Submersible robot system and method of operating the system

Abstract

In a first aspect, embodiments described herein relate to a submersible robotic system for inspecting and / or servicing an object (e.g., a naval vessel) in a marine environment. In some embodiments, the robotic system includes a housing, a plurality of suction mechanisms disposed within the housing, an illumination device, and an imaging device. In some applications, each suction mechanism includes a plurality of suction devices constructed and arranged to secure the system to the object, a magnetic switch motor for switching the suction devices on and off, at least one servicing element, and a body rotation motor for moving the servicing element over the surface of the object.

Description

[Technical Field]

[0001] (CROSS-REFERENCE TO RELATED APPLICATIONS) This application claims priority to and benefit of U.S. Provisional Application No. 63 / 435,957, filed December 29, 2022, which is incorporated herein by reference in its entirety.

[0002] The present disclosure relates generally to robotic systems and methods for inspecting and / or maintaining the surface of an object in an environment. More particularly, the present disclosure relates to robotic systems and methods for operating in hostile environments, such as underwater. [Background technology]

[0003] Biofouling refers to the fouling of submersible structures (e.g., ship hulls) and involves the primary deterioration of the submerged surface due to the accumulation of materials and organisms such as algae, plants, barnacles, microorganisms, and others on the submerged surface. A particular concern when treating submersible portions of marine and naval vessels is that the accumulation of such materials and organisms creates drag on the vessel's propulsion system. Specifically, the more severe the fouling, the higher the drag and, correspondingly, the larger the propulsion force that must be used simply to overcome the additional drag. The additional force required has a direct impact on fuel consumption, wear, maintenance, and monitoring of multiple mechanical subsystems that may not operate at optimal levels. Additional consequences of increased fuel consumption include increased pollution and associated greenhouse gas emissions.

[0004] Equally problematic is the growing number of regulatory actions addressing the introduction of invasive alien species into waterways and bodies of water. More specifically, in some instances, marine and naval vessels with excessive accumulations of materials and organisms (e.g., algae, plants, barnacles, microorganisms, etc.) on subsurface surfaces may be denied entry to ports of entry and / or fined due to the organisms being invasive and their potential impact on local ecosystems. Summary of the Invention [Means for solving the problem]

[0005] In a first aspect, embodiments described herein relate to a submersible robotic system for inspecting and / or servicing an object (e.g., a naval vessel) in a marine environment. In some embodiments, the robotic system includes a housing, a plurality of suction mechanisms disposed within the housing, an illumination device, and an imaging device. In some applications, each suction mechanism includes a plurality of suction devices constructed and arranged to secure the system to the object, a magnetic switch motor for switching the suction devices on and off, at least one servicing element, and a body rotation motor for moving the servicing element across the surface of the object.

[0006] In a second aspect, embodiments described herein relate to a method for inspecting and / or maintaining an object (e.g., a naval vessel) in a marine environment. In some embodiments, the method includes providing a robotic system for servicing a surface of the object and operating the robotic system to service the object. In some embodiments, the robotic system includes a housing, a plurality of suction mechanisms disposed within the housing, an illumination device, and an imaging device. In some applications, each suction mechanism includes a plurality of suction devices constructed and arranged to secure the system to the object, a magnetic switch motor for switching the suction devices on and off, at least one servicing element, and a body rotation motor for moving the servicing element across the surface of the object. [Effects of the Invention]

[0007] The present disclosure will be more fully appreciated in connection with the following detailed description of the embodiments taken in conjunction with the accompanying drawings. [Brief explanation of the drawings]

[0008] [Figure 1] 1 illustrates a top perspective view of the front of a system for inspecting and / or servicing an object in a marine environment, according to some embodiments of the disclosure. [Figure 2] 2 illustrates a bottom perspective view of the rear of the system shown in FIG. 1 according to some embodiments of the disclosure. [Figure 3A] 2 illustrates a first side view of the rear of the system shown in FIG. 1 according to some embodiments of the disclosure. [Figure 3B] 3B illustrates a second side view of the system shown in FIG. 3A according to some embodiments of the disclosure. [Figure 4A] 1A-1C show elevational (cross) cross-sectional views of suction mechanisms according to some embodiments of the disclosure. [Figure 4B] 4B shows a first view of the mechanical portion of the suction mechanism of FIG. 4A according to some embodiments of the disclosure. [Figure 4C] 4B illustrates a second view of the mechanical portion of the suction mechanism of FIG. 4A, according to some embodiments of the disclosure. [Figure 5] 2 illustrates an exemplary cleaning path for the system shown in FIG. 1 according to some embodiments of the disclosure. [Figure 6] FIG. 2 illustrates a bottom perspective view of the system shown in FIG. 1 removably attached to a mooring bar, according to some embodiments of the disclosure. [Figure 7A] 2 illustrates a top (plan) view of the system shown in FIG. 1 removably attached to a mooring bar, according to some embodiments of the disclosure. [Figure 7B] 7B illustrates a top (plan) view of the system shown in FIG. 1 detached from the mooring bar of FIG. 7A according to some embodiments of the disclosure. [Figure 8] 2 shows a block diagram of the system shown in FIG. 1 according to some embodiments of the disclosure. [Figure 9] 1 illustrates an overview of a detection system hierarchy according to some embodiments of the disclosure. [Figure 10] 1 illustrates an overview of ocean current detection associated with various seawater velocities and surface roughness according to some embodiments of the disclosure. [Figure 11] 10 illustrates a surface weld line on a hull and a protrusion from an inner rib weld for use in determining position, according to some embodiments of the disclosure. [Figure 12]12 illustrates details of the surface weld line and protrusion from the inner rib weld of FIG. 11 according to some embodiments of the disclosure. [Figure 13] 1 illustrates an exemplary embodiment of the use of ultrasonic or eddy current devices for position determination on the surface of a vessel, according to some embodiments of the disclosure. [Figure 14A] 1 illustrates an exemplary embodiment of a physical distance and angle sensor on a flat surface, according to some embodiments of the disclosure. [Figure 14B] 1 illustrates an exemplary embodiment of a physical distance and angle sensor on a curved surface, according to some embodiments of the disclosure. [Figure 15] 1 illustrates the relationship between magnetic flux and air gap according to some embodiments of the disclosure. [Figure 16] 1 shows the relationship between magnetic flux and time for various air gap distances (mm) between a 150 lb switchable magnet and a ½ inch thick steel plate, according to some embodiments of the disclosure. [Figure 17] FIG. 1 shows a block diagram of a machine learning module that may be used with the present system. DETAILED DESCRIPTION OF THE INVENTION

[0009] Aspects of the present disclosure may be used to reduce the level of biofouling (e.g., on ship hulls and elsewhere) by providing robotic systems and methods for inspecting the condition of ship hull surfaces, removing potential sources of biofouling from underwater surfaces, and maintaining these surfaces clean. In embodiments, (e.g., fossil) fuel consumption data before and after a cleaning operation may be used to demonstrate a reduction in carbon emissions to generate carbon credits. For example, a Carbon Intensity Indicator (CII) or similar reliable and verifiable measure may be used to measure how efficiently a marine or naval vessel transports its cargo in terms of grams of carbon dioxide emitted per cargo load and nautical mile. The difference in the CII before and after a cleaning or maintenance operation may be used to provide a number and value for carbon credits in terms of the amount of fossil fuel saved and carbon dioxide emitted.

[0010] In some embodiments, the systems and methods provided herein for inspecting and / or maintaining the surface of an object (e.g., a naval vessel) in a (e.g., marine) environment can be used to overcome several obstacles to current inspection and / or maintenance operations on marine vessels. For example, current inspection / maintenance operations can be performed within port facilities (i.e., in harbors) where the size of the inspection / maintenance system can be more effectively handled and supported by divers along with land-based auxiliary systems, and where the velocity of seawater has little or no effect on the performance of the cleaning cart system. However, inspection / maintenance operations in harbors can be limited. For example, environmental regulations may not permit or limit the extent to which inspection / maintenance operations can be performed in harbors. The performance of inspection / maintenance operations in harbors can be affected by divers as well as the availability of mooring sites. Furthermore, when permitted, low visibility due to, for example, turbid water in harbors can affect inspection / maintenance operations. Finally, harbor inspection / maintenance systems are generally relatively expensive and lack flexibility in their implementation. Indeed, harbor inspection / maintenance systems tend to be large, requiring extensive mooring space and auxiliary equipment.

[0011] The present disclosure provides a more flexible system that is (i) more economical, (ii) smaller in size, (iii) operates in open waters outside of ports while the vessel (e.g., marine or naval) is stationary or underway and the inspection / maintenance system and the object to be cleaned may be subject to stronger currents, and (iv) is desirably performed without the use of divers. When used to perform inspection / maintenance operations outside of ports, the system may be applied when the object (e.g., naval vessel) is waiting to enter the port, thus making more effective use of waiting time.

[0012] While the present disclosure is described for application in a marine environment and for steel objects, the environment and composition of the object are used for illustrative purposes only, and one skilled in the art will recognize that the inspection and maintenance robotic system described below is not limited to application in a marine environment and / or to steel objects.

[0013] Submersible self-propelled system for inspection / maintenance of objects Referring now to FIG. 1 , a submersible self-propulsion system 10 is illustrated. For simplicity and purposes of illustration, not limitation, the object is described as a marine or naval vessel, and more specifically, the iron hull (hereinafter “hull”) of a marine or naval vessel, although those skilled in the art will recognize that there are countless objects with which the system 10 may be used, including, but not limited to, industrial values. Also, while an embodiment is described in which the object is submerged in seawater and subject to currents, for purposes of illustration, not limitation, those skilled in the art will recognize that the system 10 may be used with other fluids, such as fresh water, lubricating fluids, dielectric fluids, fossil fuels, and the like, and exposed to a variety of fluid conditions. Furthermore, while an embodiment is described in connection with an application in which the vessel is constructed of steel, those skilled in the art will recognize that the teachings of the present disclosure may also be applied to industrial values ​​constructed of non-ferrous materials, as well as surfaces having specialized coatings.

[0014] The system 10 has a size and dimensions that reduce the weight of the system 10 to a few pounds (or kilograms), which may reduce the manufacturing, operating, maintenance, replacement, or other costs of the system 10. The system 10 may be deployed by a single user in some embodiments. In some embodiments, multiple systems 10 may be used as, but not limited to, a swarm. In some embodiments, the systems 10 in a swarm of systems 10 may be controlled individually and / or collectively. By way of example and not limitation, multiple systems 10 may operate with swarm intelligence, such as but not limited to particle swarm optimization, ant colony optimization, bee swarm optimization, etc. A swarm of systems 10 may operate to generate a map of the hull, cooperatively clean the hull, etc. Each system 10 in a swarm of systems 10 perceives each other system 10 in the swarm, the surrounding environment, etc. In some embodiments, a swarm of systems 10 may be autonomous, i.e., capable of operating independently without requiring direction from other swarm members. Systems 10 in a swarm of systems 10 may have a coordinated aspect by autonomously seeking new tasks once an initial task is completed, such as, for example, cleaning an area, mapping a portion of a vessel, collecting data, etc. A swarm of systems 10 may be scalable, for example, by moving systems 10 laterally away from the center of gravity of the swarm. In some embodiments, a swarm of systems 10 may be elastic, for example, when one or more members of the swarm are removed, the swarm may adjust its operations to compensate for this removal.

