Description:FIELD OF INVENTION
[0001] The present disclosure relates to automated inspection and cleaning systems, and more particularly to an artificial intelligence-driven system and method for automated inspection and cleaning of internal surfaces of conduits.
BACKGROUND
[0002] Conduits, including those used in industrial systems, residential plumbing networks, and projectile‑launching devices, are subject to internal surface degradation over time due to continuous operational stresses and repeated use. The internal surfaces of such conduits accumulate various forms of contamination and damage, including rust formation, carbon deposits, copper fouling, foreign material buildup, and structural defects such as cracks and pitting. These conditions can adversely affect the performance and efficiency of the conduit system and may impact overall operational readiness of the equipment in which the conduit is integrated. Conventional maintenance practices typically involve periodic inspection and cleaning to address these internal surface conditions.
[0003] Manual barrel cleaning methods rely on human operators using rods, brushes, and cleaning solvents to remove deposits from the internal barrel surface. These manual approaches are labour-intensive and produce inconsistent results that depend on operator skill and judgment. The time required for manual cleaning of fused copper fouling and other stubborn deposits can extend to several hours or even days. Visual inspection of barrel interiors performed by human operators is inherently subjective, as different operators may assess the same surface condition differently, leading to variability in maintenance decisions.
[0004] Existing automated systems, such as crawler-based robots, can traverse the internal bore of a barrel to perform cleaning and inspection operations. However, these systems typically operate at low traversal speeds and require extended cleaning cycles, often in the range of eight to ten hours for fused copper fouling removal. The crawler mechanisms employed in existing systems generate high mechanical drag due to friction from tracks or wheels contacting the barrel surface.
[0005] Current automated inspection approaches in barrel maintenance systems provide limited defect detection capabilities as such approaches still rely on human to understand and detect defects inside the conduit. Many existing systems lack the ability to perform precise spatial localization of detected surface conditions along the barrel length. Without accurate positional mapping of defects, maintenance operations cannot be targeted to specific affected regions, resulting in uniform cleaning approaches that treat the entire barrel surface regardless of actual condition. This uniform treatment approach leads to unnecessary cleaning in unaffected regions, increased consumption of cleaning fluids and consumables, and extended maintenance cycle times.
[0006] Existing barrel inspection systems that incorporate imaging capabilities often rely on basic image analysis techniques that do not provide quantified measurements of defect severity, coverage area, or geometric parameters. The absence of pixel-level defect segmentation and quantification limits the ability to generate objective, repeatable assessments of barrel condition. Furthermore, existing systems typically do not integrate inspection outputs with cleaning control logic, preventing the execution of targeted cleaning operations based on detected defect locations and severity levels.
[0007] Digital record-keeping and traceability of barrel inspection data presents another challenge in current maintenance practices. Manual inspection methods produce subjective assessments that are difficult to standardize and compare over time. Even automated systems may lack structured reporting capabilities that enable historical comparison of barrel condition across multiple inspection cycles, limiting the ability to track barrel degradation patterns and make informed lifecycle management decisions.
[0008] In view of the above-mentioned problem, it is desirable to develop a system and method that overcomes all the above-mentioned problems.
SUMMARY
[0009] This summary is provided to introduce a selection of concepts, in a simplified format, that are further described in the detailed description of the invention. This summary is neither intended to identify key or essential inventive concepts of the invention nor is it intended for determining the scope of the invention.
[0010] In a first aspect, a system for automated inspection and cleaning of an inner surface of a conduit is provided. The system comprises: a crawler configured to traverse longitudinally within the conduit along an internal diameter of the conduit; an imaging unit mounted on the crawler, the imaging unit configured to capture image data of the inner surface of the conduit; a position tracking mechanism configured to generate axial position data of the crawler during traversal within the conduit, a cleaning mechanism coupled to the crawler and configured to perform cleaning operations on the inner surface of the conduit; a fluid dispensing unit configured to dispense cleaning fluid assisting the cleaning operations; and a processing unit communicatively coupled to the imaging unit, the position tracking mechanism, the cleaning mechanism, and the fluid dispensing unit, the processing unit comprising an image processing module configured to: receive the image data captured by the imaging unit; detect surface anomalies on the inner surface of the conduit based on analysis of the image data, wherein the surface anomalies comprise at least one of rust, carbon deposits, copper fouling, foreign deposits, or cracks; classify the surface anomalies on the inner surface of the conduit; perform micro level segmentation of detected surface anomalies to quantify defect parameters comprising of defect area; correlate detected surface anomalies with corresponding axial position data from the position tracking mechanism to generate spatially localised inspection outputs; wherein the processing unit is further configured to control the cleaning mechanism and the fluid dispensing unit to execute cleaning operations on the inner surface of the conduit.
[0011] In a second aspect, a method for automated inspection and cleaning of an inner surface of a conduit is provided. The method comprises: traversing a crawler longitudinally within the conduit along an internal diameter of the conduit; capturing, by an imaging unit mounted on the crawler, image data of the inner surface of the conduit during traversal; generating, by a position tracking mechanism, axial position data of the crawler during traversal; processing, by an image processing module, the captured image data to detect and classify surface anomalies on the inner surface of the projectile launching device, wherein the surface anomalies comprise at least one of rust, carbon deposits, copper fouling, foreign deposits, and cracks; performing pixel-level segmentation of detected surface anomalies to quantify defect parameters comprising at least one of defect area, coverage percentage, and severity; correlating detected surface anomalies with corresponding axial position data to generate spatially localised inspection outputs; and executing cleaning operations on the inner surface of the conduit.
[0012] To further clarify the advantages and features of the present invention, a more particular description of the invention will be rendered by reference to specific embodiments thereof, which are illustrated in the appended drawings. It is appreciated that these drawings depict only typical embodiments of the invention and are therefore not to be considered limiting of its scope. The invention will be described and explained with additional specificity and detail in the accompanying drawings.
BRIEF DESCRIPTION OF FIGURES
[0013] These and other features, aspects, and advantages of the present invention will become better understood when the following detailed description is read with reference to the accompanying drawings in which like characters represent like parts throughout the drawings, wherein:
[0014] Figure 1 illustrates a block diagram of a system for automated inspection and cleaning of a conduit, according to aspects of the present disclosure.
[0015] Figure 2 illustrates a block diagram of a processing unit of the system of Figure 1, according to aspects of the present disclosure.
[0016] Figure 3 illustrates a block diagram of an image processing module of the processing unit of Figure 2, according to aspects of the present disclosure.
[0017] Figure 4 illustrates a flowchart of a method for inspecting and cleaning an inner surface of a conduit, according to aspects of the present disclosure.
[0018] Further, skilled artisans will appreciate that elements in the drawings are illustrated for simplicity and may not have necessarily been drawn to scale. For example, the flow charts illustrate the method in terms of the most prominent steps involved to help improve understanding of aspects of the present invention. Furthermore, in terms of the construction of the device, one or more components of the device may have been represented in the drawings by conventional symbols, and the drawings may show only those specific details that are pertinent to understanding the embodiments of the present invention so as not to obscure the drawings with details that will be readily apparent to those of ordinary skill in the art having benefit of the description herein.
DETAILED DESCRIPTION
[0019] For the purpose of promoting an understanding of the principles of the invention, reference will now be made to the embodiment illustrated in the drawings and specific language will be used to describe the same. It will nevertheless be understood that no limitation of the scope of the invention is thereby intended, such alterations and further modifications in the illustrated system, and such further applications of the principles of the invention as illustrated therein being contemplated as would normally occur to one skilled in the art to which the invention relates. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skilled in the art to which this invention belongs. The system, methods, and examples provided herein are illustrative only and not intended to be limiting.
[0020] For example, the term “some” as used herein may be understood as “none” or “one” or “more than one” or “all.” Therefore, the terms “none,” “one,” “more than one,” “more than one, but not all” or “all” would fall under the definition of “some.” It should be appreciated by a person skilled in the art that the terminology and structure employed herein is for describing, teaching, and illuminating some embodiments and their specific features and elements and therefore, should not be construed to limit, restrict or reduce the spirit and scope of the present disclosure in any way.
[0021] For example, any terms used herein such as, “includes,” “comprises,” “has,” “consists,” and similar grammatical variants do not specify an exact limitation or restriction, and certainly do not exclude the possible addition of one or more features or elements, unless otherwise stated. Further, such terms must not be taken to exclude the possible removal of one or more of the listed features and elements, unless otherwise stated, for example, by using the limiting language including, but not limited to, “must comprise” or “needs to include.”
[0022] Whether or not a certain feature or element was limited to being used only once, it may still be referred to as “one or more features” or “one or more elements” or “at least one feature” or “at least one element.” Furthermore, the use of the terms “one or more” or “at least one” feature or element do not preclude there being none of that feature or element, unless otherwise specified by limiting language including, but not limited to, “there needs to be one or more...” or “one or more element is required.”
[0023] Unless otherwise defined, all terms and especially any technical and/or scientific terms, used herein may be taken to have the same meaning as commonly understood by a person ordinarily skilled in the art.
[0024] Reference is made herein to some “embodiments.” It should be understood that an embodiment is an example of a possible implementation of any features and/or elements of the present disclosure. Some embodiments have been described for the purpose of explaining one or more of the potential ways in which the specific features and/or elements of the proposed disclosure fulfil the requirements of uniqueness, utility, and non-obviousness.
[0025] Use of the phrases and/or terms including, but not limited to, “a first embodiment,” “a further embodiment,” “an alternate embodiment,” “one embodiment,” “an embodiment,” “multiple embodiments,” “some embodiments,” “other embodiments,” “further embodiment”, “furthermore embodiment”, “additional embodiment” or other variants thereof do not necessarily refer to the same embodiments. Unless otherwise specified, one or more particular features and/or elements described in connection with one or more embodiments may be found in one embodiment, or may be found in more than one embodiment, or may be found in all embodiments, or may be found in no embodiments. Although one or more features and/or elements may be described herein in the context of only a single embodiment, or in the context of more than one embodiment, or in the context of all embodiments, the features and/or elements may instead be provided separately or in any appropriate combination or not at all. Conversely, any features and/or elements described in the context of separate embodiments may alternatively be realized as existing together in the context of a single embodiment.
[0026] Any particular and all details set forth herein are used in the context of some embodiments and therefore should not necessarily be taken as limiting factors to the proposed disclosure.
[0027] Embodiments of the present invention will be described below in detail with reference to the accompanying drawings.
[0028] Figure 1 illustrates a block diagram of a system 100 for automated inspection and cleaning of a conduit, according to aspects of the present disclosure. The system 100 is configured to be deployed for automated inspection and cleaning of conduits in various applications. In one implementation, the system 100 is configured for cleaning conduits such as barrels of projectile launching devices. Projectile launching devices include gun barrels of various calibres, wherein the internal bore surface accumulates deposits, fouling, and surface degradation during operation. The system 100 addresses maintenance requirements for such projectile launching devices by providing automated inspection and cleaning capabilities within the barrel interior.
[0029] In another implementation, the system 100 is configured for cleaning conduits such as industrial pipes or residential pipes. Industrial pipes include pipelines used in manufacturing, processing, or transport applications where internal surface conditions affect operational performance. Residential pipes include plumbing conduits where internal deposits, corrosion, or contamination affect fluid flow or water quality. The system 100 is adaptable to inspect and clean the inner surfaces of such pipe structures.
[0030] The system 100 comprises components configured to traverse, inspect, and clean the inner surface of the conduit. The system 100 integrates imaging, position tracking, cleaning, and fluid dispensing functions within a coordinated platform that enables both inspection and cleaning operations to be performed within the conduit interior.
[0031] The system 100 includes a crawler 102 configured to traverse longitudinally within the conduit along an internal diameter of the conduit. The crawler 102 is introduced into the conduit and moves along a longitudinal axis of the conduit to enable inspection and cleaning operations to be performed across the full length of the conduit interior.
[0032] The crawler 102 comprises a linear axial carriage with a low-drag architecture. The low-drag architecture reduces mechanical resistance during movement within the conduit, enabling high-speed axial traversal. The linear axial carriage configuration provides controlled and repeatable movement along the longitudinal axis of the conduit. The low-drag structural layout of the crawler 102 reduces friction compared to tracked or wheeled crawler designs, thereby enabling faster traversal speeds during both inspection and cleaning operations.
[0033] The system 100 is scalable for the conduit having diameters ranging from 76 mm to 155 mm. The same design of the crawler 102 is scaled appropriately to accommodate conduits within this diameter range. The scalability enables the system 100 to be deployed for inspection and cleaning of conduits of varying internal diameters without requiring fundamentally different crawler configurations.
[0034] The system 100 supports interchangeable cleaning and inspection heads that are mountable on the crawler 102. The interchangeable heads enable the crawler 102 to be configured for different operational requirements, including inspection-focused operations, cleaning-focused operations, or combined inspection and cleaning operations. The interchangeable head configuration provides flexibility in adapting the system 100 to specific maintenance tasks and conduit conditions.
[0035] Further, the system 100 includes an imaging unit 104 mounted on the crawler 102. The imaging unit 104 is configured to capture image data of the inner surface of the conduit during traversal of the crawler 102 within the conduit interior. In an embodiment, the imaging unit 104 comprises at least one camera and at least one illumination source. The at least one camera comprises a high-resolution camera configured to acquire detailed image data of the internal surface of the conduit. The at least one illumination source provides controlled illumination that enables clear capture of the internal surface under varying surface reflectivity and contamination conditions. The controlled illumination compensates for variations in surface reflectivity that occur due to different surface materials, coatings, or contamination states present on the inner surface of the conduit. The combination of the high-resolution camera and controlled illumination enables the imaging unit 104 to acquire image data of sufficient quality for subsequent analysis regardless of the surface condition encountered during traversal.
[0036] The imaging unit 104 captures image data either continuously or at predefined positional intervals as the crawler 102 progresses through the conduit. In continuous capture mode, the imaging unit 104 acquires image data throughout the traversal of the crawler 102, providing comprehensive coverage of the inner surface along the full length of the conduit. In interval-based capture mode, the imaging unit 104 acquires image data at predefined positional intervals, enabling targeted image acquisition at specific locations within the conduit. The selection between continuous capture and interval-based capture is determined based on inspection requirements and operational parameters for a given inspection operation.
[0037] The system 100 includes a position tracking mechanism 106 configured to generate axial position data of the crawler 102 during traversal within the conduit. The position tracking mechanism 106 monitors the movement of the crawler 102 and records precise axial position data throughout the inspection process. The axial position data generated by the position tracking mechanism 106 indicates the location of the crawler 102 along the longitudinal axis of the conduit at any given time during traversal.
[0038] In an embodiment, the position tracking mechanism 106 is embodied as an encoder-based position tracking mechanism. The encoder-based position tracking mechanism generates position data by detecting rotational or linear displacement of components associated with the movement of the crawler 102. The encoder-based implementation provides accurate and repeatable position measurements that enable precise determination of the axial location of the crawler 102 within the conduit interior. The position tracking mechanism 106 continuously captures encoder readings during traversal of the crawler 102. The continuous capture of encoder data enables real-time tracking of the position of the crawler 102 as the crawler 102 moves through the conduit. The encoder readings are used to calculate the precise distance travelled by the crawler 102 from a reference point, such as the entry point of the conduit, thereby providing positional coordinates along the length of the conduit.
[0039] The axial position data generated by the position tracking mechanism 106 enables spatial localisation of detected surface conditions along the conduit length. By correlating image data captured by the imaging unit 104 with corresponding axial position data from the position tracking mechanism 106, the system 100 generates spatially localised inspection outputs that associate detected anomalies with specific locations within the conduit. The encoder-linked position tracking enables position-specific inspection and maintenance actions to be performed based on the precise location of detected surface conditions.
