:

○ DQL agent command processing

○ Safety validation

○ Resource allocation

○ Action coordination

● Error Handling:

○ Fault detection

○ Graceful degradation

○ Recovery procedures

○ Backup systems activation

● Performance Optimization:

○ Resource utilization monitoring

○ Processing latency management

○ System capacity optimization

○ Quality of service maintenance

7. Self-Correction & Learning

The self-correction and learning systems enable continuous improvement of all AI components through operational experience, ensuring optimal performance and reliability over time.

7.1 Continuous Learning Framework

● Model Update System:

○ Performance monitoring

○ Training data collection

○ Model retraining triggers

○ Validation procedures

● Adaptation Mechanisms:

○ Parameter adjustment

○ Feature importance updates

○ Decision threshold optimization

○ Policy refinement

7.2 Error Detection and Correction

● Performance Monitoring:

○ Prediction accuracy tracking

○ Decision quality assessment

○ Resource efficiency metrics

○ Environmental impact monitoring

● Error Response:

○ Root cause analysis

○ Correction strategy selection

○ Implementation verification

○ Impact assessment

7.3 Knowledge Integration

● Experience Collection:

○ Successful operation patterns

○ Failure case documentation

○ Environmental condition effects

○ Economic outcome correlation

● System Updates:

○ Model retraining

○ Policy updates

○ Parameter optimization

○ Rule refinement

7.4 Safety and Reliability

● Performance Boundaries:

○ Operating limits definition

○ Risk assessment

○ Safety constraint enforcement

○ Emergency response triggers

● Validation Framework:

○ Update Verification

○ Performance regression testing

○ Safety compliance checking

○ Reliability assessment

7.5 Long-term Optimization

● Strategic Improvement:

○ Performance trend analysis

○ Resource optimization

○ Cost efficiency enhancement

○ Environmental impact reduction

● System Evolution:

○ Capability expansion

○ Algorithm refinement

○ Feature enhancement

○ Integration optimization

8. Performance & Safety Systems

This system provides comprehensive monitoring, evaluation, and safety mechanisms, ensuring reliable operation while maintaining clear performance metrics and robust fallback procedures.

8.1 Performance Metrics Framework

● Economic Performance:

○ Cost per unit mineral collected

○ Energy efficiency metrics

○ Time-to-collection ratios

○ ROI calculations

○ Operation optimization scores

● Environmental Monitoring:

○ Seabed disruption measurements

○ Area impact assessments

○ Recovery time predictions

○ Cumulative effect tracking

● Prediction Performance:

o Classification accuracy

o Quantity estimation error rates o Value prediction accuracy

o Confidence score reliability

• System Health Metrics:

o Component status tracking

o Performance degradation detection o Resource utilization monitoring o Operation efficiency scores

8.2 Fallback Systems

• Traditional Algorithmic Backups:

o Rule-based navigation systems o Basic mineral detection algorithms o Simple collection strategies

o Manual control capabilities

• Graceful Degradation:

o Performance level transitions o Functionality prioritization

o Resource reallocation

o Minimal operation modes

8.3 Emergency Response System

• Emergency Protocols:

o System shutdown procedures o Safe state transitions

o Resource conservation modes o Communication fallbacks

• Crisis Management:

o Damage mitigation procedures o Recovery protocols

o Resource preservation

o Environmental protection measures

8.4 Fault Tolerance Architecture

• Redundancy Systems:

o Component redundancy

o Data backup mechanisms

o Processing failover

o Communication redundancy

• Error Handling:

o Fault detection algorithms

o Error containment strategies

o Recovery procedures

o System restoration protocols

8.5 Safety Enforcement

• Operational Boundaries:

o Movement restrictions

o Collection limits

o Environmental thresholds ○ Resource constraints

● Safety Protocols:

○ Collision avoidance

○ Environmental protection rules

○ Equipment protection

○ Emergency responses

● Compliance Monitoring:

○ Regulatory requirement tracking

○ Safety standard adherence

○ Environmental compliance

○ Performance standard maintenance

8.6 System Validation

● Regular Testing:

○ Component testing

○ Integration testing

○ Performance validation

○ Safety system verification

● Documentation:

○ Performance reports

○ Incident logging

○ Improvement tracking

○ Compliance documentation

9. Economic Optimization Systems

This section details the economic decision-making capabilities of the AI system, providing comprehensive specifications for market adaptation, cost optimization, and value maximization strategies.