[0015] System 10 may include multiple sensors. One or more of the sensors in system 10 may include, but are not limited to, sensing devices (sensors) including imaging devices, cameras, inertial measurement units (IMUs), depth and / or pressure sensors, temperature sensors, ultrasound and / or ultrasonic devices, turbidity sensors, hydrophones, global positioning devices, chemical sensors, scanning SQUID microscopes, corrosion sensors, paint and / or coating thickness sensors, and / or other sensors. The sensors in system 10 may be constructed and arranged to provide data to the computing device of system 10 and / or to one or more external computing devices, such as, but not limited to, smartphones, laptops, desktops, tablets, servers, etc. The sensor data in the form of relative and absolute magnitudes of system 10 may include, but is not limited to, image data, depth data, pressure data, turbidity data, GPS data, chemical data, and / or other forms of data generated by any of the sensors described above and that may be transmitted to a user via one or more external computing devices. Data generated by one or more sensors may be used by system 10 to perform various tasks, including, but not limited to, (i) remote inspection and location mapping, (ii) cleaning (e.g., large) industrial surfaces, (iii) performing proactive troubleshooting, and (iv) performing maintenance and / or other tasks. In some embodiments, data and information generated by multiple sensors may be used to evaluate one or more of the integrity, structure, surface quality, and other factors related to the surface undergoing inspection and / or cleaning, including, but not limited to, a ship's hull. In some embodiments, a representation of the acquired data may be provided to a user through a four-dimensional (x, y, z, and time) representation of the surface on which system 10 is deployed. In some embodiments, the data provided through the four-dimensional representation of the surface may include coordinates, time elements, sensor data, and the like. By way of example and not limitation, sensor data may be mapped to coordinates on the four-dimensional representation of the surface. Machine learning models, such as those described below with reference to FIG. 17, may reconstruct and / or generate future predictions about the relative and / or absolute magnitudes of variables described throughout this disclosure based on the four-dimensional representation of the surface.In some embodiments, the machine learning model may generate one or more representations of a surface through two or more variables.

[0016] In some embodiments, the system 10 may include sensors and communications that enable a user and / or the system 10 to characterize the surface of the hull while the hull is in motion and / or in a steady fluid flow. By way of example, a computing device of the system 10 and / or a computing device in communication with the system 10 may be configured to characterize the hull's geometry, thickness, surface area, and / or other characteristics. In some embodiments, once the hull's surface is characterized, the system 10 may be configured to generate and / or calculate its own absolute and / or relative position relative to the hull. The absolute position may include a GPS position, Cartesian and / or polar coordinates of the hull's position, and / or other locations. The relative position may include distance, height, etc., relative to one or more features of the hull, such as, but not limited to, a rib, a weld line, etc. As a non-limiting example, the relative position may include a distance of approximately 10 feet from the first rib of the hull. One or more sensors of the system 10 may generate position data. Location data may include, but is not limited to, GPS coordinates, depth, height, elevation above sea level, proximity, and the like. The position data allows the system 10 to know its precise location. Precise location may include absolute or relative location in some embodiments. Precise location may be a position determination within a tolerance of approximately 0.1% in some embodiments. Knowing its precise location allows the system 10 to navigate a programmed path along the vessel. The position data may include a time element. The time element may include hours, seconds, minutes, hours, etc. from a reference point, and / or other forms of time information. As a non-limiting example, a relative location may include a distance of approximately 10 feet from the first rib of the vessel at a time element of 10 minutes from initial deployment of the system 10. The programmed path may include route instructions or other lateral movement guides for the system 10 to follow on the vessel. The programmed path may include linear, non-linear, geometric, and / or other paths. In some embodiments, the programmed path may include one or more grid patterns. In some embodiments, the system 10 may adjust the programmed path based on sensor data, such as, but not limited to, depth data, proximity data, Hall Effect data, time data, etc. The time data may include one or more timestamps of one or more operations of the system 10, measured times while the system 10 performs various tasks, coordinates and / or relative positions of the system 10 at various times relative to an initial starting point, etc. The timestamp may include a particular point in a timeline of the operation of the system 10, such as the length of an operation, which may include cleaning, mapping, inspection, etc. The timestamp may be relative to an initial starting point of one or more operations of the system 10. The initial starting point may include deployment of the system 10 into a denial environment, such as underwater. In other embodiments, the initial starting point may begin when the system 10 begins an operation, such as cleaning, traversing, mapping, etc. In some embodiments, the initial starting point may be set by a user. The timestamp may be specific to an absolute location, such as a GPS location, and / or a relative location, such as within the proximity of one or more features of the surface. The features may include, but are not limited to, a hull rib, a vessel rotor, a vessel valve, the vessel height, and / or other features. As a non-limiting example of time data, the time data may include approximately 45 minutes of cleaning a surface area, followed by approximately 10 minutes of mapping operations, followed by approximately 2 minutes of traversing the surface. The time data may be used by a computing device of the system 10 and / or a computing device in communication with the system 10 to determine deviations from a programmed path. Deviations from a programmed path may include, but are not limited to, the distance of the system 10 from a relative position, the distance of the system 10 from an absolute position, deviations of the system 10 from a relative and / or absolute position over time, and / or other combinations of time and position. By way of example and not limitation, the time data may be used to determine whether the system 10 is deviating from an ideal programmed path. The ideal programmed path may include absolute and / or relative position of the system 10 with respect to one or more timestamps, such as seconds, minutes, hours, and / or other time measurements, and may include the cleaning efficiency of the system 10, the cleaning rate of the system 10, etc. The cleaning efficiency may include the rate of increase in cleanliness of an object surface per hour, such as minutes, hours, etc. The cleaning rate may include the surface area served by the system 10 per hour. As a non-limiting example, cleaning efficiency can be 80% efficient with the ideal amount of biofouling removed from a surface per hour. The ideal amount of biofouling removed can be, but is not limited to, about 1 kg per hour.As another non-limiting example, the cleaning rate may be 10 square meters of surface every 10 minutes. Comparison of time data and / or relative, absolute, or other positions of the system 10 to the ideal programmed path may be used to determine deviation of the system 10 from the ideal programmed path and / or determine the path the system 10 will take in time. As a non-limiting example, the system 10 may be programmed with an ideal path for cleaning a surface, such as a ship's hull, in a 30-minute time period. The system 10 may clean the surface of the ship's hull with only a 19-minute deviation from the cleaning time of the ship's hull.

[0017] System 10 may operate locally without communication to an external computing device. Continuing with this example, once system 10 has completed a cleaning operation on a surface, a comparison of the relative and / or absolute positions of system 10 and the time data associated with those positions may be made against parameters of an ideal program path, which may determine ground truth data for the path taken by system 10. The ground truth data for the path taken by system 10 may be the actual path taken by system 10, which may differ from the ideal path taken by system 10. In some embodiments, the ground truth data may be determined by obtaining data from system 10 after a cleaning operation and comparing the data for the actual path taken by system 10 to one or more parameters of the ideal path through a computing device and / or machine learning model. If additional cleaning, surface mapping, and / or other operations are required, a comparison of system 10's ground truth data to system 10's ideal path may be used to determine the actual cleaning operation of system 10 and / or for other decisions. In a swarm-configured embodiment, when a determination is made that additional cleaning and / or mapping is required, one or more swarm members may be configured to finish those tasks not completed by the departed swarm member. The comparison of the ground truth data of the system 10 to the ideal path of the system 10 may include a comparison of one or more thresholds, such as absolute and / or relative position deviation in millimeters, meters, or other distances, cleaning efficiency deviation, cleaning speed deviation, and / or other parameters. The absolute and / or relative position deviation may include deviation in time data, such as seconds, minutes, hours, etc., and one or more relative and / or absolute positions associated with the time data.

[0018] In some embodiments, the system 10 may be configured to determine deviations, comparisons between ideal paths and actual paths taken, etc., either locally or via communication with one or more external computing devices. Based on one or more deviation thresholds being met, such as distance from absolute and / or relative position with respect to time data, cleaning operation completeness, surface mapping completeness, and / or other parameters, the system 10 may self-adjust. The system 10 may be configured to optimize cleaning operations locally or remotely via communication with one or more external computing devices. Optimization may include the use of one or more machine learning models, objective functions, loss functions, etc. Optimization may include maximizing the cleanliness gain of a surface, such as a surface area, served by the system 10 while minimizing the total time spent cleaning. By way of example, the system 10 may prioritize biofouling hotspots on a surface while minimizing the time spent cleaning low-biofouling areas of the surface.

[0019] In some embodiments, the system 10 may utilize a path machine learning model to calculate the most efficient path. The path machine learning model may be trained with training data that correlates sensor data and / or program path with one or more adjusted paths. The training data may be received from user input, an external computing device, and / or through prior iterative processing. In some embodiments, the system 10 may train and / or deploy the path machine learning model locally. In other embodiments, the path machine learning model may be trained and / or deployed, and the output of the path machine learning model may be communicated to the system 10. The path machine learning model may be comprised of depth, hull movement, debris, hull cleanliness, suction strength, and / or other parameters that may affect the path of the system 10. The path machine learning model may be used to determine, without limitation, deviations from the ideal program path, ground truth data, optimization of cleaning operations, and / or other determinations as described above. By way of example, the path machine learning model may generate a model of the actual path taken by the system 10 compared to the ideal path of the system 10. The path machine learning model may generate one or more deviation thresholds. In other embodiments, the path machine learning model may accept a deviation threshold from a user input. Those skilled in the art, upon reading this disclosure, will recognize that the more accurately the system 10 is positioned, the greater the cleaning efficiency, since the system 10 does not need to repeat the path to ensure 100% coverage of the hull's surface area. In some embodiments, a combination of the path machine learning model and the onboard sensors of the system 10 may be combined. For example, and without limitation, those skilled in the art, upon reading this disclosure, will recognize that not all areas of a surface accumulate biofouling or other debris at the same rate, and this information may be used to inform the path of the system 10. The path machine learning model may determine the cleaning speed and / or cleaning path of the system 10 based on the cleanliness level of a surface, such as a hull. As described in more detail below, the cleanliness level may be determined by image sensor data and / or other data received from one or more sensors of the system 10. The path machine learning model or other process may determine the cleaning speed of the system 10. The cleaning speed may include the time the system 10 may spend cleaning a surface area.The cleaning rate may be specific to one or more portions of the surface. In some embodiments, the path machine learning model may determine biofouling hot spots or peaks on the surface. The heat spots may be used by the path machine learning model to determine a cleaning path for the system 10. The path machine learning model may determine one or more cleaning actions of the system 10, such as one or more paths, power output of a cleaning device, such as a brush, as described below, and / or other parameters. By way of example and not limitation, the cleaning path machine learning model may determine that a particular spot on the surface may require multiple passes from the system 10, varying the revolutions per minute (RPM) of one or more brushes of the system 10.

[0020] The application and use of system 10 in a marine environment where system 10 operates while submerged presents several unique design factors. For example, when submerged, the slight positive buoyancy of system 10 combined with the action of water velocity along with subsurface forces can tend to push system 10 up off the surface of the vessel, which can adversely affect the ability of system 10 to inspect and / or service the vessel.