[0040] The system 100 includes a fluid dispensing unit 108 configured to dispense cleaning fluid, also referred to as cleaning media, assisting the cleaning operations. The fluid dispensing unit 108 is coupled to the crawler 102 and operates in coordination with the cleaning mechanism 110 to deliver cleaning media to the inner surface of the conduit during cleaning operations. In an embodiment, the fluid dispensing unit 108 comprises a pump mechanism configured to dispense at least one of cleaning fluid or oil. The pump mechanism provides controlled delivery of cleaning media to the inner surface of the conduit. The cleaning fluid comprises solutions formulated to dissolve, loosen, or remove deposits such as carbon fouling, copper fouling, rust, and other contaminants from the inner surface. The oil comprises lubricating or protective fluids that are applied to the inner surface following cleaning operations or as part of maintenance procedures. The pump mechanism enables precise control over the volume and rate of cleaning fluid or oil dispensed during cleaning operations.
[0041] The system 100 includes a processing unit 112 configured to synchronise pump activation with axial position data from the position tracking mechanism 106. The synchronisation between pump activation and axial position data enables the fluid dispensing unit 108 to dispense cleaning fluid or oil at specific locations within the conduit based on the current position of the crawler 102. The processing unit 112 receives axial position data from the position tracking mechanism 106 and controls the pump mechanism of the fluid dispensing unit 108 to activate dispensing at appropriate times and locations during traversal of the crawler 102.
[0042] The synchronisation of pump activation with axial position data enables targeted application of cleaning media at locations where cleaning is required. In full-coverage cleaning mode, the pump mechanism is activated to dispense cleaning fluid or oil uniformly along the entire length of the conduit as the crawler 102 traverses the conduit interior. In inspection-guided selective cleaning mode, the pump mechanism is selectively activated only at axial positions corresponding to detected surface anomalies, thereby avoiding unnecessary dispensing of cleaning media in unaffected regions of the conduit. The duration and intensity of pump activation are governed by the processing unit 112 based on operational parameters and, where applicable, severity metrics associated with detected surface anomalies at corresponding axial positions.
[0043] The system 100 includes a cleaning mechanism 110 coupled to the crawler 102 and configured to perform cleaning operations on the inner surface of the conduit. The cleaning mechanism 110 is mounted on the crawler 102 and operates in coordination with the fluid dispensing unit 108 to remove deposits, fouling, and contaminants from the inner surface of the conduit during cleaning operations.
[0044] The cleaning mechanism 110 comprises components configured to mechanically engage with the inner surface of the conduit to dislodge and remove surface deposits. The cleaning mechanism 110 is configured to remove fused copper deposits, carbon fouling, rust formations, and other contaminants that accumulate on the inner surface of the conduit during operation. The configuration of the cleaning mechanism 110 is optimised for rapid removal of fused deposits, enabling completion of cleaning operations within reduced timeframes compared to existing automated solutions.
[0045] The cleaning mechanism 110 is activated uniformly across the conduit length or selectively at specific locations based on inspection results. In the full-coverage cleaning mode, the cleaning mechanism 110 operates continuously as the crawler 102 traverses the full length of the conduit, providing comprehensive cleaning coverage of the entire inner surface. In the inspection-guided selective cleaning mode, the cleaning mechanism 110 is activated only at specific axial positions within the conduit where surface anomalies have been detected during inspection operations.
[0046] The selective activation of the cleaning mechanism 110 is controlled by the processing unit 112 based on spatially localised inspection outputs. The processing unit 112 references the axial position data from the position tracking mechanism 106 and compares the current position of the crawler 102 with recorded anomaly locations from inspection operations. When the crawler 102 reaches an axial position corresponding to a detected surface anomaly, the processing unit 112 activates the cleaning mechanism 110 to perform cleaning operations at that location. The cleaning mechanism 110 remains inactive when the crawler 102 traverses regions of the conduit where no surface anomalies were detected, thereby avoiding unnecessary cleaning actions in unaffected regions.
[0047] The selective activation of the cleaning mechanism 110 based on inspection results reduces wear on the cleaning mechanism 110 and minimises consumption of cleaning media dispensed by the fluid dispensing unit 108. The targeted cleaning approach enabled by selective activation provides effective treatment of identified defect or deposit regions while preserving unaffected portions of the inner surface.
[0048] The processing unit 112 is communicatively coupled to the imaging unit 104, the position tracking mechanism 106, the cleaning mechanism 110, and the fluid dispensing unit 108. The processing unit 112 manages the operation of the system 100 by receiving data from the imaging unit 104 and the position tracking mechanism 106, and by controlling the cleaning mechanism 110 and the fluid dispensing unit 108 based on operational parameters and inspection outputs.
[0049] The communicative coupling between the processing unit 112 and the imaging unit 104 enables the processing unit 112 to receive image data captured by the imaging unit 104 during traversal of the crawler 102 within the conduit. The processing unit 112 processes the received image data to detect and classify surface anomalies on the inner surface of the conduit.
[0050] The communicative coupling between the processing unit 112 and the position tracking mechanism 106 enables the processing unit 112 to receive axial position data generated by the position tracking mechanism 106. The processing unit 112 correlates the received image data with corresponding axial position data to generate spatially localised inspection outputs that associate detected surface anomalies with specific locations within the conduit.
[0051] The communicative coupling between the processing unit 112 and the cleaning mechanism 110 enables the processing unit 112 to control activation of the cleaning mechanism 110 during cleaning operations. The processing unit 112 activates the cleaning mechanism 110 uniformly along the conduit length or selectively at specific axial positions based on inspection results.
[0052] The communicative coupling between the processing unit 112 and the fluid dispensing unit 108 enables the processing unit 112 to control dispensing of cleaning fluid or oil during cleaning operations. The processing unit 112 synchronises pump activation of the fluid dispensing unit 108 with axial position data from the position tracking mechanism 106 to enable targeted or uniform application of cleaning media.
[0053] The system 100 communicates with a remote electronic device 124 via a network 126. The network 126 provides a communication pathway between the system 100 and the remote electronic device 124, enabling data exchange between the processing unit 112 and the remote electronic device 124. The network 126 comprises a wireless or wired network connection that supports transmission of inspection data, operational parameters, and control instructions between the system 100 and the remote electronic device 124.
[0054] The remote electronic device 124 receives inspection data, reports, and operational information from the system 100 via the network 126. The remote electronic device 124 enables remote monitoring of conduit inspection and cleaning operations by providing access to real-time operational status and inspection outputs generated by the processing unit 112. The remote electronic device 124 provides data storage capabilities for inspection records, enabling historical comparison and lifecycle traceability of conduit inspection and maintenance activities. The remote electronic device 124 enables further analysis of conduit inspection and cleaning operations by providing computational resources or user interfaces for reviewing and processing inspection data transmitted from the system 100.
[0055] Figure 2 illustrates a block diagram of the processing unit 112 of the system 100 of Figure 1, according to aspects of the present disclosure. The processing unit 112 includes a processor 202 and a memory 204. The memory 204 contains data 206 and modules 208. The processor 202 is configured to execute instructions stored in the memory 204 to perform the operations of the processing unit 112. The processor 202 comprises one or more processing elements configured to perform computational operations associated with image processing, position tracking correlation, cleaning control, and report generation functions of the system 100.
[0056] The processor 202 is implemented using various processing architectures. In one implementation, the processor 202 comprises a microprocessor configured to execute software instructions for controlling the operations of the system 100. In another implementation, the processor 202 comprises a microcomputer or microcontroller configured to provide integrated processing and control functions within a compact form factor suitable for deployment within the crawler 102. In a further implementation, the processor 202 comprises a digital signal processor configured to perform signal processing operations on image data received from the imaging unit 104. In yet another implementation, the processor 202 comprises a central processing unit configured to execute general-purpose computational operations. In an additional implementation, the processor 202 comprises a state machine or logic circuitry configured to perform dedicated control functions for the cleaning mechanism 110 and the fluid dispensing unit 108.
[0057] The processor 202 includes a general-purpose processor such as a central processing unit or application processor configured to execute software instructions for general computational tasks. The processor 202 includes a graphics processing unit or visual processing unit configured to perform parallel processing operations on image data captured by the imaging unit 104. The graphics processing unit accelerates image processing operations by performing parallel computations on pixel data. The processor 202 includes an AI-dedicated processor such as a neural processing unit configured to execute artificial intelligence models for surface anomaly detection and classification. The neural processing unit provides hardware acceleration for neural network inference operations, enabling real-time processing of image data during traversal of the crawler 102 within the conduit.
[0058] The memory 204 stores the data 206 and the modules 208 that are accessed and executed by the processor 202 during operation of the processing unit 112. The memory 204 comprises volatile memory and non-volatile memory components.
[0059] The volatile memory comprises memory elements that retain stored data while power is supplied and lose stored data when power is removed. The volatile memory includes static random access memory configured to provide high-speed data access for frequently accessed data and instructions. The volatile memory includes dynamic random access memory configured to provide larger storage capacity for image data, position data, and intermediate processing results generated during inspection and cleaning operations.
[0060] The non-volatile memory comprises memory elements that retain stored data even when power is removed. The non-volatile memory includes read-only memory configured to store firmware and boot instructions for initialising the processing unit 112. The non-volatile memory includes erasable programmable read-only memory configured to store configuration parameters and calibration data that are retained across power cycles. The non-volatile memory includes flash memory configured to store inspection reports, defect records, and operational logs generated during inspection and cleaning operations. The non-volatile memory includes hard disks configured to provide large-capacity storage for historical inspection data and image archives. The non-volatile memory includes optical disks and magnetic tapes configured to provide archival storage for long-term retention of inspection records and maintenance history.
[0061] The data 206 stored in the memory 204 comprises image data received from the imaging unit 104, axial position data received from the position tracking mechanism 106, inspection outputs generated by the image processing module 210, structured inspection reports, and operational parameters for controlling the cleaning mechanism 110 and the fluid dispensing unit 108. The data 206 includes pre-stored threshold ranges associated with pre-defined anomaly classes used for surface anomaly classification. The data 206 includes trained model parameters for artificial intelligence models used in surface anomaly detection and classification.
[0062] The modules 208 stored in the memory 204 comprise software components that are executed by the processor 202 to perform the functions of the processing unit 112. The modules 208 include the image processing module 210, the cleaning operation controlling module 212, and the report generation module 214. The modules 208 are loaded into volatile memory and executed by the processor 202 during operation of the system 100.
[0063] The artificial intelligence models used by the image processing module 210 for surface anomaly detection and classification are trained using neural network architectures. The neural network architectures include convolutional neural networks configured to extract spatial features from image data for detecting and classifying surface anomalies. The neural network architectures include deep neural networks configured to learn hierarchical representations of surface conditions from training data. The neural network architectures include recurrent neural networks configured to process sequential image data captured during traversal of the crawler 102. The neural network architectures include restricted Boltzmann machines and deep belief networks configured to learn probabilistic representations of surface anomaly patterns. The neural network architectures include bidirectional recurrent deep neural networks configured to process image sequences in both forward and backward directions for improved anomaly detection. The neural network architectures include generative adversarial networks configured to generate synthetic training data for augmenting training datasets. The neural network architectures include deep Q-networks configured to learn optimal inspection and cleaning strategies through interaction with the conduit environment.
[0064] The artificial intelligence models are trained using various training techniques. Supervised learning is used to train models using labelled training data comprising image samples annotated with corresponding surface anomaly classifications. Unsupervised learning is used to train models to identify patterns and clusters in image data without labelled annotations. Semi-supervised learning is used to train models using a combination of labelled and unlabelled training data, enabling effective model training when labelled data is limited. Reinforcement learning is used to train models to learn optimal cleaning strategies by receiving feedback based on cleaning effectiveness and operational efficiency.
[0065] The image processing module 210 is configured to receive the image data captured by the imaging unit 104. The image processing module 210 is communicatively coupled to the imaging unit 104 and receives image data as the crawler 102 traverses the conduit interior. The image processing module 210 processes the received image data to detect and classify surface anomalies on the inner surface of the conduit.
[0066] Figure 3 illustrates a block diagram of the image processing module 210 of the processing unit of Figure 2, according to aspects of the present disclosure. The image processing module 210 is operatively coupled to multiple specialized detection modules configured to analyze image data captured during conduit inspection operations. The specialized detection modules perform targeted analysis functions for detecting and characterizing specific types of surface anomalies present on the inner surface of the conduit.
[0067] The image processing module 210 receives image data from the imaging unit 104 during traversal of the crawler 102 within the conduit. The received image data comprises high-resolution images of the inner surface captured by the at least one camera of the imaging unit 104 under controlled illumination conditions. The image processing module 210 processes the received image data using artificial intelligence-based computer vision models to identify and classify surface anomalies present in the captured images.
[0068] The image processing module 210 distributes the received image data to the specialized detection modules for parallel or sequential analysis based on inspection objectives. Each specialized detection module is configured to detect and characterize a specific category of surface anomaly, enabling comprehensive analysis of the inner surface condition through coordinated operation of the multiple detection modules. The image processing module 210 aggregates outputs from the specialized detection modules to generate consolidated inspection results that characterize the overall condition of the inner surface of the conduit.
[0069] The image processing module 210 is configured to detect surface anomalies on the inner surface of the conduit based on analysis of the image data received from the imaging unit 104. The image processing module 210 applies artificial intelligence-based computer vision techniques to the captured image data to identify regions of the inner surface that exhibit characteristics indicative of surface anomalies. The detection of surface anomalies is performed automatically during traversal of the crawler 102 within the conduit, enabling real-time identification of surface conditions as the imaging unit 104 captures image data of the inner surface.
[0070] The surface anomalies detected by the image processing module 210 comprise at least one of rust, carbon deposits, copper fouling, foreign deposits, or cracks. Rust comprises oxidation formations on the inner surface that result from corrosion processes affecting the conduit material. Carbon deposits comprise accumulations of carbonaceous material on the inner surface that result from combustion residues or operational byproducts. Copper fouling comprises copper-based deposits that accumulate on the inner surface, particularly in projectile launching device applications where copper jacket material transfers to the barrel interior during projectile passage. Foreign deposits comprise non-native substances present on the inner surface that are not attributable to the conduit material or expected operational residues. Cracks comprise discontinuities in the surface material that extend into the conduit wall, including thermally induced cracking, stress-related fractures, and surface fissures.
[0071] The image processing module 210 analyses the captured image data to identify visual signatures associated with each category of surface anomaly. The visual signatures comprise colour characteristics, texture patterns, geometric features, and intensity variations that distinguish anomalous regions from unaffected portions of the inner surface. The image processing module 210 compares extracted image features against learned representations of surface anomaly characteristics to determine whether detected regions correspond to rust, carbon deposits, copper fouling, foreign deposits, or cracks.
[0072] The image processing module 210 is configured to perform micro level segmentation of detected surface anomalies to quantify defect parameters comprising defect area. The micro level segmentation comprises pixel-level analysis of the captured image data to isolate and delineate the boundaries of detected surface anomalies with high spatial precision. The pixel-level segmentation identifies individual pixels within the captured image data that correspond to anomalous surface regions, enabling precise determination of the spatial extent of each detected surface anomaly.