9.1 Market Intelligence System

9.1.1 Real-time Market Data Integration

● Commodity Price Monitoring:

○ Direct feeds from major mineral exchanges

○ Historical price pattern analysis

○ Forward contract pricing

○ Spot market tracking

○ Price volatility assessment

● Global Market Analysis:

○ Supply-demand equilibrium tracking

○ Regional market differentials

○ Currency exchange impact

○ Transportation cost indices

○ Competitive activity monitoring

9.1.2 Predictive Market Modeling

● Short-term Price Predictions:

○ 24-hour price movement forecasting

○ Weekly trend analysis

○ Volatility prediction

○ Market sentiment integration

o Risk factor assessment

• Long-term Market Trends:

o Seasonal pattern recognition

o Multi-year cycle analysis o Technology impact assessment o Regulatory change predictions o Industry development tracking

9.2 Operational Cost Optimization

9.2. 1 Dynamic Cost Modeling

• Direct Operational Costs:

o Energy consumption per ton

o Equipment wear rates

o Maintenance scheduling o Labor cost optimization o Consumables usage tracking

• Indirect Cost Factors:

o Weather impact assessment

o Depth-based cost variations

o Equipment eficiency decay

o Support service optimization

o Insurance cost management

9.2.2 Resource Allocation Optimization

• Equipment Utilization:

o Operating hours optimization o Maintenance timing

o Capacity utilization rates o Performance degradation tracking o Replacement timing optimization • Personnel Resource Management:

o Shift optimization

o Skill requirement planning

o Training need prediction o Safety cost assessment o Productivity optimization

9.3 Value Maximization Framework

9.3.1 Grade-Based Optimization

• Mineral Grade Assessment:

o Real-time grade analysis o Grade distribution mapping

o Processing requirement assessment o Recovery rate optimization

o Blend optimization strategies

• Value-based Selection:

o Grade cutof optimization

o Processing method selection

o Recovery method optimization o Storage strategy optimization o Transport timing optimization

9.3.2 Collection Strategy Optimization

• Collection Planning:

o Optimal collection sequence o Equipment selection logic

o Collection rate optimization o Storage capacity management o Processing queue optimization • Operational Timing:

o Market-based timing strategies o Weather window optimization o Equipment availability matching o Resource availability alignment o Maintenance schedule integration

9.4 DQL Agent Economic Integration

9.4.1 Economic State Space

• Market Conditions:

o Current prices and trends

o Market volatility indicators o Supply-demand metrics

o Competition indicators

o Future price predictions

• Operational Status:

o Current costs and eficiency o Resource quality metrics

o Equipment status

o Personnel availability o Storage capacity status

9.4.2 Economic Action Space

• Collection Decisions:

o Target selection criteria

o Collection intensity control o Equipment configuration

o Processing method selection o Storage allocation

• Timing Optimization:

o Operation scheduling o Market timing strategies

o Maintenance planning o Resource allocation timing

o Transport coordination

9.5 Adaptive Economic Strategy System

9.5.1 Short-term Optimization

• Real-time Adjustments:

o Collection rate modification o Grade cutof adaptation o Equipment utilization adjustment o Resource reallocation o Cost control measures

• Daily Operations:

o Shift planning optimization

o Equipment deployment o Maintenance scheduling

o Resource utilization

o Performance monitoring

9.5.2 Long-term Strategy

• Strategic Planning:

o Equipment lifecycle management o Technology upgrade planning o Market position optimization o Resource development planning o Capability enhancement

• Risk Management:

o Market risk assessment o Operational risk evaluation

o Technology risk management o Regulatory compliance planning o Environmental risk consideration

9.6 Performance Metrics and Reporting

9.6.1 Economic Performance Metrics

• Financial Metrics:

o Return on investment (ROI) o Operating margin analysis

o Cost per ton metrics

o Revenue optimization o Profit maximization

• Operational Metrics:

o Equipment efficiency o Resource utilization

o Processing effectiveness

o Collection rate optimization o Quality management

9.6.2 Reporting and Analysis

• Performance Reporting:

o Real-time dashboards o Trend analysis

o Variance reporting

o Predictive analytics

o Strategic recommendations

• Continuous Improvement:

o Performance optimization

o Cost reduction initiatives

o Efficiency improvements

o Process optimization

○ Technology integration

10. Environmental Protection Systems

10.1 Environmental Impact Assessment

10.1.1 Physical Impact Monitoring

● Seabed Disturbance Metrics:

○ Topographical change measurement (±0.1mm precision)

○ Sediment dispersion tracking (particle size 1-1000μm)

○ Substrate composition analysis

○ Structural integrity assessment

○ Recovery rate monitoring

● Water Column Effects:

○ Turbidity measurement (0-1000 NTU range)

○ Chemical composition analysis

○ Temperature variation (±0.1°C precision)

○ Current pattern disruption

○ Sediment plume tracking

10.1.2 Ecosystem Impact Assessment

● Biodiversity Monitoring:

○ Species population tracking

○ Habitat distribution mapping

○ Community structure analysis

○ Behavioral pattern monitoring

○ Migration pattern tracking

● Ecosystem Health Indicators:

○ Primary productivity measurement

○ Food web interaction analysis

○ Species diversity indices

○ Population density tracking

○ Ecosystem resilience assessment

10.2 Preventive Protection Framework

10.2.1 Pre-operation Assessment

● Habitat Mapping:

○ High-resolution 3D mapping

○ Sensitive area identification

○ Species distribution analysis

○ Ecological corridor mapping

○ Seasonal variation assessment

• Risk Assessment:

o Vulnerability analysis

o Impact prediction modeling

o Recovery potential evaluation o Cumulative efect assessment o Mitigation strategy development

10.2.2 Operational Controls

• Activity Limitations:

o Speed restrictions

o Collection depth control

o Equipment pressure limits

o Noise level management

o Vibration control

• Protection Zones:

o Dynamic exclusion areas

o Bufer zone management

o Seasonal restriction zones

o Recovery area protection

o Migration corridor preservation

10.3 Real-time Monitoring Systems

10.3.1 Continuous Environmental Monitoring

• Physical Parameters:

o Temperature profiles

o Pressure variations

o Current measurements

o Turbidity monitoring

o Chemical composition

• Biological Indicators:

o Species presence detection

o Behavioral monitoring

o Population movement tracking o Interaction pattern analysis

o Stress response indicators

10.3.2 Adaptive Response System

• Threshold Management:

o Impact limit monitoring

o Warning level triggers

o Emergency shutdown criteria o Recovery period requirements o Cumulative impact limits

• Response Actions:

o Activity modification protocols o Equipment adjustment procedures o Operation suspension criteria o Alternative method selection o Impact mitigation measures

10.4 Al-driven Protection Mechanisms 10.4. 1 Neural Network Environmental Analysis

• Pattern Recognition:

o Ecosystem change detection

o Behavior pattern analysis

o Impact progression tracking

o Recovery trend identification

o Risk pattern recognition

• Predictive Modeling:

o Impact prediction

o Recovery time estimation

o Risk level assessment

o Cumulative efect projection

o Mitigation efectiveness prediction

10.4.2 Protective Decision Making

• Operation Modification:

o Collection method adaptation o Route optimization

o Timing adjustment

o Equipment configuration

o Resource selection

• Impact Minimization:

o Sediment control strategies

o Noise reduction methods

o Vibration dampening

o Chemical exposure limitation o Habitat preservation techniques

10.5 Recovery Management System

10.5.1 Post-operation Monitoring

• Recovery Tracking:

o Seabed regeneration monitoring o Species recolonization assessment o Ecosystem function recovery

o Habitat restoration progress

o Water quality normalization

• Impact Assessment:

o Long-term efect monitoring

o Recovery rate analysis

o Ecosystem stability evaluation o Biodiversity recovery tracking o Function restoration assessment

10.5.2 Adaptive Management

• Recovery Support:

o Natural recovery promotion

o Assisted recovery methods

o Habitat enhancement strategies o Species support measures

o Ecosystem stabilization techniques ● Long-term Planning:

○ Recovery goal setting

○ Progress monitoring systems

○ Intervention timing optimization

○ Resource allocation planning

○ Stakeholder engagement

10.6 Integration and Reporting

10.6.1 Environmental Performance Metrics

● Impact Metrics:

○ Disturbance quantification

○ Recovery rate measurement

○ Biodiversity indices

○ Ecosystem health scores

○ Cumulative impact assessment

● Performance Indicators:

○ Protection effectiveness

○ Recovery success rates

○ Mitigation efficiency

○ Compliance levels

○ Sustainability scores

10.6.2 Documentation and Reporting

● Compliance Documentation:

○ Regulatory requirement tracking

○ Performance standard verification

○ Impact assessment reports

○ Mitigation effectiveness

○ Recovery progress documentation

● Stakeholder Communication:

○ Performance dashboards

○ Progress reports

○ Impact assessments

○ Recovery updates

○ Compliance verification

Claims

CLAIMS

1.A system for offshore mineral mining, comprising:

At least one electric and/or acoustic sensor (fig. 1; 1 a, b, fig.2; 1, a, b) that senses parameters of an area of interest on a seabed to identify mineral deposits and generates data signals based on the sensed parameters;

a collector (fig.1; 4, fig.2; 4) that selectively collects the identified mineral deposits based on the identified locations determined by the processor. The collector can, but not necessarily, comprise an at least one degree of freedom of any combinations of, but not limited to, a robotic arm, a trunk, a claw a cutting device or a suction-based system and a storage unit (fig.1; 8, fig.2; 8) to hold the collected minerals,

and/or

a communication device (fig.3; 5, fig. 4; 5) which is in communication with at least one external vessel (fig.3; 10, fig.4; 10).