[0021] In some embodiments, to counteract the system 10's tendency to be pushed up and away from the hull, the system 10 may be adapted to utilize an adsorption system that adheres to a myriad of surfaces, including surfaces (e.g., flat, convex, concave) made from steel. While embodiments are described for applications involving steel, those skilled in the art will recognize, upon reading this disclosure, that the system teachings may be modified to provide adsorption using negative pressure (i.e., suction). Additionally, the system 10 may be configured to utilize a path planning method that controls the speed, position, and operating path of the system 10 on the hull. The path planning method may take into account the magnitude of one or more external forces applied to the system 10 during operation, the position of the system 10 relative to the hull, obstacles in the path of the system 10, the effectiveness of hull cleaning, and so forth. By way of example, a horizontal orientation of the system 10 relative to the hull may conserve most of the system's energy content, but may reduce the downward pressure exerted by water currents as it flows over the system 10. Taking into account the orientation of system 10 and the pressures applied to system 10, a path may be planned to modify the alignment of system 10 to increase or decrease the level of adhesion of system 10 to the hull, which may increase or decrease the energy consumption of system 10. Path planning methods may include, without limitation, utilizing one or more machine learning models, such as any of the machine learning models described throughout this disclosure.

[0022] The system 10's method of propelling the hull relies on constantly and / or continuously applying a variable level of suction to the hull, regardless of conditions including the planarity, convexity, and / or concavity of the hull's surface or environmental influences such as the velocity of water currents impinging on the hull and the system 10. The suction level may be equivalent to the force required to hold the system 10 to the hull while also applying sufficient suction to enable the brushes 24 of the plurality of servicing elements 20A, 20B to remove unwanted debris and material from the hull's surface. The system 10 may be configured to utilize an suction machine learning model or other neural network. The suction machine learning model may be trained with training data that correlates hull conditions and / or environmental influences with the level of suction. The training data may be received from a user input, an external computing device, and / or through prior iterative processing. The suction machine learning model may be configured to input a current suction level, environmental influences, and / or hull conditions and output one or more suction levels. In some embodiments, the adsorbed machine learning model may be trained and / or deployed locally. In other embodiments, the adsorbed machine learning model may be trained and / or deployed on an external computing device, and the output of the adsorbed machine learning model may be provided to system 10.

[0023] In some embodiments, the system 10 is adapted to utilize variable levels of attraction provided by a plurality of selectively controllable and variably switched magnets 26. The controller and / or control application may be adapted to turn the selective magnetic attraction devices 26 on or on, turn the magnetic attraction devices 26 off or off, and / or control or adjust the strength of the magnetic force of the selective magnetic attraction devices 26. For example, in one embodiment, the controller and / or control application is adapted to move the system 10 using at least two switchable contact points, such as two magnetic attraction devices 26. The controller and / or control application may selectively and / or alternatively turn on one or more of the magnetic attraction devices 26. In some embodiments, the first magnetic attraction device 26 may be a first contact point that adheres to the hull when in an active or on state. The second magnetic attraction device 26 may be a second contact point that may have a lower level of attraction to the hull or be easily interruptible when in an inactive or off state. The system 10 may be capable of pivoting or rotating about the first contact point. In some embodiments, a "leapfrog" form of propulsion may be created by alternatively turning one or more magnetic attraction devices 26 on and off. The leapfrog method may ensure that at least one of the magnetic attraction devices 26 remains securely attached to the hull, preventing the system 10 from falling off the hull. In some embodiments, high-friction rubber pads may be disposed around the magnetic attraction devices 26, which may help prevent the system 10 from sliding off the surface of an object to which it may be attracted. The high-friction rubber pads may increase the static and dynamic friction of the attraction mechanism of the system 10.

[0024] In some embodiments, the system 10 may include a communication device that may be configured to emit one or more signals. The signals may include, but are not limited to, supersonic, sonic, infrared, radio, and / or other signals. The communication device may include a supersonic transmitter, a radio transmitter, an infrared transmitter, and the like. The communication device may be useful for facilitating the location and recovery of the system 10 if it becomes detached from the vessel. In one embodiment, the system 10 may include a signaling device that may generate and emit a signal to, for example, but not limited to, an autonomous underwater vehicle (AUV). The AUV may be a submersible, underwater, or airborne vehicle. In some embodiments, the AUV may include a recovery arm to recover a lost system 10.

[0025] Still referring to FIG. 1 , the system 10 includes multiple servicing elements 20A, 20B adapted to remove debris and / or other obstacles from the surface of the hull. In some embodiments, the servicing elements 20A, 20B may be adapted to remove debris regardless of the orientation of the system 10 and / or the hull, the surface of the hull, the motion of the marine vessel, and / or environmental conditions. Multiple sensors incorporated into the system 10 may enable a user and / or the system 10 to evaluate the effectiveness of cleaning and servicing operations. By way of example and not limitation, imaging devices and / or other sensors may provide information to the system 10 about the cleanliness level of the hull. In some embodiments, a cleaning machine learning model may be used by the system 10. The cleaning machine learning model may be trained with training data that correlates sensor data to hull cleanliness. The training data may be received from a user input, an external computing device, and / or through prior iterative processing. The cleaning machine learning model may be configured to input sensor data and output one or more hull cleanliness levels, such as, but not limited to, fouled, average, clean, spotless, etc., and / or a level of force to be applied by the maintenance elements 20A, 20B corresponding to the hull cleanliness level. As a non-limiting example, the cleaning machine learning model may input sensor data and output a hull cleanliness level of fouled, corresponding to a high power output or force of the maintenance elements 20A, 20B. The cleaning machine learning model may be trained and / or deployed locally to the system 10 and / or may be trained and / or deployed on a remote computing device, and the output of the cleaning machine learning model may be communicated to the system 10.

[0026] In some embodiments, the system 10 includes multiple rotating brushes 24 for maintaining the hull of a boat. The rotating brushes 24 may be configured to rotate in a clockwise and / or counterclockwise direction. In some embodiments, the system 10 may include two or more rotating brushes 24. By way of example and not limitation, the system 10 may include a first rotating brush 24 on a left side of the system 10 and a second rotating brush 24 on a right side of the system 10. In some embodiments, one or more magnetic attraction devices 26 may be disposed around the rotating brushes 24. As a non-limiting example, three magnetic attraction devices 26 may be disposed around a first periphery of the rotating brush 24 and three magnetic attraction devices 26 may be disposed around a second periphery of the second rotating brush 24. In some embodiments, the first rotating brush 24 may be configured to operate independently from the second rotating brush 24. As a non-limiting example, the first rotating brush 24 may rotate in a clockwise direction and the second rotating brush 24 may rotate in a counterclockwise direction. In some embodiments, two or more rotating brushes 24 may operate simultaneously and / or collectively. The rotating brush 24 may be circular, rectangular, and / or other geometric shapes. In some embodiments, the rotating brush 24 may have a radius of approximately 5 inches. In other embodiments, the rotating brush 24 may have a radius greater than or less than approximately 5 inches. In some embodiments, the magnetic attraction elements 26 may be selectively controlled to provide high or low magnetic forces to the surface of the hull, which may enable the rotating brush 24 to remove debris, marine life, and / or unwanted material. To facilitate matching the appropriate magnetic force to the detected debris, marine life, and / or unwanted material (e.g., biofouling), the imaging devices 33 and light-emitting elements 31 may be used to detect debris, marine life, etc. In some embodiments, the computing device of the system 10 may be configured to adjust the magnetic force applied to the surface of the hull based on sensor data received from one or more sensors, such as, but not limited to, the imaging devices 33 and / or other sensors.

[0027] In some embodiments, the system 10 may include bellows 34. The bellows 34 may be operable to maintain the concave and convex surfaces seen from bow to stern and / or port to starboard of the marine vessel. By way of example and not limitation, the system 10 may have a first bellows 34 on the right side of the system 10 and a second bellows 34 on the left side of the system 10. Each bellows 34 may be operable to rotate relative to the bottom of the system 10, such as in a clockwise and / or counterclockwise direction. Each bellows 34 may be controlled independently of the others. Adjusting the angle of the bellows 34 allows the system 10 to traverse various concave and / or convex portions of the hull. A computing device of the system 10 may detect various features of the hull's surface and adjust the angle of the bellows 34 accordingly.

[0028] Depending on the nature and extent of debris on the hull, the stiffness or bristle hardness of the rotating brush 24 may be advantageously adjusted to be softer or stiffer. In some embodiments, for example, but not limited to, when the extent of biofouling is slight, a single rotating brush 24 may be combined with a wiper. In some embodiments, for certain types of debris, marine life, and / or unwanted material that cannot be removed by the rotating brush 24, a specially designed cleaning attachment may be used instead. For example, in some applications, the rotating brush 24 may be replaced with a rotating cutter blade, for example, but not limited to, removing barnacles and mussels. In other embodiments, the rotating brush 24 may be replaced with an abrasive brush that may be used to reduce surface friction on the object after general cleaning is completed. In some embodiments, the system 10 may have a combination of rotating brushes 24, rotating cutter blades, and / or abrasive brushes. One or more of the rotating brushes 24, rotating cutter blades, and / or abrasive brushes may be activated by the computing device of the system 10 based on sensor data, such as, but not limited to, image data, suction data, etc.

[0029] In some embodiments, system 10 may include a non-rigid body. A non-rigid body may be any body that can flex, stretch, twist, or the like without impediment. For example, system 10 may have a flexible skeletal body. A flexible skeletal body may include one or more interconnected components that may be mechanically actuated to rotate, stretch, twist, or the like. In some embodiments, a flexible skeletal body may be linear, non-linear, or otherwise shaped. A flexible skeletal body may include a main body with one or more branches that may extend from the main body. In some embodiments, a flexible skeletal body may be a body in which either the primary or secondary structural elements of the body include joints or other flexible elements with variable stiffness that can be controlled to match the geometry of the surface, thereby allowing system 10 to adhere and actuate to conform to non-flat surfaces. In embodiments, the first housing portion 12 and / or the second housing portion 14 may include one or more mechanical joints that may allow one or more components of the system 10 to flex, rotate, contract, etc. The flexible skeletal body of the system 10 may allow one or more degrees of freedom. By way of example and not limitation, the flexible skeletal body of the system 10 may allow up to six or more degrees of freedom. The increased degrees of freedom enable increased sensing of the system 10. By way of example and not limitation, the system 10 may have one or more sensors in communication with one or more joints of the skeletal body that may allow additional sensing of the environment and / or positioning of the system 10. By way of non-limiting example, rather than relying on on-board sensors of the system 10, one or more components of the flexible skeletal body may be configured to detect water flow, pressure, magnetic fields, etc. In some embodiments, one or more components of the flexible skeletal body may include one or more adhesive devices, such as magnetic adhesive devices 26, which may provide adhesion of the flexible skeletal body to one or more components of the hull. The adhesive devices of the flexible skeletal body may increase the propulsive capabilities of the system 10, such as by providing more degrees of freedom. The increased sensing capabilities of the skeletal body may provide additional and / or more accurate sensor data for one or more machine learning models, such as, without limitation, any of the machine learning models described throughout this disclosure.The sensor data of the flexible skeletal body may be used by one or more machine learning models to improve, without limitation, predicting the propulsion of the system 10, detecting the adhesion level of one or more magnetic adhesion devices 26, generating a map of the hull, and the like.