[0073] The micro level segmentation performed by the image processing module 210 generates segmentation masks that define the boundaries of detected surface anomalies at pixel resolution. Each segmentation mask comprises a binary or multi-class representation indicating which pixels within the captured image correspond to anomalous surface regions and which pixels correspond to unaffected surface regions. The segmentation masks enable accurate measurement of the spatial extent of detected surface anomalies by counting the number of pixels classified as anomalous and converting the pixel count to physical area measurements based on known image scale factors.
[0074] The defect parameters quantified through micro level segmentation comprise defect area measurements that indicate the physical extent of each detected surface anomaly on the inner surface of the conduit. The defect area is computed by determining the number of pixels within the segmentation mask that correspond to the detected surface anomaly and multiplying by a scale factor that relates pixel dimensions to physical surface dimensions. The scale factor is determined based on the optical characteristics of the imaging unit 104 and the distance between the imaging unit 104 and the inner surface of the conduit during image capture.
[0075] The micro level segmentation enables the image processing module 210 to distinguish between surface anomalies of different sizes and to quantify the severity of detected conditions based on the measured defect area. Surface anomalies with larger defect areas indicate more extensive surface degradation or contamination, while surface anomalies with smaller defect areas indicate localised conditions that affect limited portions of the inner surface. The quantified defect area measurements are included in the spatially localised inspection outputs generated by the processing unit 112 and are used to inform cleaning operation decisions in the inspection-guided selective cleaning mode.
[0076] The image processing module 210 is operatively coupled to a cleaning time estimation module 309. The cleaning time estimation module 309 is configured to analyse deposit on the inner surface of the conduit. The cleaning time estimation module 309 receives image data from the image processing module 210 and processes the received image data to determine extent of deposits present on the inner surface at specific locations within the conduit.
[0077] The cleaning time estimation module 309 employs an artificial intelligence-based model, such as Machine Learning models, to analyse colour of deposits on the surface. In an embodiment, the cleaning time estimation module 309 receives image data and determine the colour of depositions present on the inner surface. The cleaning time estimation module 309 analyses the colour characteristics of detected depositions to estimate the required cleaning time, wherein darker coloured depositions indicate more severe accumulations requiring extended cleaning duration, and lighter coloured depositions indicate less severe accumulations requiring reduced cleaning duration. The cleaning operation controlling module 212 receives cleaning time estimates from the cleaning time estimation module 309 and adjusts the cleaning duration at each axial position where surface anomalies have been detected. Surface anomalies with darker coloured depositions receive extended cleaning duration, while surface anomalies with lighter coloured depositions receive reduced cleaning duration. The adjustment of cleaning parameters based on deposition colour enables targeted cleaning execution that minimizes unnecessary cleaning actions while ensuring adequate treatment of identified defect or deposit regions.
[0078] The artificial intelligence-based model, such as ML model, deployed in the cleaning time estimation module 309 is trained to recognise extent of deposit accumulation patterns based on visual characteristics extracted from the captured image data. The artificial intelligence-based model processes pixel-level information from segmented deposit regions to compute severity of deposition that characterise the concentration and distribution of deposits on the inner surface. The cleaning time estimation module 309 automatically determines a corresponding cleaning duration and intensity required for effective removal. The processing unit 112 is configured to adjust at least one of cleaning duration or cleaning intensity based on the severity of detected surface anomalies at corresponding axial positions using the cleaning time estimation module 309.
[0079] The cleaning operation is dynamically adjusted based on surface condition severity. The processing unit 112 references the severity information associated with each detected surface anomaly and adjusts the cleaning parameters in real-time as the crawler 102 traverses the conduit. The dynamic adjustment enables targeted cleaning execution that minimizes unnecessary cleaning actions while ensuring adequate treatment of identified defect or deposit regions. Regions with greater severity receive extended cleaning duration and increased cleaning intensity, while regions with lower lesser severity receive reduced cleaning duration and decreased cleaning intensity. The dynamic adjustment of cleaning parameters based on surface condition severity optimises cleaning effectiveness and reduces consumption of cleaning media in regions where intensive cleaning is not required.
[0080] The image processing module 210 comprises a rust detection module 302 configured to perform image-level rust detection, rust localisation, and pixel-level segmentation of rusted regions. The rust detection module 302 is operatively coupled to the image processing module 210 and receives image data captured by the imaging unit 104 during traversal of the crawler 102 within the conduit.
[0081] The rust detection module 302 performs AI-based rust detection using advanced computer vision and deep learning techniques. The rust detection module 302 analyses the received image data to identify regions of the inner surface that exhibit visual characteristics indicative of rust formation. Rust formations comprise oxidation products that develop on metallic surfaces of the conduit due to corrosion processes, and the rust detection module 302 is configured to detect such formations based on colour signatures, texture patterns, and spatial distributions characteristic of oxidised surface regions.
[0082] The rust detection module 302 supports image-level rust classification. Image-level rust classification comprises analysis of the captured image data to determine whether rust is present within the image and to classify the overall rust condition depicted in the image. The image-level rust classification provides a categorical assessment of rust presence and severity for each captured image, enabling rapid screening of the inner surface condition during traversal of the crawler 102.
[0083] The rust detection module 302 supports rust localisation using bounding boxes. Rust localisation comprises identification of the spatial location of rusted regions within the captured image data. The rust detection module 302 generates bounding boxes that define rectangular regions within the captured image containing detected rust formations. Each bounding box specifies the coordinates of the detected rust region within the image frame, enabling identification of where rust formations are located on the inner surface relative to the field of view of the imaging unit 104.
[0084] The rust detection module 302 supports pixel-level segmentation to precisely identify the shape and extent of rusted regions. Pixel-level segmentation comprises classification of individual pixels within the captured image data as either corresponding to rusted surface regions or corresponding to unaffected surface regions. The pixel-level segmentation generates segmentation masks that delineate the precise boundaries of detected rust formations, enabling accurate measurement of rust coverage area and characterisation of rust formation geometry.
[0085] The rust detection module 302 supports encoder-linked positional mapping. Encoder-linked positional mapping comprises correlation of detected rust formations with axial position data generated by the position tracking mechanism 106. The rust detection module 302 receives encoder readings from the position tracking mechanism 106 and associates each detected rust formation with the corresponding axial position of the crawler 102 at the time of image capture. The encoder-linked positional mapping enables generation of spatially localised inspection outputs that specify the precise location of detected rust formations along the length of the conduit, supporting position-specific maintenance actions and targeted cleaning operations at identified rust locations.
[0086] The image processing module 210 comprises a surface condition detection module 304 configured to detect and quantify carbon deposits, copper fouling, dirt, and contaminants. The surface condition detection module 304 is operatively coupled to the image processing module 210 and receives image data captured by the imaging unit 104 during traversal of the crawler 102 within the conduit.
[0087] The surface condition detection module 304 is configured for advanced image processing-based detection and quantification of surface conditions associated with cleaning requirements. The surface condition detection module 304 analyses the received image data to identify regions of the inner surface that exhibit visual characteristics indicative of carbon deposits, copper fouling, dirt, and contaminants. The surface condition detection module 304 employs image processing techniques to extract features from the captured image data that distinguish contaminated surface regions from unaffected surface regions.
[0088] Carbon deposits comprise accumulations of carbonaceous material on the inner surface that result from combustion residues or operational byproducts. The surface condition detection module 304 detects carbon deposits based on colour signatures and texture patterns characteristic of carbonaceous accumulations. The surface condition detection module 304 quantifies detected carbon deposits by computing coverage area and distribution patterns within the captured image data.
[0089] Copper fouling comprises copper-based deposits that accumulate on the inner surface. The surface condition detection module 304 detects copper fouling based on colour characteristics and reflectivity patterns associated with copper material transferred to the inner surface. The surface condition detection module 304 quantifies detected copper fouling by measuring the spatial extent and intensity of copper-related visual signatures within the captured image data.
[0090] Dirt comprises particulate matter and debris present on the inner surface. The surface condition detection module 304 detects dirt based on visual characteristics that distinguish particulate accumulations from the underlying surface material. Contaminants comprise substances present on the inner surface that are not attributable to expected operational residues. The surface condition detection module 304 detects contaminants based on visual anomalies that indicate the presence of foreign substances on the inner surface.
[0091] The surface condition detection module 304 performs segmentation of detected surface conditions to isolate affected regions and compute quantitative parameters. The surface condition detection module 304 generates segmentation outputs that delineate the boundaries of detected carbon deposits, copper fouling, dirt, and contaminants at pixel resolution. The quantitative parameters computed by the surface condition detection module 304 comprise coverage percentage, spatial distribution, and severity metrics for each detected surface condition.
[0092] The surface condition detection module 304 supports customization based on inspection requirements. The customization enables the surface condition detection module 304 to be configured for different inspection objectives, detection sensitivities, and surface condition categories based on the specific requirements of a given inspection operation. The customization comprises adjustment of detection thresholds, selection of target surface condition categories, and configuration of quantification parameters to match the inspection requirements for the conduit being inspected. The customization capability enables the surface condition detection module 304 to be adapted for different conduit types, operational environments, and maintenance objectives without requiring modification of the underlying detection algorithms.
[0093] The image processing module 210 comprises a crack detection module 306 configured to perform crack detection with pixel-level segmentation of crack regions, and to estimate crack width and crack length. The crack detection module 306 is operatively coupled to the image processing module 210 and receives image data captured by the imaging unit 104 during traversal of the crawler 102 within the conduit.
[0094] The crack detection module 306 is configured for precise identification and characterisation of cracks on the inner surface of the conduit using deep learning-based image analysis. The crack detection module 306 analyses the received image data to identify regions of the inner surface that exhibit visual characteristics indicative of crack formations. Cracks comprise discontinuities in the surface material that extend into the conduit wall, including thermally induced cracking, stress-related fractures, and surface fissures that develop due to operational stresses, thermal cycling, or material degradation.
[0095] The crack detection module 306 performs crack classification to categorise detected cracks based on visual characteristics and morphological features. The crack classification comprises analysis of crack appearance, orientation, branching patterns, and surface characteristics to determine the type and nature of detected crack formations. The crack classification enables differentiation between different crack categories, including thermally induced cracks, fatigue cracks, and stress corrosion cracks, based on their distinctive visual signatures.
[0096] The crack detection module 306 performs object detection to identify and localise crack formations within the captured image data. The object detection comprises identification of regions within the captured image that contain crack formations and generation of bounding regions that define the spatial extent of detected cracks within the image frame. The object detection enables rapid identification of crack locations within the field of view of the imaging unit 104.
[0097] The crack detection module 306 performs pixel-level segmentation of crack regions to accurately delineate crack boundaries. The pixel-level segmentation comprises classification of individual pixels within the captured image data as either corresponding to crack regions or corresponding to unaffected surface regions. The pixel-level segmentation generates segmentation masks that define the precise boundaries of detected crack formations at pixel resolution, enabling accurate characterisation of crack geometry and spatial extent.
[0098] The crack detection module 306 estimates geometric attributes including crack length and width to quantify severity of detected cracks. The crack length estimation comprises measurement of the linear extent of detected crack formations along their principal axis. The crack detection module 306 analyses the segmentation mask of each detected crack to determine the maximum linear dimension of the crack region, providing a crack length measurement that characterises the extent of the crack formation along the inner surface.
[0099] The crack width estimation comprises measurement of the transverse dimension of detected crack formations perpendicular to their principal axis. The crack detection module 306 analyses the segmentation mask of each detected crack to determine the width of the crack opening at various points along the crack length. The crack width measurement characterises the severity of the crack formation by indicating the degree of surface separation at the crack location.
[0100] The geometric attribute estimation performed by the crack detection module 306 enables quantification of crack severity based on measured crack dimensions. Cracks with greater length and width measurements indicate more severe surface degradation that affects larger portions of the inner surface and represents deeper penetration into the conduit material. The quantified crack dimensions are included in the spatially localised inspection outputs generated by the processing unit 112 and are used to assess the structural condition of the conduit and to inform maintenance decisions.
[0101] The crack detection module 306 associates detected cracks with exact conduit coordinates using encoder-based position tracking. The crack detection module 306 receives axial position data from the position tracking mechanism 106 and correlates each detected crack with the corresponding axial position of the crawler 102 at the time of image capture. The encoder-linked positional mapping enables generation of spatially localised defect reporting and analysis that specifies the precise location of detected cracks along the length of the conduit.
[0102] The image processing module 210 comprises a foreign deposition detection module 308 configured to identify non-native substances, isolate deposit regions, and compute coverage percentage. The foreign deposition detection module 308 is operatively coupled to the image processing module 210 and receives image data captured by the imaging unit 104 during traversal of the crawler 102 within the conduit.
[0103] The foreign deposition detection module 308 is configured for detection and quantification of foreign deposits on the internal surface of the conduit. The foreign deposition detection module 308 analyses the received image data to identify regions of the inner surface that exhibit visual characteristics indicative of non-native substances. Non-native substances comprise materials present on the inner surface that are not attributable to the conduit material itself or to expected operational residues such as carbon deposits or copper fouling. The foreign deposition detection module 308 distinguishes non-native substances from native surface materials and expected operational deposits based on colour signatures, texture patterns, reflectivity characteristics, and spatial distribution features that differ from those associated with the conduit material and anticipated residue types.
[0104] The foreign deposition detection module 308 isolates deposit regions within the captured image data. The isolation of deposit regions comprises segmentation of the captured image data to delineate the boundaries of detected foreign deposits at pixel resolution. The foreign deposition detection module 308 generates segmentation outputs that define the spatial extent of each detected foreign deposit region, enabling precise characterisation of deposit location and geometry on the inner surface. The isolation of deposit regions enables the foreign deposition detection module 308 to distinguish individual deposit formations from surrounding surface regions and to analyse each deposit region independently for quantification purposes.
[0105] The foreign deposition detection module 308 computes coverage percentage for detected foreign deposits. The coverage percentage comprises a quantitative metric indicating the proportion of the inspected surface area that is affected by foreign deposit accumulation. The foreign deposition detection module 308 calculates the coverage percentage by determining the total area of isolated deposit regions relative to the total inspected surface area within the captured image data. The coverage percentage provides a normalised measure of deposit extent that enables comparison of deposit severity across different inspection operations and different regions of the conduit.
[0106] The foreign deposition detection module 308 computes spread metrics for detected foreign deposits. The spread metrics characterise the spatial distribution of detected foreign deposits across the inspected surface area. The spread metrics indicate whether detected deposits are concentrated in localised regions or distributed across broader portions of the inner surface. The spread metrics enable assessment of deposit distribution patterns that inform maintenance decisions and cleaning operation planning.
[0107] The foreign deposition detection module 308 computes severity metrics for detected foreign deposits. The severity metrics characterise the intensity and degree of deposit accumulation at detected deposit locations. The severity metrics are computed based on visual characteristics of the detected deposits, including deposit thickness indicators, colour intensity, and texture features that correlate with deposit accumulation levels. The severity metrics enable differentiation between light deposit accumulations and heavy deposit accumulations, supporting prioritisation of cleaning operations based on deposit severity.