2.The system of claim 1, wherein the at least one electric and/or acoustic sensor (fig.1; 1, a, b, fig.2; 1, a, b) can measure at least one of, but are not limited to, electric conductivity, resistivity, dielectric properties, centroid frequency, ultrasonic characteristics, sounds, velocities, density, acoustic impedance, attenuation, of seabed material to identify mineral deposits.

3.The system of claim 1, wherein the at least one sensor (fig.1; 7, fig. 2; 7 and/or fig.3;7, fig.

4;7) includes an acoustic sensor that senses echo data from the seabed to identify the mineral deposits.

4.The system of claim 1, wherein the mineral deposits to be identified and can be separated, but not limited to, manganese crusts, nodules or sulphide deposits from other matter based on electric and/or acoustic parameters (fig.1; 1, 2, 4, fig.2; 1, 2, 4).

5.The system of claim 1, further comprising a vehicle that is maneuverable (fig.1; 3) in water and/or over the seabed or a vehicle that is maneuverable (fig.3; 3) in water and/or over the seabed.

6.The system of claim 5, wherein the vehicle (fig.1) includes a storage unit (fig.1; 8) that holds the mineral deposits collected by the collector (fig 1; 4, fig 2; 4).

7.The system of claim 5, wherein a vehicle (fig.1) include(s) a communication device (fig. 1; 5, fig.2; 5) and/or a vehicle (fig.3) include(s) a communication device (fig.3; 5, fig.4; 5) configured to transmit and receive information with a master control station and/or at least one external vessel (fig.3; 10, fig.4; 10).

8.The system of claim 5, wherein a vehicle (fig.1) include(s) at least one environmental monitoring sensor (fig.1; 7) to ensure compliance with regulations and to minimize environmental impact by the collector (fig 1; 4, fig 2; 4).

9.The system of claim 1, wherein a separator (fig.1; 4, fig.2; 4), based on output data from the at least one processor (2), that separates and/or collects the identified mineral deposits from other matter that is identified on the seabed to differentiate and/or collect the identified mineral deposits from the other matter, wherein the separator is operated by the processor (fig.

1; 2, fig.2; 2) to differentiate and/or collect the identified economically sound concentrations of mineral deposits from other matter.

10.The system to all previous claims, wherein the at least one processor (fig.1; 2, fig.2; 2 and/or fig.3; 2, fig. 4; 2) can include at least one self-correcting algorithm.

11.The system to all previous claims, wherein the at least one processor (fig.1; 2, fig.2; 2 and/or fig.3; 2, fig. 4; 2) includes at least one Learning (ML) and/or AI algorithm to make high-level navigation and collection strategy decisions and constitute a mapping and data collection framework to maintain historical information and optimize future operations, characterized by at least one of:

- Simulation System

- Mapping and Data Collection System

- Sensor Data Processing System

- DQL Navigation and Decision Agent

- Classification and Prediction System

- System Integration Architecture

- Self-Correction and Learning System

- Performance and Safety System

- Economic Optimization System

- Environmental Protection System.

12.The system to all previous claims, wherein the at least one vessel (fig.1, fig.2 and/or fig. 3, fig. 4) can include a power unit (6) and/or a locking device (9).

13.A method for offshore mineral mining, comprising:

sensing, using at least one electric and/or acoustic sensor (fig. 1; 1, a, b, fig.2; 1, a, b), parameters of an area of interest on a seabed to identify mineral deposits and generating data signals based on the sensed parameters;

processing, using at least one processor (fig.1; 2, fig.2; 2) the data signals and determining locations on the seabed where economically sound concentrations, determined using quantifiable thresholds of mineral deposits are identified, wherein the processor utilizes signal processing to identify the economic sound concentrations of the mineral deposits based on the data signals; and

selectively collecting, using a collector (fig.1; 4, fig. 1; 4), the identified mineral deposits based on the identified locations determined by the processor,

and/or

a communication device (fig.3; 5, fig. 4; 5) which is in communication with at least one external vessel (fig.3, 10, fig.4, 10).