[0030] With continued reference to FIG. 1 , the system 10 may be capable of interacting with one or more valuable objects. The valuable objects may include, but are not limited to, sea chests, intake and outflow pipes, sacrificial anodes, and the like. The system 10 may be capable of interacting with one or more valuable objects using a robotic manipulation process. The robotic manipulation process may include, but is not limited to, activation of any of the elements of the system 10 described throughout this disclosure. In some embodiments, the system 10 may be configured to collect information related to maintenance related to the surface of the hull, the substrate, or any other aspect of the environment. The related information may include, but is not limited to, any of the sensor data described throughout this disclosure. The robotic manipulation process operations may be performed on a planned or ad-hoc basis. In some applications, the system 10 may utilize a valuable object machine learning model to predict when the valuable object may require proactive intervention. Proactive intervention may include individual removal of the valuable object or other manipulation processes. The removal and / or manipulation of the valuable object enables the system 10 to continuously clean the hull. A value machine learning model may be trained with training data that correlates sensor data with one or more value items requiring proactive intervention. The training data may be received from user input, an external computing device, and / or through prior iterative processing. The value machine learning model may be configured to input sensor data and output an indication of one or more value items requiring proactive intervention. System 10 may train and / or deploy the value machine learning model locally and / or may be in communication with an external computing device that may train and / or deploy the value machine learning model.

[0031] The system 10 may be operable to collect data about the surface of the hull to create a temporal mapping of the hull's surface. The temporal mapping may be a geographic layout of the hull over a period of time. A period of time may include, but is not limited to, minutes, hours, days, etc. In some embodiments, the temporal mapping may be a real-time geographic layout of the hull. The temporal mapping, which may be generated for specific or arbitrary time intervals, allows a user and / or the system 10 to track any and all changes in the condition of the hull's surface through the use of precise location and sensor data. This feature may enable a user and / or the system 10 to perform one or more maintenance actions to prevent further surface, structural, or other deterioration of the hull.

[0032] In some embodiments, system 10 includes a housing portion. As used in this disclosure, a "housing" is a structure that houses components and / or subcomponents of a system or device and is part of the system or device itself. As described in more detail below, the housing portion of system 10 can house one or more elements of system 10. The housing portion can have a first or upper housing portion 12 and a second or lower housing portion 14. The first housing portion 12 can be opposite the second housing portion 14. In some embodiments, the first housing portion 12 can be configured to removably engage with the second housing portion 14. The engagement between the first housing portion 12 and the second housing portion 14 can provide an interior space that can house one or more elements of system 10. In an exemplary embodiment, system 10 can be approximately 12 inches long, approximately 6 inches wide, and approximately 4 inches high. In some embodiments, system 10 can be greater than or less than 12 inches long, 6 inches wide, and approximately 4 inches high.

[0033] The first housing portion 12 may be hydrodynamically designed. By way of example and not limitation, the first housing portion 12 may be shaped as a shell or other hydrodynamic shape, which may have a structure and arrangement that allows the system 10 to remain on the surface of an object, even in high currents. In some embodiments, the first housing portion 12 may have a plurality of indentations and / or grooves, which may enhance the hydrodynamics of the system 10. By way of example and not limitation, the first housing portion 12 may have a plurality of thin lines that are parallel to each other and equidistant from each other. The system 10 may be constructed and arranged to provide a slight positive buoyancy that allows the system 10 to float if detached during use. Alternatively, or in addition, the system 10 may include an airbag and gas source disposed within the interior space between the first housing portion 12 and the second housing portion 14. The airbag and gas, which may be compressed air, may be operable to inflate the airbag and raise the system 10 to the surface of the seawater should the system 10 become detached and begin to sink. In some embodiments, the airbag system may be deployed when the system 10 reaches a predetermined depth, such as, but not limited to, about 20 meters.

[0034] Still referring to FIG. 1 , the first housing portion 12 and the second housing portion 14 may be configured to engage with one another, which may provide an airtight and / or watertight seal. The airtight and / or watertight seal created by the engagement of the first housing portion 12 and the second housing portion 14 may prevent seawater, air, and / or other fluids from entering the interior of the system 12. Alternatively or optionally, the first housing portion 12 and the second housing portion 14 may be configured to house a sealing device. The sealing device may include, but is not limited to, a gasket, an O-ring, and / or other sealing device. In some embodiments, the sealing device may be disposed in a groove formed in the periphery of one or both of the first housing portion 12 and the second housing portion 14.

[0035] Still referring to FIG. 1 , in some embodiments, the first housing portion 12 may include a heat sink feature 11. The heat sink feature 11 may be made of a metal or other element capable of conducting heat. The heat sink feature 11 may be operable to cool heat-generating components disposed in the first housing portion 12 and / or the second housing portion 14. In some embodiments, the heat sink feature 11 may be in the form of a thin plate. The heat sink feature 11 may be operable to cool the communication port 16. The communication port 16 may be an electrical port through which a communication cable 60 connects to a component disposed within the interior space of the system 10. While the port is referred to as the “communication port 16” and the cable entering the communication port 16 is referred to as the “communication cable 60” for purposes of simplicity, those skilled in the art will recognize that hard wiring for one or more of a power connection, an electronic connection, a communication connection, etc. may be included in the communication cable 60 and enclosed in a watertight enclosure or cover. While the communication port 16 is depicted as a hollow cylinder, this is done for purposes of illustration and not limitation. Those skilled in the art will recognize that there are countless sizes, shapes, and locations for communication ports 16 .

[0036] In some embodiments, a slip ring 15 may be rotatably mounted to the distal end of the communication port 16. The slip ring 15 may include an opening for the communication cable 60 to enter the communication port 16. The slip ring 15 may be adapted to rotate about a longitudinal axis that passes through the center of the communication port 16 to prevent tangling.

[0037] As shown in Figures 2, 3A, and 3B, an arched portion 17 may be formed in the rear portion of the first housing section 12. In some implementations, as shown below with reference to Figure 4, the arched portion 17 may be constructed and arranged to house all or a portion of a circular connection port 30 that may be provided for selectively and removably attaching any tether rod 40 to the system 10. In some embodiments, the connection port 30 may include a cylindrical tether port 32 having a plurality of detents 37 disposed on an inner circumferential surface of the tether port 32. The plurality of detents 37 may securely retain and / or releasably attach the tether rod 40 to the system 10.

[0038] Referring again to FIG. 1 , the front portion of the second housing portion 14 may define an opening 35. The opening 35 may be oval, circular, oblong, and / or other geometric shapes. In some embodiments, the opening 35 may be constructed and arranged to house a light-emitting element 31. The light-emitting element 31 may be configured to illuminate a portion of an environment. The light-emitting element 31 may include one or more light-emitting diodes (LEDs) and / or other light-emitting elements. The environment illuminated by the light-emitting element 31 may include the hull and portions thereof. In some embodiments, the opening 35 may house an image capture device 33. The image capture device 33 may be a camera, video recorder, and / or other device. The image capture device 33 may be configured to capture individual, sequential, and / or other images and / or videos of the environment of the system 10. In some embodiments, the image capture device 33 may capture images and / or videos in the direction of travel of the system 10. In some embodiments, the imaging device 33 may be a wide-angle camera configured to provide real-time images of the hull. The imaging device 33 may be rotatable about an x-axis, a y-axis, etc., which may enable the imaging device 33 to capture images at different angles. Images obtained by the imaging device 33 may be used by a computing device of the system 10 and / or a computing device in communication with the system 10 to create a map of the hull and / or provide a visual inspection of the condition of the hull. In other embodiments, the (e.g., oval) aperture 35 may further be adapted to house one or more of the following, among other things, to provide data for operation, to provide data for assessing the extent of cleaning performed, or for other purposes: a magnetometer, a Hall sensor, structured light spectroscopy, fluorescence spectroscopy, other types of spectroscopy, etc.

[0039] As shown in FIGS. 3A and 3B , the second housing portion 14 may include a pair of flexible bellows 34, each including an extendable / retractable watertight element. As described above with reference to FIG. 1 , the bellows 34 may be rotatable. In some embodiments, one of the bellows 34 is disposed at the distal end of the second housing portion 14, and the other bellows 34 is disposed at the proximal end of the second housing portion 14. The bellows 34 may be constructed and arranged to ensure joint compliance and watertightness when the system 10 encounters concave and convex surfaces of the hull. The bellows 34 may be constructed and arranged to accommodate changes in orientation of the plurality of servicing elements 20A, 20B and the suction mechanism 50 as the system 10 traverses the concave, convex, and / or double-curved surfaces of the hull (described in more detail below with reference to FIG. 4 ). The articulation of the bellows 34 may be adaptable to extend and retract to maintain a watertight seal between the bottom of the system 10 and the surface of the hull. In some embodiments, an accordion-like waterproofing element may be coupled to each of the bellows 34 and the second housing portion 14 such that the waterproofing element unfolds when the bellows 34 is extended and compresses within the interior space of the system 10 when the extended bellows 34 is retracted.

[0040] Referring to Figure 3A, an exemplary embodiment is shown in which the system 10 is operating on a relatively flat surface. Figure 3B, on the other hand, shows an exemplary embodiment in which the system 10 is operating on a convex surface, so that both servicing elements 20A, 20B (and their corresponding suction mechanisms 50) are extended. The angle of one or both of the servicing elements 20A, 20B can be selectively changed to be at the same or different angles relative to the bottom surface 38 of the system 10 to accommodate a variety of concave and convex surfaces on the hull.

[0041] 3A and 3B, the rear portion of the second or lower housing portion 14 may also be configured to house all or a portion of the connection port 30. In some embodiments, a docking guidance system 39 may be disposed, for example, directly below the connection port 30. The docking guidance system 39 may include one or more rails that allow the system 10 to navigate to and dock at a central location.

[0042] Referring again to FIG. 2 , the bottom surface 38 of the second or lower housing portion 14 may be configured and arranged to house a plurality of maintenance elements 20A, 20B, which may include, by way of example and not limitation, a pair of optional rotating brushes 24. A pair of portions 18 may be formed on the front and rear of the second or lower housing portion 14. In particular, the portions 18 may be located between the maintenance elements 20A, 20B to prevent large objects from entering and potentially clogging the maintenance elements 20A, 20B. In some embodiments, the pair of portions 18 may be substantially triangular. The pair of portions 18 may provide additional surface area and internal spacing for housing the connection port 30, the opening 35, and some or all of the sensors. While FIG. 2 shows only two maintenance elements 20A, 20B and two optional rotating brushes 24, this is done for illustrative purposes only. Those skilled in the art will recognize that the size (i.e., diameter), number, bristle thickness, and position of the plurality of maintenance elements 20A, 20B may vary.

[0043] In one embodiment, the plurality of servicing elements 20A, 20B may be constructed and arranged to disperse biofilm from the hull into the surrounding water. In operation, the attraction mechanism 50 of one servicing element 20A (described below with reference to FIG. 4A ) attracts the system 10 to the hull, while the other, non-stationary, rotating servicing element 20B rotates its brushes 24 to actively clean the hull surface. In some embodiments, the brushes 24 may protrude between about 1 mm and about 8 mm beyond the bottom surfaces of the plurality of magnets 26 used to selectively attract the servicing elements 20A, 20B to the hull. In other embodiments, the brushes 24 may protrude less than 1 mm or more than 8 mm beyond the bottom surfaces of the plurality of magnets 26. This protrusion allows the non-stationary attraction mechanism 50 of the system 10 to slide across and clean the hull surface. When the magnetic attraction device 26 of the non-stationary attraction mechanism 50 is disabled or only loosely engaged, the rotating brushes 24 are selectively pressed against the hull with a force of up to about 600 pounds. Advantageously, this force is selectively controllable to apply a force between 1 lb and 600 lb. Those skilled in the art will recognize that the 600 lb force is for a given size system 10 and that the force will increase or decrease depending on the size of the system 10.