[0108] The coverage percentage, spread metrics, and severity metrics computed by the foreign deposition detection module 308 are included in inspection reporting outputs generated by the processing unit 112. The inspection reporting outputs comprise structured data records that document the detected foreign deposits, their spatial locations, and their quantified characteristics. The inspection reporting outputs enable informed maintenance decisions by providing objective quantification of foreign deposit conditions on the inner surface of the conduit.
[0109] The image processing module 210 is configured to classify the surface anomalies on the inner surface of the conduit. The classification of surface anomalies comprises determination of the specific anomaly type corresponding to each detected surface region based on analysis of visual characteristics extracted from the captured image data. The image processing module 210 assigns detected surface anomalies to pre-defined anomaly classes that correspond to the categories of surface conditions identified during inspection, including rust, carbon deposits, copper fouling, foreign deposits, and cracks.
[0110] The image processing module 210 is configured to apply a predefined region of interest to a field of view of at least one camera of the imaging unit 104 to define a controlled inspection region. The predefined region of interest comprises a bounding circle that is applied to the camera's field of view during surface anomaly detection operations. The bounding circle defines a controlled inspection region corresponding to a target surface area on the inner surface of the conduit. The bounding circle delineates a specific portion of the captured image data that is analysed for surface anomaly detection, enabling focused inspection of a defined surface region as the crawler 102 traverses the conduit interior.
[0111] The controlled inspection region defined by the bounding circle corresponds to a target surface area that is positioned within the field of view of the at least one camera of the imaging unit 104. The target surface area comprises the portion of the inner surface that is captured within the bounding circle region during image acquisition. As the crawler 102 traverses the conduit, the bounding circle region moves along the inner surface, enabling sequential inspection of target surface areas along the full length of the conduit. The application of the bounding circle provides a consistent and repeatable inspection region that enables standardised analysis of surface conditions across different portions of the inner surface.
[0112] The image processing module 210 is configured to extract pixel colour values within the region of interest. The extraction of pixel colour values comprises real-time acquisition of colour and intensity information from pixels located within the white colour bounding circle region of the captured image data. The pixel colour values represent the colour characteristics and intensity levels of the inner surface at each pixel location within the controlled inspection region. The real-time extraction of pixel colour values enables continuous monitoring of surface conditions as the crawler 102 traverses the conduit, with colour and intensity information being acquired and processed during traversal.
[0113] When a copper, carbon, or rust deposit enters the bounding circle region, the deposit causes a localised variation in pixel intensity and colour values within the captured image data. The localised variation comprises changes in the colour characteristics and intensity levels of pixels corresponding to the deposit region compared to pixels corresponding to unaffected surface regions. The image processing module 210 detects these localised variations by analysing the extracted pixel colour values within the bounding circle region and identifying regions where the colour and intensity characteristics differ from expected values for unaffected surface material.
[0114] The image processing module 210 is configured to compare the extracted pixel colour values with pre-stored threshold ranges associated with pre-defined anomaly classes to classify detected regions. The pre-stored threshold ranges comprise model-defined threshold ranges representing characteristic colour and intensity signatures of specific surface anomalies. Each pre-defined anomaly class is associated with a corresponding set of threshold ranges that define the expected colour and intensity characteristics for that category of surface anomaly. The threshold ranges represent the colour signatures and intensity patterns that are characteristic of rust formations, carbon deposits, copper fouling, foreign deposits, and other surface anomaly types.
[0115] The comparison of extracted pixel colour values against pre-stored threshold ranges enables classification of detected surface regions based on their colour and intensity characteristics. The image processing module 210 determines whether the captured pixel values fall within one or more predefined ranges associated with specific anomaly classes. Upon determining that the captured pixel values fall within a predefined range corresponding to a particular anomaly class, the image processing module 210 classifies the detected region as the corresponding anomaly type. The classification generates an anomaly identification output that specifies the type of surface anomaly detected at the corresponding location within the bounding circle region.
[0116] The image processing module 210 correlates detected surface anomalies with corresponding axial position data from the position tracking mechanism 106 to generate spatially localised inspection outputs. Concurrently with the anomaly detection process, encoder readings associated with the movement of the crawler 102 are continuously captured when an anomaly is present within the white bounding circle region. The continuous capture of encoder readings during anomaly detection enables simultaneous detection and positional measurement of surface anomalies. The encoder data is used to calculate the precise distance or positional location of the detected anomaly along the inspected surface, such that the detection and positional measurement occur simultaneously.
[0117] The simultaneous detection and positional measurement enabled by the continuous encoder capture provides accurate spatial localisation of each detected surface anomaly. The image processing module 210 receives the encoder readings from the position tracking mechanism 106 and associates each detected anomaly with the corresponding axial position of the crawler 102 at the time of detection. The correlation of detected anomalies with encoder-derived distance values enables generation of spatially localised inspection outputs that specify the precise location of each detected surface anomaly along the length of the conduit.
[0118] Based on the classified anomaly, the extracted pixel data, and the corresponding encoder-derived distance values, the image processing module 210 generates inspection outputs comprising at least an anomaly type, an anomaly coverage or severity percentage, and a distance or location at which the anomaly occurred. The inspection outputs provide synchronized surface anomaly detection and spatial localisation using combined vision-based analysis and encoder-based positioning. The spatially localised inspection outputs enable position-specific maintenance actions and targeted cleaning operations at identified anomaly locations within the conduit.
[0119] The modules 208 include a cleaning operation controlling module 212 configured to control the cleaning mechanism 110 and the fluid dispensing unit 108 to execute cleaning operations on the inner surface of the conduit. The cleaning operation controlling module 212 is stored in the memory 204 of the processing unit 112 and is executed by the processor 202 to manage cleaning operations performed by the system 100.
[0120] The cleaning operation controlling module 212 is communicatively coupled to the cleaning mechanism 110. The communicative coupling between the cleaning operation controlling module 212 and the cleaning mechanism 110 enables the cleaning operation controlling module 212 to transmit control signals that activate, deactivate, and modulate the operation of the cleaning mechanism 110 during cleaning operations. The cleaning operation controlling module 212 controls the timing, duration, and intensity of cleaning actions performed by the cleaning mechanism 110 based on operational parameters and, where applicable, inspection outputs generated by the image processing module 210.
[0121] The cleaning operation controlling module 212 is communicatively coupled to the fluid dispensing unit 108. The communicative coupling between the cleaning operation controlling module 212 and the fluid dispensing unit 108 enables the cleaning operation controlling module 212 to transmit control signals that govern the dispensing of cleaning fluid or oil during cleaning operations. The cleaning operation controlling module 212 controls the pump mechanism of the fluid dispensing unit 108 to regulate pump activation, flow rate, and dispensing duration based on operational parameters and positional data received from the position tracking mechanism 106.
[0122] The cleaning operation controlling module 212 coordinates the operation of the cleaning mechanism 110 and the fluid dispensing unit 108 to execute cleaning operations on the inner surface of the conduit. The coordination comprises synchronised activation of the cleaning mechanism 110 and the fluid dispensing unit 108 such that cleaning media is dispensed at appropriate times and locations relative to the cleaning actions performed by the cleaning mechanism 110. The cleaning operation controlling module 212 receives axial position data from the position tracking mechanism 106 and uses the received position data to synchronise the operation of the cleaning mechanism 110 and the fluid dispensing unit 108 with the current position of the crawler 102 within the conduit.
[0123] The cleaning operation controlling module 212 is configured to operate in a full-coverage cleaning mode. In the full-coverage cleaning mode, the cleaning mechanism 110 and the fluid dispensing unit 108 are activated to perform cleaning operations along an entire length of the conduit independent of the spatially localised inspection outputs generated by the image processing module 210. The full-coverage cleaning mode provides comprehensive cleaning coverage of the inner surface without reference to inspection data or defect localisation information.
[0124] In the full-coverage cleaning mode, the cleaning operation controlling module 212 activates the cleaning mechanism 110 to perform cleaning actions continuously as the crawler 102 traverses the full length of the conduit. The cleaning mechanism 110 operates throughout the traversal to engage with the inner surface and remove deposits, fouling, and contaminants present along the entire conduit length. The continuous activation of the cleaning mechanism 110 ensures uniform cleaning treatment of the inner surface regardless of the presence or absence of detected surface anomalies at specific locations.
[0125] In the full-coverage cleaning mode, the cleaning operation controlling module 212 activates the fluid dispensing unit 108 to dispense cleaning fluid or oil along the entire length of the conduit. The pump mechanism of the fluid dispensing unit 108 operates to deliver cleaning media uniformly as the crawler 102 traverses the conduit interior. The uniform dispensing of cleaning media ensures consistent application of cleaning fluid or oil across the full extent of the inner surface.
[0126] The full-coverage cleaning mode operates independent of the spatially localised inspection outputs. The cleaning operation controlling module 212 does not reference inspection data, defect locations, or severity metrics when controlling the cleaning mechanism 110 and the fluid dispensing unit 108 in the full-coverage cleaning mode. The cleaning operations are executed based on traversal position and operational parameters rather than inspection-derived information about surface conditions at specific locations.
[0127] During the cleaning phase in the full-coverage cleaning mode, no inspection or AI analysis is performed while the crawler 102 traverses the conduit and executes cleaning actions. The image processing module 210 and the associated detection modules are disabled or inactive during the full-coverage cleaning operation. The processing unit 112 focuses exclusively on controlling cleaning execution without performing concurrent image capture, surface anomaly detection, or defect classification operations. The deactivation of inspection and AI analysis functions during the cleaning phase enables the processing unit 112 to dedicate computational resources to cleaning control operations.
[0128] The processing unit 112 is configured to receive an instruction indicative of controlling the cleaning mechanism 110 and the fluid dispensing unit 108. The instruction comprises a control command that specifies operational parameters for the cleaning mechanism 110 and the fluid dispensing unit 108, including activation of the full-coverage cleaning mode. The processing unit 112 receives the instruction from the remote electronic device 124 via a user interface. The user interface comprises an interface presented on the remote electronic device 124 that enables an operator to transmit control commands to the processing unit 112 via the network 126. The user interface enables selection of operational modes, configuration of cleaning parameters, and initiation of cleaning operations from a remote location.
[0129] The processing unit 112 is configured to operate at least one of the cleaning mechanism 110 and the fluid dispensing unit 108 based on the received instruction. Upon receiving the instruction from the remote electronic device 124, the processing unit 112 interprets the instruction and transmits corresponding control signals to the cleaning mechanism 110 and the fluid dispensing unit 108. The control signals activate, deactivate, or modulate the operation of the cleaning mechanism 110 and the fluid dispensing unit 108 in accordance with the parameters specified in the received instruction. The remote control capability enables an operator to initiate and manage full-coverage cleaning operations from the remote electronic device 124 without direct physical interaction with the system 100.
[0130] The cleaning operation controlling module 212 is configured to operate in an inspection-guided selective cleaning mode. In the inspection-guided selective cleaning mode, the cleaning mechanism 110 and the fluid dispensing unit 108 are selectively activated only at axial positions corresponding to detected surface anomalies based on the spatially localised inspection outputs generated by the image processing module 210. The inspection-guided selective cleaning mode enables targeted cleaning operations that focus cleaning actions on regions of the inner surface where surface anomalies have been detected, while avoiding unnecessary cleaning actions in unaffected regions of the conduit.
[0131] The inspection-guided selective cleaning mode operates in two coordinated software-controlled phases comprising an inspection phase followed by a cleaning phase. The two-phase workflow separates the inspection and cleaning operations into distinct sequential phases, with inspection data generated during the inspection phase being used to guide cleaning actions during the subsequent cleaning phase. The coordination between the inspection phase and the cleaning phase is managed by the processing unit 112, which stores inspection outputs from the inspection phase and references the stored outputs during the cleaning phase to control selective activation of the cleaning mechanism 110 and the fluid dispensing unit 108.
[0132] During the inspection phase, the crawler 102 traverses the conduit and the imaging unit 104 captures image data of the inner surface. The image processing module 210 processes the captured image data to detect and classify surface anomalies on the inner surface. The rust detection module 302, the surface condition detection module 304, the crack detection module 306, and the foreign deposition detection module 308 analyse the captured image data to identify rust formations, carbon deposits, copper fouling, cracks, and foreign deposits present on the inner surface. The position tracking mechanism 106 generates axial position data during traversal, and the image processing module 210 correlates detected surface anomalies with corresponding axial position data to generate spatially localised inspection outputs.
[0133] The spatially localised inspection outputs generated during the inspection phase comprise positional coordinates along the conduit, type of detected surface condition or defect, and quantified severity metrics such as coverage percentage or extent for each detected surface anomaly. The report generation module 214 compiles the spatially localised inspection outputs into a structured inspection report that is stored in the memory 204 of the processing unit 112. The structured inspection report provides a record of detected surface anomalies and their corresponding axial positions that is referenced during the subsequent cleaning phase.
[0134] During the cleaning phase, the crawler 102 traverses the conduit and the cleaning operation controlling module 212 executes cleaning actions based on the stored inspection report. The cleaning operation controlling module 212 continuously compares real-time axial position data from the position tracking mechanism 106 with recorded anomaly locations from the structured inspection report. As the crawler 102 traverses the conduit, the cleaning operation controlling module 212 monitors the current axial position of the crawler 102 and determines whether the current position corresponds to a location where a surface anomaly was detected during the inspection phase.
[0135] When the axial position of the crawler 102 corresponds to a recorded anomaly location from the structured inspection report, the cleaning operation controlling module 212 selectively activates the cleaning mechanism 110 to perform cleaning operations at that axial position. The cleaning mechanism 110 engages with the inner surface at the identified anomaly location to remove deposits, fouling, or contaminants detected during the inspection phase. The cleaning operation controlling module 212 also selectively activates the fluid dispensing unit 108 to dispense cleaning fluid or oil at the axial position corresponding to the detected surface anomaly. The pump mechanism of the fluid dispensing unit 108 is actuated in synchronisation with the position of the crawler 102 to ensure targeted treatment of the identified region.
[0136] When the axial position of the crawler 102 does not correspond to a recorded anomaly location, the cleaning operation controlling module 212 maintains the cleaning mechanism 110 and the fluid dispensing unit 108 in an inactive state. The cleaning mechanism 110 does not perform cleaning actions and the fluid dispensing unit 108 does not dispense cleaning media when the crawler 102 traverses regions of the conduit where no surface anomalies were detected during the inspection phase. The selective activation based on recorded anomaly locations avoids unnecessary application of cleaning actions and cleaning media in unaffected regions of the conduit.
[0137] The cleaning intensity and fluid dispensing parameters during the cleaning phase are guided by the severity metrics recorded in the structured inspection report. The cleaning operation controlling module 212 references the quantified severity metrics associated with each detected surface anomaly and adjusts the cleaning duration and intensity at corresponding axial positions based on the recorded severity. Surface anomalies with higher severity metrics receive extended cleaning duration and increased cleaning intensity, while surface anomalies with lower severity metrics receive reduced cleaning duration and decreased cleaning intensity. The cleaning time estimation module 309 provides deposit severity analysis that informs the determination of cleaning parameters for effective removal of detected deposits.
[0138] During the cleaning phase, no inspection or AI analysis is performed. The image processing module 210 and the associated detection modules are inactive during the cleaning phase, and the processing unit 112 focuses on controlling cleaning execution based on the previously generated inspection outputs. The separation of inspection and cleaning into distinct phases enables the processing unit 112 to dedicate computational resources to cleaning control operations during the cleaning phase without concurrent image processing overhead.