14. A non-transitory memory storing an executable program for supporting offshore mineral mining, the program causing at least one processor of a computer (fig.1; 2, fig.2; 2 and/or fig. 3; 2, fig.4; 2) to perform the steps of:

- receiving data signals (fig.1; 1, fig.2; 1 and/or fig. 3; 1, fig.4; 1 and/or fig.1; 7, fig.

2; 7 and/or fig.3; 7, fig. 4; 7) generated by at least one electric and/or acoustic sensor (fig.1; 1 a, b, fig.2; 1, a, b) and related to parameters of an area of interest on a seabed identifying mineral deposits;

- processing the data (fig.1; 2, fig.2; 2 and/or fig. 3; 2, fig. 4; 2) to determine locations on the seabed where economic sound concentrations of mineral deposits are identified,

- the processing utilizing signal processing, to identify the economic sound concentrations of mineral deposits based on the data signals; and

- instructing a collector (fig.1; 4, fig.1; 4) to selectively collect the identified mineral deposits based on the identified locations determined by the processor,

and/or

- a communication device (fig.3; 5, fig.4; 5) which is in communication with at least one user and/or external vessel (fig. 3, 10, fig.4, 10).

15. A non-transitory memory storing a computer (fig.1; 2, fig.2; 2) and/or (fig.3; 2, fig.4; 2) providing an executable program for supporting offshore mineral mining, the program causing at least one processor (fig.1; 2, fig. 2; 2) and/or (fig.3; 2, fig.4; 2) to perform the steps of claim 14, to execute and/or control at least one of the following tasks, components or functionalities;

- Data Collection: Gather acoustic and/or electric parameter data from the area of

interest.

- Data Preprocessing:

- Filter the acoustic data to remove noise and enhance signal quality.

- Normalize electric sensor data to account for variations in sensor performance and environmental factors.

- Data Labeling: Data to be classified into categories (e.g., manganese crusts, nodules, sulphide deposits, other matter) based on ground truth information obtained from e.g. samples or previous surveys.

- Signal Processing: Use the labeled dataset to train a supervised model, e.g. a Convolutional Neural Network (CNN) model, which can recognize patterns in the feature set that are indicative of different materials.

- Real Time Classification: Apply the trained model to new data to classify it into one of the categories. Train the model e.g. on labeled datasets where the ground truth (manganese crusts, nodules, sulphide deposits, or other matter) is known. The model will output probabilities for each category, and the highest probability can determine the classification.

- Separation Decision/Threshold: Incorporate at least, but not limited to, a set of threshold probabilities for the model's output to decide whether the material is a manganese crust, nodules or sulphide deposit of economic value or meets the at least one threshold.

- Post-Processing: Apply additional rules or filters based on domain knowledge to refine the classification results, e.g. geological knowledge about the distribution of these deposits can help in validating the model’s output.

a. Acoustic Sensors,

b. Electric Sensors,

c) A processing Unit with Key Functionalities to run at least one separation algorithm,

d) A data Acquisition Sub-System to convert collected data from the sensors and convert them into processable formats,

e) A wavelet-Based Algorithm, f) Wavelet Regression,

g) User Interface.

Drawings

Priority

06.05.2024 NO 20240441

Patent classes

IPC classes

E21C 50/00

B63C 11/52

E02F 7/00

E02F 9/20

G01V 3/38

B63G 8/00

G01N 1/02

CPC classes

E21C 50/00

B63C 11/52

E02F 7/00

E02F 9/2025

E02F 9/205

G01V 3/38

B63G 2008/002

G01N 2001/021

G01N 2223/616

Citations

EP 3719252 B1 (B1 )

WO 2023063831 A1 (A1)

US 2022187227 A1 (A1)

WO 2023183651 A1 (A1)

EP 4005917 A1 (A1)

Applicants, owners and inventors

Applicants

Modi Vivendi AS

Putten 50

1676 KRÅKERØY

NO (FREDRIKSTAD Municipality, Østfold county)

Org. number: 988979139

Inventors

Gunnar Myhr

Putten 50

1676 KRÅKERØY

NO (FREDRIKSTAD Municipality, Østfold county)

Agent

No agent

Status history

Payment history

Oppositions and appeals

No oppositions or appeals

Administrative limitation

No administrative limitation

Liens and licences

No liens or licenses registered

Gazette entries

No Gazette entries found

Only events from 2024.03.04 and onwards are shown.