[0044] In some embodiments, each of the service elements 20A, 20B includes a (e.g., circular or disk-shaped) base portion 22. In some variations, a mounting or fastening device 28 may be used to removably and securely mount the base portion 22 to the second or lower housing portion 14. The diameter and wall thickness of the (circular or disk-shaped) base portion 22 may be varied to provide the system 10 with a desired size, weight, and serviceability.

[0045] 2, an optional rotating brush 24, e.g., a brush ring, may be fixedly or removably attached to the outer periphery of the substrate portion 22. Those skilled in the art will recognize that the diameter of the rotating brush 24 may increase or decrease the size of the system 10.

[0046] A plurality of selectively controllable, i.e., mechanically switchable, magnetic attraction devices 26 may be disposed on the base portion 22, which in some embodiments may be a planetary ring. The magnetic attraction devices 26 may be disposed between the rotating brushes 24 and the mounting or fastening devices 28 to prevent or minimize biofouling from entering the attraction area. In some embodiments, the attraction devices 26 may be magnets (e.g., 95 lbs) adapted to apply a magnetic field of up to about 250 lbs of force to a ferrous surface. The plurality of selectively controllable, i.e., mechanically switchable, magnetic attraction devices 26 on each of the maintenance elements 20A, 20B may be switched together, for example, from zero magnetic field to maximum magnetic field strength. In some embodiments, the strength of the magnetic fields of the maintenance elements 20A, 20B may be selectively adjustable to apply a high static attraction force to the hull of the fixed maintenance element 20A of the system 10, pre-equip the rotating brushes 24 of the non-fixed maintenance element 20B, or the like. 2 shows only three selectively controllable (e.g., magnetic) attraction devices 26, this is done for illustrative purposes only, and one skilled in the art will recognize that the size, number, magnetic strength, and location of the selectively controllable (e.g., magnetic) attraction devices 26 may be varied.

[0047] 4A-4C, a suction mechanism 50 is shown. For purposes of this discussion, FIG. 5 illustrates a typical operation of the suction mechanism 50 in relation to a fixed servicing element 20A and a non-fixed servicing element 20B. Those skilled in the art will recognize that the terms "fixed servicing element 20A" and "non-fixed servicing element 20B" are used for convenience, as each servicing element 20A, 20B must alternatively transition from a fixed state to a non-fixed state in order for the system 10 to translate within the hull.

[0048] A suction mechanism 50 may be provided to control the movement and operation of each of the maintenance elements 20A, 20B. In some embodiments, each suction mechanism 50 of the plurality of suction mechanisms 50 may be disposed within an interior space defined by the first housing portion 12 and the second housing portion 14. In some embodiments, the suction mechanism 50 includes a unique housing 52 that defines a void in an inner portion 54. In some embodiments, the void in the inner portion 54 allows the suction mechanism 50 to maintain neutral buoyancy.

[0049] In some embodiments, the attraction mechanism 50 includes a plurality of selectively controllable magnetic attraction devices 26, a magnet switching motor 51, a body rotation motor 53, and a magnetic switching shaft 55. A switch bar 56 may be magnetically coupled to each of the attraction devices 26 of the maintenance elements 20A, 20B. Optionally, each of the attraction devices 26 and / or the entire attraction mechanism 50 may include a ring, such as, but not limited to, a silicone ring, to increase surface friction with the stationary maintenance element 20A.

[0050] The magnetic switching motor 51 may be constructed and arranged to selectively (e.g., mechanically) turn on and off the magnetic attraction devices 26 associated with the corresponding stationary or non-stationary service elements 20A, 20B. The body rotation motor 53 may be adapted to enable clockwise and counterclockwise rotation of (e.g., individual) planetary mechanical systems 57, which in some embodiments may include circular service elements 20A, 20B rotated about a switching shaft 55 by a plurality of interconnected gears 59 and a planetary ring 58 securely fastened to the outermost gear of the plurality of interconnected gears 59. In operation, the body rotation motor 53 may apply torque to and rotate the switching shaft 55. Rotation of the switching shaft 55 may rotate the plurality of interconnected gears 59, which may in turn rotate the planetary ring 58. Rotation of the planetary ring 58 may cause rotation of the brushes 24 of the non-stationary service element 20B.

[0051] The non-stationary servicing element 20B can be a mobile and cleaning / servicing element. In some embodiments, the brushes 24 rotate counterclockwise so that biofouling being removed from the hull surface is pushed away from the center of the system, i.e., the stationary servicing element 20A. If any of the debris is magnetic, it may be attracted to, and possibly adhere to, one or more of the attractors 26. Because the attractors 26 are alternatively turned on and off, collected debris may fall off the affected attractor 26 when turned off. While the brushes 24 of the non-stationary servicing element 20B move clockwise (or alternatively counterclockwise) around the stationary servicing element 20A, the non-stationary servicing element 20B itself may rotate in the opposite direction to the suction mechanism 50 of the stationary servicing element 20A.

[0052] How to move 5, a method 75 of operation on a hull 70 is shown. In a first step, the magnet switching motor 51 of the stationary maintenance element 20A may be controlled to turn on the attraction device 26 of the stationary maintenance element 20A to secure the stationary maintenance element 20A to the hull, while the magnet switching motor 51 of the non-stationary maintenance element 20B may be controlled to turn off the attraction device 26 of the non-stationary maintenance element 20B to rotate it (e.g., clockwise) around the stationary maintenance element 20A. In a next step, the magnet switching motor 51 and the body rotation motor 53 of the stationary maintenance element 20A may be synchronized to rotate the non-stationary maintenance element 20B (e.g., clockwise) around the stationary maintenance element 20A, while the magnet switching motor 51 and the body rotation motor 53 of the non-stationary maintenance element 20B are synchronized to rotate the brushes 24 of the non-stationary maintenance element 20B.

[0053] When the maintenance element 20B has completed cleaning and maintenance of the hull to the extent possible, the brush 26 stops rotating and the magnet switching motor 51 of the maintenance element 20B can be controlled to turn on the suction device 26 of the maintenance element 20B to secure the maintenance element 20B to the hull, while the magnet switching motor 51 of the maintenance element 20A can be controlled to turn off the suction device 26 of the maintenance element 20A to rotate it around the maintenance element 20B (counterclockwise).

[0054] In the next step, the magnet switching motor 51 and body rotation motor 53 of service element 20A may be synchronized to rotate service element 20A (e.g., counterclockwise) around service element 20B, while the magnet switching motor 51 and body rotation motor 53 of service element 20A are synchronized to rotate brushes 24 of service element 20A. This process may be repeated until the system 10 requires a 180 degree change of direction.

[0055] In some embodiments, the magnet-switching motor 51 and the body-rotating motor 53 can run at the same revolutions per minute (RPM) value to rotate the attraction mechanism 50 around the stationary maintenance element 20A by locating the motors 51, 53 on the non-rotating portion, eliminating the need for slip rings. In some variations, controlling the magnet-switching motor 51 and the body-rotating motor 53 to run at the same RPM can be achieved using encoder feedback.

[0056] Mooring System 6, 7A, and 7B, there is shown an exemplary embodiment of a mooring system 100. The mooring system 100 may allow for live streaming of data (e.g., via communication cable 60). However, when unmoored, the system 10 is not constrained by mooring controls and may therefore navigate to hard-to-reach areas such as the bottom of a vessel (e.g., the keel).

[0057] In some embodiments, the mooring system 100 includes a mooring rod 40 including a long portion 42 having a proximal end 44 and a distal end 46. In some implementations, the proximal end may include an opening 48 through which a communication cable may be routed. A plurality of suction devices (e.g., magnets) 43 may be disposed on a bottom portion 41 of the long portion 42 near the distal end 46 to secure the mooring rod 40 to the hull of the vessel. A docking portion disposed on the distal end of the long portion 42 may be constructed and arranged to engage with the connecting portion 30 of the system 10.

[0058] Central Processing System and Memory 8, features for controlling system 10 will be described. In some embodiments, system 10 may include one or more of the following: a central processing system 81 electrically and / or electronically coupled to each other via a bus 89 and operatively coupled to magnet switching motor 51 and body rotation motor 53 of first and second attraction mechanisms 50, a plurality of sensing devices 82, memory or storage 83, a master controller 84, a communication link 85, an imaging device 86, an illumination device 87, and a power source 88.

[0059] Implementations of the subject matter and operations described herein may be implemented in digital electronic circuitry, or in computer software, firmware, or hardware, including the structures disclosed herein and structural equivalents thereof, or in any combination of one or more of these. Implementations of the subject matter described herein may be implemented as one or more computer programs, i.e., one or more modules of computer program instructions encoded in a computer storage medium (i.e., memory 83) for execution by or to control the operation of a data processing device (i.e., central processing system 81). Alternatively, or in addition, the program instructions may be encoded in an artificially generated propagated signal, such as a machine-generated electrical, optical, or electromagnetic signal, that is generated to encode information for transmission to an appropriate receiving device for execution by the data processing device (i.e., central processing system 81). The computer storage medium (i.e., memory 83) may be, or may be included in, a computer-readable storage device, a computer-readable storage substrate, a random or serial access memory array or device, or a combination of one or more of these. Also, although the computer storage medium (i.e., memory 83) is not a propagating signal, the computer storage medium (i.e., memory 83) may be a source or destination of computer program instructions encoded in an artificially generated propagating signal. The computer storage medium (i.e., memory 83) may also be or be included in one or more separate physical components or media (e.g., multiple CDs, disks, or other storage devices).

[0060] The operations described herein may be implemented as operations performed by a data processing device (i.e., central processing system 81) on data stored in one or more computer-readable storage devices (i.e., memory 83) or received from other sources.

[0061] The term "data processing device" includes all types of devices, apparatuses, and machines for processing data, including, by way of example, programmable processing devices (i.e., processors), computers, systems-on-chips, or any combination or combination of the above. Data processing devices (i.e., central processing systems 81) may include dedicated logic circuitry, such as FPGAs (field-programmable gate arrays) or ASICs (application-specific integrated circuits). In addition to hardware, data processing devices (i.e., central processing systems 81) may also include code that creates an execution environment for the computer program, such as processor firmware, protocol stacks, database management systems, operating systems, cross-platform runtime environments, virtual machines, or a combination of one or more of these. Devices and execution environments may implement a variety of different computing model infrastructures, such as distributed computing and grid computing infrastructures.

[0062] A computer program (also known as a program, software, software application, script, or code) may be written in any form of programming language, including compiled or interpreted languages, declarative or procedural languages, and may be deployed in any form including components, subroutines, objects, or other units suitable for use in a computing environment, either as stand-alone programs or as modules. A computer program may, but need not, correspond to a file in a file system. A program may be stored in a portion of a file that holds other programs or data (e.g., one or more scripts stored in a markup language resource), in a single file dedicated to the program, or in multiple cooperating files (e.g., files storing one or more modules, subprograms, or code portions). A computer program may be deployed to be executed on one computer or on multiple computers, either located at one site or distributed across multiple sites and interconnected by a communications network.