[0139] Upon completion of the selective cleaning cycle, the cleaning operation controlling module 212 exits the inspection-guided selective cleaning mode. The completion of the selective cleaning cycle occurs when the crawler 102 has traversed the full length of the conduit and cleaning actions have been executed at all axial positions corresponding to detected surface anomalies recorded in the structured inspection report. The integrated inspection-and-cleaning approach provided by the inspection-guided selective cleaning mode enables precise localisation of surface anomalies, targeted remediation at identified locations, and efficient conduit maintenance through coordinated software control.
[0140] The modules 208 include a report generation module 214 configured to generate a structured inspection report comprising at least anomaly type, quantified severity metrics, coverage percentage, and positional coordinates for each detected surface anomaly. The report generation module 214 is stored in the memory 204 of the processing unit 112 and is executed by the processor 202 to compile inspection outputs from the image processing module 210 into structured digital inspection records.
[0141] The report generation module 214 receives inspection outputs from the image processing module 210 during and following inspection operations performed by the system 100. The inspection outputs comprise detection results from the rust detection module 302, the surface condition detection module 304, the crack detection module 306, the foreign deposition detection module 308, and the cleaning time estimation module 309. The report generation module 214 aggregates the received inspection outputs and organises the aggregated data into a structured format suitable for storage, retrieval, and analysis.
[0142] The structured inspection report generated by the report generation module 214 comprises anomaly type information for each detected surface anomaly. The anomaly type information specifies the classification of each detected surface anomaly as determined by the image processing module 210, including rust, carbon deposits, copper fouling, foreign deposits, cracks, corrosion-induced pitting, material chipping, thermally induced cracking, and surface contamination. The anomaly type information enables identification of the specific category of surface condition present at each detected location within the conduit.
[0143] The structured inspection report comprises quantified severity metrics for each detected surface anomaly. The quantified severity metrics comprise numerical measurements that characterise the degree and intensity of each detected surface anomaly. The severity metrics include defect area measurements computed through micro level segmentation, colour characteristics determined by the cleaning time estimation module 309, crack width and crack length measurements computed by the crack detection module 306, and severity classifications derived from visual characteristic analysis. The quantified severity metrics enable objective assessment of the condition of the inner surface at each detected anomaly location.
[0144] The structured inspection report comprises coverage percentage for each detected surface anomaly. The coverage percentage indicates the proportion of the inspected surface area affected by each detected surface anomaly relative to the total inspected area. The coverage percentage is computed based on pixel-level segmentation outputs that delineate the spatial extent of detected anomalies within the captured image data. The coverage percentage provides a normalised measure of anomaly extent that enables comparison of surface conditions across different inspection operations and different regions of the conduit.
[0145] The structured inspection report comprises positional coordinates for each detected surface anomaly. The positional coordinates specify the axial location of each detected surface anomaly along the length of the conduit as determined through correlation with encoder-derived position data from the position tracking mechanism 106. The positional coordinates enable precise spatial localisation of detected anomalies and support position-specific maintenance actions and targeted cleaning operations at identified anomaly locations.
[0146] The report generation module 214 compiles the inspection outputs into structured digital inspection records including type of surface condition detected, quantified severity, exact position within the conduit, and visual evidence. The type of surface condition detected comprises the anomaly classification assigned by the image processing module 210 based on analysis of visual characteristics extracted from the captured image data. The quantified severity comprises the severity metrics computed through pixel-level segmentation and geometric attribute estimation performed by the detection modules of the image processing module 210. The exact position within the conduit comprises the axial positional coordinates derived from encoder readings captured by the position tracking mechanism 106 during anomaly detection.
[0147] The structured digital inspection records include visual evidence for each detected surface anomaly. The visual evidence comprises image data captured by the imaging unit 104 that depicts the detected surface anomaly at the corresponding axial position within the conduit. The visual evidence includes annotated images showing bounding boxes, segmentation masks, or other visual indicators that highlight the detected anomaly regions within the captured image data. The inclusion of visual evidence in the structured digital inspection records enables verification of detected anomalies and provides documentary support for maintenance decisions based on inspection results.
[0148] The structured digital inspection records are suitable for maintenance planning, quality assessment, and historical comparison. The structured format of the inspection records enables systematic review of detected surface conditions and supports informed decision-making regarding maintenance actions. The inspection records provide data for quality assessment by documenting the condition of the inner surface at the time of inspection and enabling evaluation of surface quality against defined acceptance criteria. The inspection records enable historical comparison by providing a baseline record of surface conditions that is compared with subsequent inspection results to track changes in surface condition over time.
[0149] The processing unit 112 is configured to store the structured inspection report in a digital format for historical comparison and lifecycle traceability. The digital format comprises a data structure that organises the inspection report contents in a manner suitable for electronic storage, retrieval, and processing. The digital format enables the structured inspection report to be stored in the memory 204 of the processing unit 112 or transmitted to the remote electronic device 124 via the network 126 for storage in external data repositories.
[0150] The storage of the structured inspection report in digital format enables historical comparison of inspection results across multiple inspection operations. The digital format enables retrieval and comparison of inspection reports generated at different times for the same conduit, supporting identification of changes in surface condition over the operational lifetime of the conduit. The historical comparison capability enables tracking of surface degradation trends, assessment of maintenance effectiveness, and prediction of future maintenance requirements based on observed condition changes.
[0151] The storage of the structured inspection report in digital format enables lifecycle traceability of conduit inspection and maintenance activities. The digital inspection records provide a documented history of surface conditions detected during each inspection operation, maintenance actions performed in response to detected conditions, and outcomes of maintenance activities. The lifecycle traceability enables reconstruction of the inspection and maintenance history of each conduit, supporting accountability, compliance verification, and analysis of maintenance practices over the operational lifetime of the conduit.
[0152] The system 100 monitors execution parameters throughout inspection and cleaning operations to ensure stable and controlled operation. The processing unit 112 tracks operational parameters including crawler position, traversal speed, imaging unit status, cleaning mechanism activation state, and fluid dispensing status during inspection and cleaning operations. The monitoring of execution parameters enables detection of operational anomalies and supports controlled execution of inspection and cleaning cycles.
[0153] Upon completion of the defined inspection or cleaning cycle, the crawler 102 is withdrawn in a controlled manner. The processing unit 112 controls the withdrawal of the crawler 102 from the conduit interior following completion of the inspection or cleaning operation. The controlled withdrawal ensures that the crawler 102 exits the conduit without causing damage to the inner surface or to the components of the system 100.
[0154] Upon completion of the defined inspection or cleaning cycle, all operational and inspection data is securely stored for traceability and future reference. The processing unit 112 stores the structured inspection report, operational logs, and associated data in the memory 204 or transmits the data to the remote electronic device 124 for secure storage. The secure storage of operational and inspection data enables retrieval of inspection results and operational records for future reference, supporting maintenance planning, quality assessment, and lifecycle traceability of conduit inspection and maintenance activities.
[0155] Figure 4 illustrates a flowchart for a method 400 for automated inspection and cleaning of an inner surface of a conduit. The method 400 comprises a sequence of steps that are performed by the system 100 to inspect and clean the inner surface of the conduit. The method 400 integrates inspection and cleaning operations within a coordinated workflow, enabling targeted maintenance actions based on detected and localised surface conditions.
[0156] The method 400 begins with a step 402 where a crawler is traversed longitudinally within the conduit along an internal diameter of the conduit. The crawler 102 is introduced into the conduit and moves along the longitudinal axis of the internal bore. The traversal of the crawler 102 enables inspection and cleaning operations to be performed across the full length of the conduit interior. The crawler 102 comprises a linear axial carriage with a low-drag architecture that reduces mechanical resistance during movement within the conduit, as described previously. The longitudinal traversal of the crawler 102 along the internal diameter of the conduit positions the imaging unit 104, the cleaning mechanism 110, and the fluid dispensing unit 108 at successive axial locations within the conduit to enable sequential inspection and cleaning of the inner surface.
[0157] The method 400 proceeds to a step 404 where the imaging unit 104 mounted on the crawler 102 captures image data of the inner surface of the conduit during traversal. The imaging unit 104 comprises at least one camera and at least one illumination source, as described previously. During traversal of the crawler 102 within the conduit, the imaging unit 104 acquires high-resolution image data of the inner surface under controlled illumination conditions. The image data captured by the imaging unit 104 depicts the surface condition of the inner surface at each axial position along the conduit length. The capturing of image data during traversal enables continuous or interval-based acquisition of visual information representing the inner surface condition as the crawler 102 progresses through the conduit.
[0158] The method 400 continues to a step 406 where the position tracking mechanism 106 generates axial position data of the crawler 102 during traversal. The position tracking mechanism 106 is embodied as an encoder-based position tracking mechanism, as described previously. During traversal of the crawler 102 within the conduit, the position tracking mechanism 106 monitors the movement of the crawler 102 and records precise axial position data. The axial position data generated by the position tracking mechanism 106 indicates the location of the crawler 102 along the longitudinal axis of the conduit at each point during traversal. The generation of axial position data during traversal enables correlation of captured image data with corresponding positional coordinates, supporting spatial localisation of detected surface conditions along the conduit length.
[0159] The method 400 proceeds to a step 408 where the image processing module 210 processes the captured image data to detect and classify surface anomalies on the inner surface of the projectile launching device. The image processing module 210 receives the image data captured by the imaging unit 104 during traversal of the crawler 102 and applies artificial intelligence-based computer vision techniques to identify and characterise surface conditions present on the inner surface. The surface anomalies detected and classified by the image processing module 210 comprise at least one of rust, carbon deposits, copper fouling, foreign deposits, and cracks.
[0160] Detecting and classifying surface anomalies comprises performing rust detection using image-level rust detection, rust localisation, and pixel-level segmentation of rusted regions. The rust detection module 302 of the image processing module 210 performs image-level rust detection by analysing the captured image data to determine whether rust is present within each captured image and to classify the overall rust condition depicted in the image. The rust detection module 302 performs rust localisation by identifying the spatial location of rusted regions within the captured image data and generating bounding boxes that define rectangular regions containing detected rust formations. The rust detection module 302 performs pixel-level segmentation of rusted regions by classifying individual pixels within the captured image data as either corresponding to rusted surface regions or corresponding to unaffected surface regions, thereby generating segmentation masks that delineate the precise boundaries of detected rust formations and enable accurate measurement of rust coverage area.
[0161] Detecting and classifying surface anomalies comprises detecting and quantifying carbon deposits, copper fouling, dirt, and contaminants. The surface condition detection module 304 of the image processing module 210 analyses the captured image data to identify regions of the inner surface that exhibit visual characteristics indicative of carbon deposits, copper fouling, dirt, and contaminants. The surface condition detection module 304 detects carbon deposits based on colour signatures and texture patterns characteristic of carbonaceous accumulations resulting from combustion residues or operational byproducts. The surface condition detection module 304 detects copper fouling based on colour characteristics and reflectivity patterns associated with copper material transferred to the inner surface. The surface condition detection module 304 detects dirt comprising particulate matter and debris present on the inner surface and detects contaminants comprising substances present on the inner surface that are not attributable to expected operational residues. The surface condition detection module 304 quantifies detected carbon deposits, copper fouling, dirt, and contaminants by computing coverage area, distribution patterns, and severity metrics for each detected surface condition.
[0162] Detecting and classifying surface anomalies comprises performing crack detection comprising pixel-level segmentation of crack regions and estimation of crack width and crack length. The crack detection module 306 of the image processing module 210 analyses the captured image data to identify regions of the inner surface that exhibit visual characteristics indicative of crack formations, including thermally induced cracking, stress-related fractures, and surface fissures. The crack detection module 306 performs pixel-level segmentation of crack regions by classifying individual pixels within the captured image data as either corresponding to crack regions or corresponding to unaffected surface regions, thereby generating segmentation masks that define the precise boundaries of detected crack formations at pixel resolution. The crack detection module 306 performs estimation of crack width by measuring the transverse dimension of detected crack formations perpendicular to their principal axis at various points along the crack length. The crack detection module 306 performs estimation of crack length by measuring the linear extent of detected crack formations along their principal axis based on analysis of the segmentation mask of each detected crack. The estimated crack width and crack length measurements quantify the severity of detected cracks and characterise the degree of surface separation and extent of crack formations along the inner surface.
[0163] The method 400 proceeds to a step 410 where pixel-level segmentation of detected surface anomalies is performed to quantify defect parameters comprising at least one of defect area, coverage percentage, and severity. The pixel-level segmentation comprises classification of individual pixels within the captured image data as either corresponding to anomalous surface regions or corresponding to unaffected surface regions. The image processing module 210 generates segmentation masks that define the boundaries of detected surface anomalies at pixel resolution, enabling precise determination of the spatial extent of each detected surface anomaly on the inner surface of the conduit.
[0164] The defect area is quantified by determining the number of pixels within the segmentation mask that correspond to each detected surface anomaly and converting the pixel count to physical area measurements based on known image scale factors. The scale factor relates pixel dimensions to physical surface dimensions based on the optical characteristics of the imaging unit 104 and the distance between the imaging unit 104 and the inner surface of the conduit during image capture. The defect area measurement indicates the physical extent of each detected surface anomaly on the inner surface.
[0165] The coverage percentage is quantified by computing the proportion of the inspected surface area affected by each detected surface anomaly relative to the total inspected area within the captured image data. The coverage percentage is calculated based on the ratio of pixels classified as anomalous to the total number of pixels within the controlled inspection region defined by the predefined region of interest. The coverage percentage provides a normalised measure of anomaly extent that enables comparison of surface conditions across different inspection operations and different regions of the conduit.
[0166] The severity is quantified based on visual characteristics extracted from the captured image data and the computed defect parameters. The severity quantification incorporates defect area measurements, coverage percentage values, colour characteristics from the cleaning time estimation module 309, and geometric attribute measurements such as crack width and crack length from the crack detection module 306. The severity metrics characterise the degree and intensity of each detected surface anomaly and enable differentiation between surface anomalies of varying degrees of degradation or contamination. The quantified severity metrics inform cleaning operation decisions in the inspection-guided selective cleaning mode by indicating the cleaning duration and intensity required for effective treatment of each detected surface anomaly.
[0167] The method 400 continues to a step 412 where detected surface anomalies are correlated with corresponding axial position data to generate spatially localised inspection outputs. The image processing module 210 receives axial position data from the position tracking mechanism 106 and associates each detected surface anomaly with the corresponding axial position of the crawler 102 at the time of image capture. The correlation of detected surface anomalies with axial position data enables precise spatial localisation of each detected anomaly along the length of the conduit.