[0063] The processes and logic flows described herein may be implemented by one or more programmable processors that execute one or more computer programs to perform actions by operating on input data and generating output. The processes and logic flows may be implemented by dedicated logic circuitry, such as an FPGA or ASIC, or may be so implemented. In some implementations, the processes and logic flows described herein may be implemented by one or more programmable processors located remotely (e.g., on a naval vessel) of the system 10. In some implementations of the system 10, it may be advantageous for a group or multiple systems 10 to simultaneously clean / maintain a ship's hull. In such applications, a master controller may be located remotely (e.g., on a naval vessel) that is structured and arranged to receive data from each of the systems 10 in real time and control (e.g., operate) each of the systems 10.

[0064] Processors suitable for the on-board central processing system 81 and the remotely-mounted master controller and for executing computer programs include, by way of example, both general-purpose and special-purpose microprocessors and one or more processors of any type of digital computer (e.g., the NVIDIA Jetson Nano). Generally, a processor accepts instructions and data from a read-only memory or a random-access memory 83, or both. The essential elements of a computer are a processor for performing actions in accordance with the instructions and one or more memory devices 83 for storing instructions and data. Generally, a computer will also include one or more mass storage devices, such as magnetic, magneto-optical, or optical disks, for storing data, or may be operatively coupled to receive data, transmit data, or both. However, a computer need not have such devices. Suitable devices for storing computer program instructions and data include all forms of non-volatile memory, media, and memory devices, including, by way of example, semiconductor memory devices such as EPROM, EEPROM, and flash memory devices, magnetic disks such as internal hard disks or removable disks, magneto-optical disks, and CD-ROM and DVD-ROM disks. The central processing system 81 and memory 83 may be supplemented by, or incorporated in, dedicated logic circuitry.

[0065] To provide for user interaction, implementations of the subject matter described herein may be implemented in a remote master controller having a display device, such as a CRT (cathode ray tube) or LCD (liquid crystal display) monitor, for displaying information to the user, and a keyboard and pointing device, such as a mouse or trackball, for the user to provide input to the computer. Other types of devices may be used to provide user interaction as well. For example, feedback provided to the user may be some form of sensory feedback, such as visual feedback, auditory feedback, or tactile feedback, and input from the user may be accepted in any form, including acoustic, speech, or tactile input. Additionally, the computer may interact with the user by sending resources to and accepting resources from devices used by the user.

[0066] Implementations of the master controller described herein may be implemented in a computing system that includes a back-end component, such as a data server; a middleware component, such as an application server; a front-end component, such as a client computer having a graphical user interface or web browser through which a user interacts with an implementation of the subject matter described herein; or any combination of one or more such back-end, middleware, or front-end components. The remote master controllers and central processing system 81 of each system may be interconnected by any form or medium of digital data communication, such as a communications network or link 85. Examples of communications networks or links 85 include wired peer-to-peer networks (e.g., ad-hoc peer-to-peer networks), Ethernet, and / or Wi-Fi.

[0067] One or more computer systems may be configured to perform particular operations or actions by having installed on the systems software, firmware, hardware, or a combination thereof that, when run, causes the system to perform the actions. One or more computer programs may be configured to perform particular operations or actions by containing instructions that, when executed by a data processing device, cause the device to perform the actions.

[0068] While this application contains many specific implementation details, these should not be construed as limitations on the scope of the invention or what may be claimed, but rather as descriptions of features that are specific to particular implementations of particular inventions. Certain features described herein in the context of separate implementations may also be implemented in combination in a single implementation. Conversely, various features that are described in the context of a single implementation may also be implemented separately in multiple implementations or in any suitable subcombination. Also, while features are described above as working in a certain combination and are originally claimed as such, one or more features of a claimed combination may in some cases be deleted from the combination, and a claimed combination may be of a subcombination or a variation of a subcombination.

[0069] Imaging and illumination equipment Still referring to FIG. 8 , as previously mentioned, in some implementations, the system 10 may be configured to include an imaging device 86 and / or an illumination device 87. In some variations, the system 10 may be constructed and arranged to house the imaging device 86 for capturing images (e.g., discrete or continuous) of the environment (e.g., the hull) in the direction of travel of the system 10, along with an illumination device 87 having light-emitting elements 31 capable of illuminating portions of the environment (e.g., the hull) in the direction of travel of the system 10. In some implementations, the illumination device 87 may include light-emitting elements 31, such as UV lamps. Advantageously, UV lamps may be used to treat the surface of the hull to prevent organic growth and extend the time between cleanings. In some variations, the imaging device 86 may be a wide-angle camera 33 configured to provide real-time images of the hull, or a LiDAR system. Advantageously, such images may be used to create a map of the hull and perform a visual inspection of the hull's condition. The map of the hull may be a three-dimensional or two-dimensional mapping. The visual inspection of the hull condition may include identification of debris, marine life, structural integrity, etc. by system 10 and / or a computing device in communication with system 10. In some embodiments, system 10 may be configured to utilize an image recognition and / or classification mode that may identify the hull's cleanliness level, debris / marine life category, and / or other image categories. By way of example and not limitation, system 10 may input one or more images and / or videos and output bounding boxes and / or pixel-level segmentations of objects of interest. An image recognition model may be trained with training data that correlates images to one or more categories, such as, but not limited to, cleanliness level, marine life, debris, etc. Training data may be received from user input, an external computing device, and / or through prior iterative processing. In some embodiments, system 10 may train and / or deploy an image recognition process locally. In other embodiments, system 10 may communicate with an external computing device that may train and / or deploy an image recognition process and communicate the output of the image recognition process to system 10.The system 10 may utilize an image recognition process to determine the type of marine life, such as, but not limited to, barnacles, hermit crabs, muscle, and / or other marine life. The system 10 may utilize an image recognition process to determine the type of debris, such as, but not limited to, rocks, dirt, wood, biofilm, and / or other debris types. The system 10 may utilize an image recognition process to determine the level of hull cleanliness, such as, but not limited to, fouled, average, clean, etc. The system 10 may utilize one or more outputs of the image recognition process to determine traversal path, suction level, cleaning action, and / or other functions. Other embodiments may utilize structured light mounting elements and projection to further determine other visually achievable characteristics, including, but not limited to, shape, size, and other characteristics of the hull surface, rocks, dirt, wood, marine life, biofilm, and / or other debris types.

[0070] power supply 8 , power from a remote AC power source, for example, for electrical or electronic components of each system 10, may be provided via communication cable 60. Alternatively, or as a secondary power source, each system 10 may include an on-board (e.g., LiPo) battery pack 88. The size of the on-board battery pack 88 may depend on whether the system 10 is moored. For example, the size (voltage) of the battery pack 88 for an untethered system 10 may be larger than that required for a moored system 10. More specifically, each mooring rod 40 may be configured to include a large battery pack, solar panel, etc., that provides power to the system 10 when the system 10 is moored.

[0071] Adsorption mechanism controller 8, as previously described, each system 10 may include a plurality of attraction mechanisms 50 that may be alternatively controlled to selectively turn on / off the attraction devices 26 and servicing elements 20A, 20B. An attraction mechanism controller 84 is adapted to control each magnetic attraction element 26 of the attraction mechanism 50 and the servicing elements 20A, 20B so that the stationary servicing element 20A, controlled by a first attraction mechanism 50 (e.g., in a fixed state), attracts the system 10 to the hull (e.g., in a fixed state), while the other non-stationary (e.g., rotating) servicing element 20B, controlled by a second attraction mechanism 50, rotates the brushes 24 to actively clean the surface of the hull. As shown in FIG. 5, the attraction mechanism controller 84 is constructed and arranged to operate each system 10 to clean / maintain a desired path on the hull.

[0072] Detection device 8, the system 10 includes a number of sensors 82 that may be used, for example, to determine the position of the system, plan the path (i.e., maneuver) of the system, gather data about the vessel condition, etc. Exemplary sensors 82 that may be included in the system 10 may include one or more of the following: Motor encoders for positioning and magnetic engagement, motor current sensors for collision detection, surface friction feedback (from brushes 24) and / or magnetic field engagement, magnetometers for detecting gripper 26 grip, material thickness, and / or gap distance, linear distance sensors for detecting physical contact distance to weld lines or other hull features, gripper angle sensors for detecting hull curvature at discrete locations and for normalizing linear distance sensor data, ultrasonic sensors for detecting paint thickness, steel hull thickness to detect weld lines, inner weld ribs, etc., eddy current sensors for detecting paint thickness and steel hull thickness to detect weld lines, inner weld ribs, etc., optical flow sensors for detecting fluid flow, optical sensors / devices for visually operating the system and inspecting underwater conditions, inertial measurement units (IMUs) for manipulation, position determination, collision detection, etc., depth sensors, pressure sensors, temperature sensors, etc. The sensing system hierarchy is summarized in the table shown in Figure 9.

[0073] Surface friction detection and seawater velocity detection using current sensors As the system 10 cleans the hull, motor current feedback from current sensors associated with the suction mechanism 50 can be advantageously used to detect how fouled and uneven the hull's surface is, as well as the velocity of the water or other fluid. More specifically, current is used to apply torque to rotate the non-stationary servicing element 20B. If the movement of the non-stationary servicing element 20B is affected by surface irregularities and / or seawater velocity, more torque (and therefore more current) may be required to navigate the surface irregularities or overcome the seawater velocity. Accordingly, associated current sensors may be used to provide information about the surface irregularities and / or seawater velocity. A current sensing overview is provided in FIG. 10. In some embodiments, the body of the system 10 may be used as the sensor itself. By way of example, the system 10 may include one or more sensors disposed throughout the body of the system 10 that may be configured to detect various forces applied to the body of the system 10, such as, but not limited to, suction, pressure, traction, pushing, etc. The body of system 10 may have one or more flexible structures. Each flexible structure of the body of system 10 may be equipped with force sensors, encoders, and / or other sensors that may be used to calibrate system 10 and / or provide redundant sensing.

[0074] System operation and position determination using weld line and inner rib detection As shown in FIGS. 11 and 12 , the hull 1100 may include multiple surface weld lines 1110 and, in some cases, internal rib weld protrusions 1120. Typically, the surface welds 1110 are linear protrusions that may be between about 1 mm and about 10 mm in height and range in width from about 2 mm to about 25 mm. The surface weld lines 1110 allow a user to divide the surface of the hull into (individual) segments 1150 that can be used to locate and operate the system 10. The internal rib weld protrusions 1120 provide an additional means for dividing the hull into individual sections. Dividing the hull into multiple segments 1150 (approximately 40 feet by 10 feet or less) bounded by the weld lines 1110 and / or the internal rib weld protrusions 1120 (approximately 2 foot by 2 foot sections) allows a user to clean the hull in sections or segments. Additional detection of the inner rib protrusions 1120 (eg, using minute detection of surface variations with ultrasonic, eddy current, or distance sensors) can significantly reduce errors in system operation.

[0075] Advantageously, positioning the system 10 within individual segments of the hull—rather than having the entire length and width of the vessel as the boundary of the system 10—allows a user to operate the system 10 within the weld seams 1110 of the individual segments 1150. In this way, operation errors do not propagate outside of the individual segments 1150. Indeed, the system 10 can easily detect corners 1130 of the weld seam segments 1110 to facilitate path planning ( FIG. 5 ) for back-and-forth cleaning within these segments 1150. Importantly, the actual shape and geometry of the hull within the segments 1150 do not need to be pre-mapped, as the system 10 can easily navigate around the weld seams 1110 and then plan the cleaning path.

[0076] Operations using eddy currents Referring to FIG. 13 , a further exemplary method of operation is shown. While the use of the weld seam 1110 and inner rib protrusion 1120 does not require mapping of the hull surface, ultrasonic and / or eddy currents can be used to map the hull surface and, for example, create a lookup table of specific responses at known points. Preferably, the lookup table can be stored in a memory provided for that purpose. The ultrasonic and / or eddy current responses are reflected by features such as hull curvatures, weld seams 1110, inner rib protrusions 1120, and defects (e.g., scratches, holes, dents, etc.), obstructions, etc. on the hull surface. Each feature detected using ultrasonic and / or eddy currents can then be used to identify specific locations on the hull surface.

[0077] 13, an ultrasonic and / or eddy current device 1300 may be disposed in the system 10 and adapted to emit a signal 1350 that is reflected from a free surface 1310 (e.g., the inner surface of a ship's hull). When the reflected signal encounters, for example, a weld seam 1110, the signal 1350 is reflected back to the ultrasonic and / or eddy current device 1300.

[0078] Distance sensors may also be used to detect weld lines, protrusions, defects, obstructions, etc. on the surface of the hull, enabling minute detection of surface changes. For example, distance sensors may be adapted to identify curvatures, protrusions, defects, obstructions, etc. on the surface of the hull, along with the properties (e.g., orientation, thickness, etc.) of the inner rib weld protrusion 1120. These data may provide specific and distinctive characteristics that are detectable and measurable by the sensors of the system 10. The central processing system may initially use these data to determine the location of the system 10 (i.e., the individual segment of the hull). Once the location of the system 10 is determined, the central processing system may use these data to determine the path of the system 10 to efficiently clean / maintain the surface of the hull. Advantageously, the operating signals sent from the central processing system to the system 10 may be saved and stored in memory for reuse in future cleaning / maintenance operations within the same individual segment of the hull.

[0079] Physical distance and angle sensing for position determination using hull curvature 14A and 14B, there is shown the use of a physical distance sensor (e.g., a dial indicator) 1410 in combination with multiple (e.g., two) angle sensors 1420, 1430 to determine the curvature of the hull surface at or between discrete locations on the hull surface. The distance data from the physical distance sensor 1410 and the associated rotational angle of the hull surface from each of the angle sensors 1420, 1430 may be advantageously provided to a radius of curvature lookup table, which may be stored, for example, in memory provided for this purpose. As shown in FIG. 14A, when the system 10 is placed on a flat surface, the rotational angle associated with the angle sensor 1420 incorporated into the first suction mechanism 50A and the rotational angle associated with the angle sensor 1420 incorporated into the second suction mechanism 50B are each essentially zero (0). As the system 10 translates across the surface of the hull to encounter and / or follow the curvature of the hull, the physical distance sensor 1410 may record the distance from a previous point (e.g., 22.42 mm), the angle sensor 1420 associated with the first suction mechanism 50A may record a first rotation angle (1.9 degrees), and the angle sensor 1430 associated with the second suction mechanism 50B may record a second rotation angle (1.9 degrees), and these measurements may be used (e.g., by a central processing system) to calculate a radius of curvature of 2561.5 mm. A look-up table of radius of curvature magnitudes may then be used for location purposes, enabling the central processing system to narrow down where a corresponding curvature of this magnitude is located on the surface of the hull.

[0080] Adsorption and gap detection 15, as previously described, the movement and operation of the robotic system 10 requires that the fixed first suction mechanism 50 acts as a pivot point around which the non-fixed second suction mechanism 50 rotates to clean / maintain the surface of the hull while moving. To prevent the system 10 from falling off the hull surface, validating the adequacy of the suction force between the fixed suction mechanism 50 and the hull surface is important to avoid peeling or sliding of the system 10 from the hull surface.

[0081] A first method for verifying the adequacy of the adhesive force between the fixed adhesive mechanism 50 and the surface of the hull involves the use of one or more magnetometers arranged radially around the adhesive device (e.g., magnet) 26 and adapted to measure the magnetic flux adjacent to or around the adhesive device (e.g., magnet) 26. Figure 15 shows the relationship between magnetic flux and distance or gap from the surface of 1 / 4 inch, 3 / 8 inch, and 1 / 2 inch thick steel plates, and the relationship between magnetic flux and variable thickness, while Figure 16 shows the relationship between magnetic flux and time for various gap distances (mm) between a 150 lb switchable magnet and a 1 / 2 inch thick steel plate.

[0082] Alternatively, motor current feedback can be monitored to verify the adequacy of the adhesive force between the fixed adhesive mechanism 50 and the hull surface. Indeed, when the upper magnet of the fixed adhesive mechanism 50 rotates relative to a stationary magnet (e.g., a lower one), a known torque is required to rotate the upper magnet. For example, when both the upper and lower magnets are in free space, the torque required to rotate the magnet into position is greatest. However, as the magnet approaches a ferrous surface (i.e., the hull surface), the torque decreases because the magnetic flux is directed toward the ferrous surface rather than resisting the magnet. Thus, data on the relationship between the torque and current required to move the magnet depending on the air gap between the magnet and the hull surface can be used to correlate the adequacy of the adhesive force. Advantageously, the hull thickness can be evaluated based on the current required to activate the magnet.

[0083] Machine Learning Implementation 17 illustrates an exemplary embodiment of a machine learning module 1700 that may implement one or more of the machine learning processes described herein. The machine learning module 1700 may be configured to use a machine learning process to implement various decisions, calculations, processes, etc., described in this disclosure. As used in this disclosure, a "machine learning process" is a process that automatically uses training data to generate an algorithm that calculates an output when given data as input. A machine learning process is in contrast to a non-machine learning software program in which the instructions to be executed are predetermined by a user and written in a programming language.

[0084] Still referring to FIG. 17 , machine learning module 1700 may utilize training data 1704. As used herein, “training data” refers to data containing correlations that a machine learning process may use to model relationships between two or more categories of data elements. By way of example and not limitation, training data 1704 may include multiple data entries, each representing a collection of data elements recorded, received, and / or generated together. Training data 1704 may include data elements that may be correlated by co-occurrence in a given data entry, proximity in a given data entry, or otherwise. The multiple data entries in training data 1704 may reveal one or more trends in correlations between data element categories. By way of example and not limitation, a high value of a first data element belonging to a first category of data elements tends to correlate with a high value of a second data element belonging to a second category of data elements, perhaps indicating a proportionality or other mathematical relationship linking values ​​belonging to the two categories.

[0085] Multiple categories of data elements may be related to the training data 1704 by various correlations. Correlations may refer to causal and / or predictive links between categories of data elements and may be modeled as relationships, such as mathematical relationships, by machine learning processes, described in more detail below. The training data 1704 may be formatted and / or organized by categories of data elements. For example, the training data 1704 may be organized by associating data elements with one or more descriptors that correspond to the categories of the data element. As a non-limiting example, the training data 1704 may include data entered by one or more people on a standard form, such that the entry of a given data element in a given field on the form may be mapped to one or more descriptors in the category. Elements of the training data 1704 may be linked to descriptors in the category by tags, tokens, or other data elements. The training data 1704 may be provided in a fixed-length format, such as a comma-separated value (CSV) format and / or a self-describing format, which is a format that links the location of data to a category. Self-describing formats may include, but are not limited to, Extensible Markup Language (XML), JavaScript Object Notation (JSON), and others, which allow a process or device to discover the category of data.

[0086] Continuing with reference to FIG. 17 , training data 1704 may include one or more uncategorized elements. Uncategorized data in training data 1704 may include data that is unformatted or does not include descriptors for some elements of the data. In some embodiments, machine learning algorithms and / or other processes may classify training data 1704 according to one or more categorizations. Machine learning algorithms may classify training data 1704 using, for example, natural language processing algorithms, tokenization, detecting correlations in raw data, etc. In some embodiments, categories of training data 1704 may be generated using correlation and / or other processing algorithms. As a non-limiting example, in a body of text, phrases composed of several “n” compound words, such as nouns modified by other nouns, may be identified according to statistically significant occurrences of n-grams containing such words in a particular order. For example, n-grams may be categorized as linguistic elements, such as “words,” that are tracked similarly to single words, which may generate new categories as a result of statistical analysis. In a data entry that includes some text data, a person's name may be identified by reference to a list, dictionary, or other collection of words, allowing for on-the-fly categorization by a machine learning algorithm and / or automatic association of the data entry's data with a descriptor or to a given format. The ability to automatically categorize data entries may allow the same training data 1704 to be applied to two or more separate machine learning algorithms, as described in more detail below. The training data 1704 used by the machine learning module 1700 may correlate input data as described in this disclosure with output data as described in this disclosure, without limitation.

[0087] With further reference to FIG. 17 , training data 1704 may be filtered, classified, and / or selected using one or more supervised and / or unsupervised machine learning processes and / or models, as described in more detail below. In some embodiments, training data 1704 may be classified using training data classifier 1716. Training data classifier 1716 may include a classifier. A "classifier," as used in this disclosure, is a machine learning model that classifies inputs into one or more categories. Training data classifier 1716 may utilize a mathematical model, neural net, or program generated by a machine learning algorithm. The machine learning algorithm of training data classifier 1716 may include a classification algorithm. A "classification algorithm," as used in this disclosure, is one or more computer processes that generate a classifier from training data. A classification algorithm may classify inputs into categories and / or bins of data. The classification algorithm may output categories of data and / or labels associated with the data. A classifier may be configured to output data that label or identify data sets that can be clustered together. The machine learning module 1700 may generate a classifier, such as training data classifier 1716, that uses a classification algorithm. Classification may be performed using, but is not limited to, linear classifiers such as logistic regression and / or naive Bayes classifiers, nearest neighbor classifiers such as ask nearest neighbor classifiers, support vector machines, least squares support vector machines, Fisher's linear discriminant, quadratic classifiers, decision trees, boosting trees, random forest classifiers, learning vector quantization, and / or neural network-based classifiers. As a non-limiting example, the training data classifier 1716 may classify elements of the sensor data into adhesion levels.

[0088] Still referring to FIG. 17 , the machine learning module 1700 can be configured to implement a lazy learning process 1720. The lazy learning process 1720 can include a “lazy load” or “call-on-demand” process and / or protocol. The “lazy learning process” can include a process in which machine learning is performed upon receipt of input that is converted into output by combining the input with a training set and deriving an algorithm that is used to generate output on demand. By way of example, an initial simulation set can be implemented to correspond to an initial heuristic and / or “first guess” in the output and / or relationships. As a non-limiting example, the initial heuristic can include ranking associations between the input and elements of the training data 1704. The heuristic can include selecting some number of the highest-ranked associations and / or training data 1704 elements. Lazy learning may implement any suitable lazy learning algorithm, including, but not limited to, a K-nearest neighbor algorithm, a lazy naive Bayes algorithm, and others; upon reviewing this disclosure as a whole, those skilled in the art will be aware of a variety of lazy learning algorithms that may be applied to generate the outputs described in this disclosure, including, but not limited to, the lazy learning applications of machine learning algorithms described further below.