[0168] The spatially localised inspection outputs comprise positional coordinates along the conduit, type of detected surface condition or defect, and quantified defect parameters including defect area, coverage percentage, and severity for each detected surface anomaly. The positional coordinates specify the axial location of each detected surface anomaly along the longitudinal axis of the conduit as determined through correlation with encoder-derived position data from the position tracking mechanism 106. The spatially localised inspection outputs enable position-specific maintenance actions and targeted cleaning operations at identified anomaly locations within the conduit.
[0169] The correlation of detected surface anomalies with corresponding axial position data is performed concurrently with the anomaly detection process. Encoder readings associated with the movement of the crawler 102 are continuously captured during traversal, and the encoder data is used to calculate the precise distance or positional location of each detected anomaly along the inspected surface. The simultaneous detection and positional measurement enabled by the continuous encoder capture provides accurate spatial localisation of each detected surface anomaly without requiring separate positional determination operations following anomaly detection.
[0170] The spatially localised inspection outputs generated through correlation of detected surface anomalies with axial position data are compiled by the report generation module 214 into the structured inspection report. The structured inspection report stores the anomaly type, quantified defect parameters, and positional coordinates for each detected surface anomaly in a structured format within the memory 204 of the processing unit 112. The stored spatially localised inspection outputs are referenced during the cleaning phase in the inspection-guided selective cleaning mode to enable selective activation of the cleaning mechanism 110 and the fluid dispensing unit 108 at axial positions corresponding to detected surface anomalies.
[0171] The method 400 proceeds to a step 414 where cleaning operations are executed on the inner surface of the conduit. The cleaning operations comprise activation of the cleaning mechanism 110 and the fluid dispensing unit 108 to remove deposits, fouling, and contaminants from the inner surface of the conduit. The execution of cleaning operations follows the inspection and defect quantification steps described previously, enabling cleaning actions to be performed based on the inspection outputs generated during traversal of the crawler 102 within the conduit.
[0172] Cleaning operations are executed on the inner surface of the conduit based on one of a full-coverage cleaning mode or an inspection-guided selective cleaning mode. The selection between the full-coverage cleaning mode and the inspection-guided selective cleaning mode is determined based on operational requirements and maintenance objectives for a given cleaning operation. The full-coverage cleaning mode provides comprehensive cleaning coverage of the entire inner surface, while the inspection-guided selective cleaning mode provides targeted cleaning at locations where surface anomalies have been detected.
[0173] Executing cleaning operations in the full-coverage cleaning mode comprises activating a cleaning mechanism and a fluid dispensing unit to perform cleaning operations along an entire length of the conduit independent of the spatially localised inspection outputs. In the full-coverage cleaning mode, the cleaning mechanism 110 is activated to perform cleaning actions continuously as the crawler 102 traverses the full length of the conduit. The fluid dispensing unit 108 is activated to dispense cleaning fluid or oil uniformly along the entire length of the conduit during traversal. The cleaning mechanism 110 and the fluid dispensing unit 108 operate throughout the traversal to provide uniform cleaning treatment of the inner surface regardless of the presence or absence of detected surface anomalies at specific locations. The full-coverage cleaning mode operates independent of the spatially localised inspection outputs, and the cleaning operation controlling module 212 does not reference inspection data, defect locations, or severity metrics when controlling the cleaning mechanism 110 and the fluid dispensing unit 108.
[0174] Executing cleaning operations in the inspection-guided selective cleaning mode comprises comparing real-time axial position data with recorded anomaly locations from the structured inspection report. During the cleaning phase in the inspection-guided selective cleaning mode, the cleaning operation controlling module 212 receives real-time axial position data from the position tracking mechanism 106 as the crawler 102 traverses the conduit. The cleaning operation controlling module 212 compares the real-time axial position data with the recorded anomaly locations stored in the structured inspection report generated during the inspection phase. The comparison determines whether the current axial position of the crawler 102 corresponds to a location where a surface anomaly was detected during the inspection phase.
[0175] Executing cleaning operations in the inspection-guided selective cleaning mode further comprises selectively activating a cleaning mechanism and a fluid dispensing unit only at axial positions corresponding to detected surface anomalies. When the comparison indicates that the axial position of the crawler 102 corresponds to a recorded anomaly location from the structured inspection report, the cleaning operation controlling module 212 activates the cleaning mechanism 110 to perform cleaning operations at that axial position. The cleaning operation controlling module 212 also activates the fluid dispensing unit 108 to dispense cleaning fluid or oil at the axial position corresponding to the detected surface anomaly. When the axial position of the crawler 102 does not correspond to a recorded anomaly location, the cleaning mechanism 110 and the fluid dispensing unit 108 remain inactive. The selective activation based on recorded anomaly locations avoids unnecessary cleaning actions and unnecessary dispensing of cleaning media in unaffected regions of the conduit.
[0176] The method 400 further comprises adjusting at least one of cleaning duration or cleaning intensity based on the severity of detected surface anomalies at corresponding axial positions using a cleaning time estimation module 309 configured to analyse colour characteristics of deposits on the inner surface. The cleaning time estimation module 309, as described previously, analyses the colour characteristics of the deposits on the inner surface and determines corresponding cleaning parameters for effective removal based on surface condition severity. The cleaning operation controlling module 212 receives severity analysis outputs from the cleaning time estimation module 309 and adjusts the cleaning duration and cleaning intensity at each axial position where surface anomalies have been detected. Surface anomalies with higher severity receive extended cleaning duration and increased cleaning intensity, while surface anomalies with lower severity receive reduced cleaning duration and decreased cleaning intensity. The adjustment of cleaning parameters based on severity enables targeted cleaning execution that minimizes unnecessary cleaning actions while ensuring adequate treatment of identified defect or deposit regions.
[0177] The method 400 further comprises generating a structured inspection report comprising anomaly type, quantified severity metrics, and positional coordinates for each detected surface anomaly. The report generation module 214 compiles the spatially localised inspection outputs generated during the inspection phase into the structured inspection report. The structured inspection report comprises anomaly type information specifying the classification of each detected surface anomaly, including rust, carbon deposits, copper fouling, foreign deposits, and cracks. The structured inspection report comprises quantified severity metrics including defect area measurements, coverage percentage values, and geometric attribute measurements such as crack width and crack length for each detected surface anomaly. The structured inspection report comprises positional coordinates specifying the axial location of each detected surface anomaly along the length of the conduit as determined through correlation with encoder-derived position data from the position tracking mechanism 106. The structured inspection report is stored in the memory 204 of the processing unit 112 and is referenced during the cleaning phase in the inspection-guided selective cleaning mode to enable selective activation of the cleaning mechanism 110 and the fluid dispensing unit 108 at axial positions corresponding to detected surface anomalies.
[0178] The system and method for automated inspection and cleaning of an inner surface of a conduit provide AI-driven internal surface condition assessment that delivers objective quantified inspection instead of visual judgement. The AI-based computer vision models employed by the image processing module automatically detect, classify, and quantify surface anomalies present on the inner surface without reliance on subjective manual inspection. The automated detection and classification performed by the image processing module eliminates variability associated with human visual assessment, wherein different operators may reach different conclusions regarding the presence, type, or severity of surface conditions based on subjective interpretation of visual observations. The objective quantified inspection provided by the AI-driven assessment generates consistent and repeatable inspection results across multiple inspection operations, enabling reliable comparison of surface conditions over time and across different conduits. The quantified inspection outputs comprise numerical measurements of defect parameters including defect area, coverage percentage, and severity metrics that provide precise characterisation of surface conditions rather than qualitative descriptions subject to interpretation.
[0179] The system and method provide spatially localised defect mapping along the conduit length using encoder-linked localisation. The position tracking mechanism generates axial position data during traversal of the crawler within the conduit, and the image processing module correlates detected surface anomalies with corresponding axial position data to generate spatially localised inspection outputs. The encoder-linked localisation enables precise determination of the axial location of each detected surface anomaly along the longitudinal axis of the conduit, providing positional coordinates that specify where each detected condition is located within the conduit interior. The spatially localised defect mapping enables position-specific maintenance actions by identifying the exact locations within the conduit where surface anomalies are present. The spatial localisation supports targeted cleaning operations in the inspection-guided selective cleaning mode by providing the positional information required to selectively activate the cleaning mechanism and the fluid dispensing unit at axial positions corresponding to detected surface anomalies. The encoder-linked localisation provides accurate and repeatable positional measurements that enable consistent spatial mapping across multiple inspection operations.
[0180] The system and method provide pixel-level defect detection and severity measurement through micro level segmentation of detected surface anomalies. The pixel-level segmentation classifies individual pixels within the captured image data as either corresponding to anomalous surface regions or corresponding to unaffected surface regions, enabling precise delineation of the boundaries of detected surface anomalies at pixel resolution. The pixel-level detection provides accurate measurement of defect parameters including defect area and coverage percentage by counting pixels classified as anomalous and converting pixel counts to physical measurements based on known image scale factors. The pixel-level severity measurement enables differentiation between surface anomalies of varying degrees of degradation or contamination based on quantified defect parameters rather than qualitative assessment. The crack detection module estimates geometric attributes including crack width and crack length through pixel-level analysis, providing dimensional measurements that characterise the severity of detected crack formations. The pixel-level detection and severity measurement provide the quantitative data required to inform cleaning operation decisions and to assess the structural condition of the conduit based on measured defect characteristics.
[0181] The system and method provide inspection-guided selective cleaning intelligence that optimises cleaning effectiveness. The inspection-guided selective cleaning mode references the spatially localised inspection outputs generated during the inspection phase and selectively activates the cleaning mechanism and the fluid dispensing unit only at axial positions corresponding to detected surface anomalies. The selective activation based on inspection results focuses cleaning actions on regions of the inner surface where surface anomalies have been detected, while avoiding unnecessary cleaning actions in unaffected regions of the conduit. The inspection-guided selective cleaning intelligence reduces wear on the cleaning mechanism by limiting activation to locations where cleaning is required based on detected surface conditions. The selective cleaning approach minimises consumption of cleaning media dispensed by the fluid dispensing unit by avoiding unnecessary dispensing in regions where no surface anomalies were detected. The cleaning time estimation module 309 analyses colour characteristics of the deposits on the inner surface and determines corresponding cleaning duration and intensity for effective removal based on surface condition severity, enabling dynamic adjustment of cleaning parameters that provides adequate treatment of identified defect or deposit regions while minimising excessive cleaning actions. The inspection-guided selective cleaning intelligence provides targeted remediation that addresses detected surface conditions while preserving unaffected portions of the inner surface.
[0182] The system and method provide an integrated inspection and automated cleaning workflow that enables seamless transition from diagnosis to remediation. The inspection phase and the cleaning phase are performed as two coordinated software-controlled phases within the same system, with inspection data generated during the inspection phase being used to guide cleaning actions during the subsequent cleaning phase. The integration of inspection and cleaning within a single system eliminates the requirement for separate inspection equipment and cleaning equipment, reducing equipment complexity and operational overhead associated with deploying multiple systems for inspection and cleaning operations. The seamless transition from diagnosis to remediation enables cleaning operations to be performed immediately following inspection without requiring transfer of inspection data between separate systems or manual interpretation of inspection results to determine cleaning requirements. The coordinated workflow managed by the processing unit stores inspection outputs from the inspection phase and references the stored outputs during the cleaning phase to control selective activation of the cleaning mechanism and the fluid dispensing unit, providing automated coordination between inspection and cleaning operations.
[0183] The system and method provide digital inspection records that enable repeatable inspections and lifecycle traceability. The report generation module compiles inspection outputs into structured digital inspection records comprising anomaly type, quantified severity metrics, coverage percentage, positional coordinates, and visual evidence for each detected surface anomaly. The digital format of the inspection records enables electronic storage, retrieval, and processing of inspection data, supporting systematic management of inspection information across multiple inspection operations. The digital inspection records enable repeatable inspections by providing a documented baseline of surface conditions that is compared with subsequent inspection results to verify consistency of inspection outcomes and to track changes in surface condition over time. The lifecycle traceability provided by the digital inspection records enables reconstruction of the inspection and maintenance history of each conduit, documenting surface conditions detected during each inspection operation, maintenance actions performed in response to detected conditions, and outcomes of maintenance activities. The digital inspection records support historical comparison by enabling retrieval and comparison of inspection reports generated at different times for the same conduit, supporting identification of surface degradation trends and assessment of maintenance effectiveness over the operational lifetime of the conduit.
[0184] The system and method provide a low-drag architecture that enables high-speed traversal within the conduit. The crawler comprises a linear axial carriage with a low-drag structural layout that reduces mechanical resistance during movement within the conduit compared to tracked or wheeled crawler designs. The reduced mechanical drag enables faster traversal speeds during both inspection and cleaning operations, reducing the time required to complete inspection and cleaning cycles. The high-speed axial traversal capability provided by the low-drag architecture enables the system to complete inspection and cleaning operations within reduced timeframes compared to existing automated solutions that employ crawler designs with higher mechanical friction.
[0185] The system and method enable rapid completion of fused copper fouling cleaning compared to existing automated solutions. The combination of the low-drag architecture enabling high-speed traversal, the inspection-guided selective cleaning intelligence enabling targeted cleaning at detected anomaly locations, and the cleaning time estimation module 309 enabling dynamic adjustment of cleaning parameters based on deposit severity provides cleaning performance that reduces cleaning cycle duration compared to existing automated conduit cleaning systems. The reduced cleaning time enables more frequent maintenance cycles within available maintenance windows and reduces the duration that conduits are unavailable for operational use during maintenance activities. The cleaning time reduction represents a reduction in cleaning cycle duration that improves maintenance efficiency and operational readiness of conduits subjected to fused copper fouling accumulation.
[0186] Features of any of the examples or embodiments outlined above may be combined to create additional examples or embodiments without losing the intended effect. It should be understood that the description of an embodiment or example provided above is by way of example only, and various modifications could be made by one skilled in the art. Furthermore, one skilled in the art will recognise that numerous further modifications and combinations of various aspects are possible. Accordingly, the described aspects are intended to encompass all such alterations, modifications, and variations that fall within the scope of the appended claims.
, Claims:WE CLAIMS:
1. A system for automated inspection and cleaning of an inner surface of a conduit, the system comprising:
a crawler configured to traverse longitudinally within the conduit along an internal diameter of the conduit;
an imaging unit mounted on the crawler, the imaging unit configured to capture image data of the inner surface of the conduit;
a position tracking mechanism configured to generate axial position data of the crawler during traversal within the conduit;
a cleaning mechanism coupled to the crawler and configured to perform cleaning operations on the inner surface of the conduit;
a fluid dispensing unit configured to dispense cleaning fluid assisting the cleaning operations; and
a processing unit communicatively coupled to the imaging unit, the position tracking mechanism, the cleaning mechanism, and the fluid dispensing unit, the processing unit comprising an image processing module configured to:
receive the image data captured by the imaging unit;
detect surface anomalies on the inner surface of the conduit based on analysis of the image data, wherein the surface anomalies comprise at least one of rust, carbon deposits, copper fouling, foreign deposits, or cracks;
classify the surface anomalies on the inner surface of the conduit,
perform micro level segmentation of detected surface anomalies to quantify defect parameters comprising of defect area,
correlate detected surface anomalies with corresponding axial position data from the position tracking mechanism to generate spatially localised inspection outputs;
where in the processing unit is further configured to control the cleaning mechanism and the fluid dispensing unit to execute cleaning operations on the inner surface of the conduit.

2. The system as claimed in claim 1, wherein the imaging unit comprises at least one camera and at least one illumination source.
3. The system as claimed in claim 1, wherein the position tracking mechanism is embodied as an encoder-based position tracking mechanism.

4. The system as claimed in claim 1, wherein the processing unit is configured to operate in a full-coverage cleaning mode in which the cleaning mechanism and the fluid dispensing unit are activated to perform cleaning operations along an entire length of the conduit independent of the spatially localised inspection outputs.

5. The system as claimed in claim 1, wherein the processing unit is configured to operate in an inspection-guided selective cleaning mode in which the cleaning mechanism and the fluid dispensing unit are selectively activated only at axial positions corresponding to detected surface anomalies based on the spatially localised inspection outputs.

6. The system as claimed in claim 3, wherein the processing unit is further configured to adjust at least one of cleaning duration or cleaning intensity based on the severity of detected surface anomalies at corresponding axial positions using a cleaning time estimation module.

7. The system as claimed in any of claims 1-6, the image processing module comprises:
a rust detection module configured to perform image-level rust detection, rust localisation, and pixel-level segmentation of rusted regions;
a surface condition detection module configured to detect and quantify carbon deposits, copper fouling, dirt, and contaminants; and
a crack detection module configured to perform crack detection with pixel-level segmentation of crack regions, and to estimate crack width and crack length.