[0089] Still referring to FIG. 17 , the machine learning processes described in this disclosure can be used to generate a machine learning model 1724. As used in this disclosure, a “machine learning model” is a mathematical and / or algorithmic representation of a relationship between inputs and outputs that is generated and stored in memory using a machine learning process, including, but not limited to, the processes described above. By way of example, inputs may be sent to the machine learning model 1724, which, once created, may generate an output according to the derived relationship. By way of example and not limitation, a linear regression model generated using a linear regression algorithm may calculate a linear combination of input data using coefficients derived during the machine learning process to compute an output. As a further non-limiting example, the machine learning model 1724 can be generated by creating an artificial neural network, such as a convolutional neural network, that includes an input layer of nodes, one or more hidden layers, and an output layer of nodes. Connections between nodes are created through a process of "training" the network, where elements from a training data 1704 set are applied to input nodes, and an appropriate training algorithm (such as Levenberg-Marquardt, conjugate gradient, simulated annealing, or other algorithms) is used to adjust the connections and weights between nodes in adjacent layers of the neural network to produce desired values ​​at the output nodes. This process is sometimes referred to as deep learning.

[0090] Still referring to FIG. 17 , the machine learning algorithm may include a supervised machine learning process 1728. As used herein, a “supervised machine learning process” refers to one or more algorithms that accept labeled input data and generate an output according to the labeled input data. By way of example, the supervised machine learning process 1728 may include the sensor data described above as input, a lateral movement path as output, and a scoring function that represents a desired relationship between the input and the output. The scoring function may maximize the probability that a given input and / or combination of input elements is associated with a given output, so as to minimize the probability that a given input is not associated with the given output. The scoring function may be expressed as a risk function that represents the “expected loss” of the algorithm relating inputs to outputs, where loss is calculated as an error function that represents the degree to which the prediction generated by the relationship is incorrect when compared to a given input-output pair contained in the training data 1704. Upon reviewing this entire disclosure, those skilled in the art will recognize various possible variations of at least the supervised machine learning process 1728 that may be used to determine the relationship between inputs and outputs. The supervised machine learning process may include the classification algorithm defined above.

[0091] 17 , the machine learning process may include an unsupervised machine learning process 1732. As used in this disclosure, an “unsupervised machine learning process” is a process that calculates relationships between one or more datasets without labeled training data. The unsupervised machine learning process 1732 may freely discover structures, relationships, and / or correlations contained in the training data 1704. The unsupervised machine learning process 1732 does not require a response variable. The unsupervised machine learning process 1732 may calculate patterns, inferences, correlations, etc. between two or more variables in the training data 1704. In some embodiments, the unsupervised machine learning process 1732 may determine the degree of correlation between two or more elements of the training data 1704.

[0092] Still referring to FIG. 17 , the machine learning module 1700 can be designed and configured to create the machine learning model 1724 using techniques for developing linear regression models. The linear regression model can include ordinary least squares regression, which attempts to minimize the square of the difference between predicted and actual results according to an appropriate criterion (e.g., a vector space distance criterion) for measuring such a difference, and the coefficients of the resulting linear equation can be modified to improve the minimization. The linear regression model can include a ridge regression method, in which the function being minimized includes a least squares function plus a term that multiplies the square of each coefficient by a scalar to penalize large coefficients. The linear regression model can include a least absolute value shrinkage selection operator (lasso) model, in which ridge regression is combined with least squares terms multiplied by a coefficient of 1 and divided by twice the number of samples. The linear regression model can include a multitasking lasso model, in which the criterion applied to the least squares terms of the lasso model is the Frobenius criterion, which is the square root of the sum of the squares of all terms. The linear regression model may include an elastic net model, a multitask elastic net model, a least angle regression model, a LARS lasso model, an orthogonal matching pursuit model, a Bayesian regression model, a logistic regression model, a stochastic gradient descent model, a perceptron model, a reluctant aggressive algorithm, a robustness regression model, a Huber regression model, or other suitable models that may occur to one of ordinary skill in the art upon review of this entire disclosure. The linear regression model may, in embodiments, be generalized to a polynomial regression model, whereby a polynomial (e.g., quadratic, cubic, or higher order equation) that provides the best predicted output / actual output fit is found. As will be apparent to one of ordinary skill in the art upon review of this entire disclosure, methods similar to those described above may be applied to minimize an error function.

[0093] With continued reference to FIG. 17 , the machine learning algorithm may include, but is not limited to, linear discriminant analysis. The machine learning algorithm may include quadratic discriminant analysis. The machine learning algorithm may include kernel ridge regression. The machine learning algorithm may include support vector machines, including, but not limited to, regression processes based on support vector classification. The machine learning algorithm may include stochastic gradient descent algorithms, including classification and regression algorithms based on stochastic gradient descent. The machine learning algorithm may include nearest neighbor algorithms. The machine learning algorithm may include various forms of latent space regularization, such as variational regularization. The machine learning algorithm may include Gaussian processes, such as Gaussian process regression. The machine learning algorithm may include cross-decomposition algorithms, including partial least squares and / or canonical correlation analysis. The machine learning algorithm may include naive Bayes methods. The machine learning algorithm may include decision tree-based algorithms, such as decision tree classification or regression algorithms. The machine learning algorithm may include ensemble methods, such as bagging meta-estimators, forests of randomized trees, Adaboost, gradient tree boosting, and / or voting classification methods. The machine learning algorithm may include a neural net algorithm, including a convolutional neural net process.

[0094] In the foregoing description, for purposes of explanation, specific terms were used to provide a thorough understanding of the invention. However, it will be apparent to one skilled in the art that specific details are not required to practice the invention. Thus, the foregoing description of specific embodiments of the invention has been presented for purposes of illustration and description. It is not intended to be exhaustive or to limit the invention to the precise form disclosed. Obviously, many modifications and variations are possible in light of the above teachings. The embodiments were chosen and described to best explain the principles of the invention and its practical application, thereby enabling others skilled in the art to best utilize the invention, including various modifications, and various embodiments, as suited to the particular uses contemplated. Various combinations and arrangements of the disclosed embodiments are possible. It is intended that the following claims and their equivalents define the scope of the invention. [Explanation of symbols]

[0095] 10 Submersible Self-Propulsion System 11 Heatsink Features 12 First housing part 14 Second housing part 15 slip ring 16 communication ports 17 Arched section 18 parts 20A, 20B Maintenance Elements 22 Board part 24 Rotating Brush 26 Magnetic adsorption device 28 Mounting or fastening devices 30 circular connection ports 31 Light-emitting elements 32 Cylindrical Mooring Port 33 Imaging device 34 Bellows 35 Opening part 37 Detent 38 bottom 39 Docking Guidance System 40 Mooring pole 41 Bottom part 42 Long part 43 Adsorption device 44 proximal end 46 distal end 48 Opening 50 Adsorption mechanism 50A First suction mechanism 50B 2nd suction mechanism 51 Magnet switching motor 52 Housing 53 Body rotation motor 54 Inner part 55 Magnetic Switching Shaft 56 Switch Bar 57 Planetary Mechanical Systems 58 Planetary Ring 59 Interconnecting Gear 60 Communication Cable 70 Hull 75 Operation method 81 Central Processing System 82 Detection Device 83 memory 84 Suction mechanism controller 85 Communication Links 86 Imaging device 87 Irradiation device 88 Power supply 89 Bus 100 Mooring System 1110 welding line 1120 Protrusion 1130 Corner 1150 segments 1300 Ultrasonic and / or Eddy Current Devices 1350 signal 1410 Physical Distance Sensor 1420 Angle Sensor 1430 Angle Sensor 1700 Machine Learning Module 1704 training data 1708 Output 1712 Input 1716 Training Data Classifier 1720 Lazy Learning Process 1724 machine learning models 1728 Supervised Machine Learning Process 1732 Unsupervised Machine Learning Process

Claims

1. 1. A submersible robotic system for inspecting and / or servicing an object in a marine environment, comprising: Housing and a sensor configured to generate sensor data; a processor disposed in the housing and in communication with the sensor; a plurality of suction mechanisms disposed within the housing, each suction mechanism in communication with the processor; a plurality of magnetically attractive devices constructed and arranged to secure the system to the object; a magnetic switch motor for switching the magnetic attraction device on and off; at least one maintenance element; a body rotation motor for moving the servicing element across the surface of the object; an adsorption mechanism comprising: Equipped with the processor is configured to command the magnetic switch motor of each attraction mechanism based on the sensor data, and the command of the magnetic switch motor of each attraction mechanism causes the system to move laterally across the surface of the object. Submersible robot system.

2. 2. The submersible robotic system of claim 1, wherein the processor is configured to adjust a magnetic field of at least one magnetic attraction device of the plurality of magnetic attraction devices to adjust attraction of the system to the surface of the object.

3. 2. The submersible robotic system of claim 1, wherein the magnetic switch is configured to alternatively turn on and off at least two magnetic attraction devices.

4. 4. The submersible robotic system of claim 3, wherein the at least two magnetic attraction devices include a first pivot axis and a second pivot axis, and the system moves laterally across the surface of the object through alternative switching of the first pivot axis and the second pivot axis on and off.

5. The submersible robotic system of claim 1 , wherein the processor is configured to generate a temporal mapping of the object based on the sensor data.

6. The submersible robotic system of claim 1 , wherein the processor is further configured to adjust a program path based on the sensor data.

7. 10. The submersible robotic system of claim 1, further comprising an illumination device configured to illuminate an environment of the system.

8. The submersible robotic system of claim 1 , wherein the sensor is an imaging device.

9. The submersible robotic system of claim 1 , wherein the object is a vessel.

10. The submersible robotic system of claim 1 , wherein the sensor is configured to detect a magnetic field, and the processor is configured to determine a weld seam on the object based on data generated by the sensor.

11. 1. A method for inspecting and / or servicing an object in a marine environment, comprising: providing a submersible robotic system for servicing the object, the robotic system comprising: Housing and a sensor configured to generate sensor data; a processor disposed in the housing and in communication with the sensor; a plurality of suction mechanisms disposed within the housing, each suction mechanism in communication with the processor; a plurality of magnetically attractive devices constructed and arranged to secure the system to the object; a magnetic switch motor for switching the magnetic attraction device on and off; at least one maintenance element; a body rotation motor for moving the servicing element across the surface of the object; an adsorption mechanism comprising: and operating the submersible robotic system on the object to maintain the object; and A method that encompasses

12. The method of claim 11 , wherein the manipulating comprises alternatively turning on and off at least two magnetically attractive devices.

13. The method of claim 11 , further comprising cleaning the surface of the object with the at least one servicing element.

14. The method of claim 11 , further comprising generating, by the processor, a temporal mapping of the object based on the sensor data.

15. The method of claim 11 , further comprising adjusting, by the processor, a program path based on the sensor data.

16. The method of claim 11 , wherein the submersible robotic system further comprises an illumination device configured to illuminate an environment of the system.

17. The method of claim 11 , wherein the sensor is an imager.

18. The method of claim 11 , wherein the object is a watercraft.

19. The operation further comprises: determining a level of cleanliness of the surface of the object; adjusting the movement of the at least one servicing element based on the determined cleanliness level of the surface of the object; The method of claim 11 , comprising:

20. The operation further comprises: detecting eddy currents with the sensor; determining, with the processor, a weld seam on the object based on the eddy currents; The method of claim 11 , comprising:

Images