8. The system as claimed in claim 7, wherein the image processing module further comprises a foreign deposit detection module configured to identify non-native substances, isolate deposit regions, and compute coverage percentage.

9. The system as claimed in claim 1, wherein:
the image processing module is configured to apply a predefined region of interest to a field of view of at least one camera of the imaging unit to define a controlled inspection region, and
wherein the image processing module is configured to extract pixel colour values within the region of interest and compare the extracted pixel colour values with pre-stored threshold ranges associated with pre-defined anomaly classes to classify detected regions.

10. The system as claimed in claim 1, wherein the processing unit is further configured to generate a structured inspection report comprising at least anomaly type, quantified severity metrics, coverage percentage, and positional coordinates for each detected surface anomaly.

11. The system as claimed in claim 10, wherein the processing unit is configured to store the structured inspection report in a digital format for historical comparison and lifecycle traceability.

12. The system as claimed in claim 1, wherein:
the fluid dispensing unit comprises a pump mechanism configured to dispense at least one of cleaning fluid or oil, and
wherein the processing unit is configured to synchronise pump activation with axial position data from the position tracking mechanism.

13. The system as claimed in claim 12, wherein the processing unit is configured to:
receive an instruction indicative of controlling the cleaning mechanism and the fluid dispensing mechanism, wherein the processing unit is configured to receive the instruction from a remote electronic device via a user interface; and
operate at least one of the cleaning mechanisms and the fluid dispensing mechanism based on the received instruction.

14. The system as claimed in claim 1, wherein the crawler comprises a linear axial carriage with a low-drag architecture, and wherein the system is scalable for the conduit having diameters ranging from 76 mm to 155 mm.

15. A method for automated inspection and cleaning of an inner surface of a conduit, comprising:
traversing a crawler longitudinally within the conduit along an internal diameter of the conduit;
capturing, by an imaging unit mounted on the crawler, image data of the inner surface of the conduit during traversal;
generating, by a position tracking mechanism, axial position data of the crawler during traversal;
processing, by an image processing module, the captured image data to detect and classify surface anomalies on the inner surface of the projectile launching device, wherein the surface anomalies comprise at least one of rust, carbon deposits, copper fouling, foreign deposits, and cracks;
performing pixel-level segmentation of detected surface anomalies to quantify defect parameters comprising at least one of defect area, coverage percentage, and severity;
correlating detected surface anomalies with corresponding axial position data to generate spatially localised inspection outputs;
executing cleaning operations on the inner surface of the conduit.

16. The method as claimed in claim 15 further comprising:
generating a structured inspection report comprising anomaly type, quantified severity metrics, and positional coordinates for each detected surface anomaly.

17. The method as claimed in claim 15 further comprising:
executing cleaning operations on the inner surface of the conduit based on one of a full-coverage cleaning mode or an inspection-guided selective cleaning mode.

18. The method as claimed in claim 17, wherein executing cleaning operations in the full-coverage cleaning mode comprises:
activating a cleaning mechanism and a fluid dispensing unit to perform cleaning operations along an entire length of the conduit independent of the spatially localised inspection outputs.

19. The method as claimed in claim 17, wherein executing cleaning operations in the inspection-guided selective cleaning mode comprises:
comparing real-time axial position data with recorded anomaly locations from the structured inspection report; and
selectively activating a cleaning mechanism and a fluid dispensing unit only at axial positions corresponding to detected surface anomalies.

20. The method as claimed in claim 19, further comprising adjusting at least one of cleaning duration or cleaning intensity based on the severity of detected surface anomalies at corresponding axial positions using a cleaning time estimation module.

21. The method as claimed in claim 15, wherein detecting and classifying surface anomalies comprises:
performing rust detection using image-level rust detection, rust localization, and pixel-level segmentation of rusted regions;
detecting and quantifying carbon deposits, copper fouling, dirt, and contaminants; and
performing crack detection comprising crack detection pixel-level segmentation of crack regions, and estimation of crack width and crack length.