Intelligent positioning system for valve hall inspection device

Abstract

The invention discloses an intelligent positioning system for a valve hall inspection device, which comprises a visual sensor and an industrial personal computer, the visual sensor and the industrial personal computer are arranged on the inspection device, and the visual sensor is electrically connected with the industrial personal computer; the industrial personal computer is provided with an image feature extraction module, an image feature matching module, a data processing module and an output module; the image feature matching module is electrically connected with the image feature extraction module and the data processing module; and the data processing module is electrically connected with the output module. According to the intelligent positioning system, the inspection device is not limited by a fixed route, and the inspection device can complete self positioning at any position. The visual sensor adopts a depth camera; and the data processing module performs iterative calculation on the two groups of matching point sets through an ICP (Inductively Coupled Plasma) algorithm. And expensive positioning equipment is replaced by the depth camera, so that the hardware cost is reduced. And the positioning precision of the intelligent positioning system is improved by combining the depth camera with feature matching and an ICP (Inductively Coupled Plasma) algorithm.

Description

Technical Field

[0001] This invention relates to the field of intelligent positioning system technology, and in particular to an intelligent positioning system for a valve hall inspection device. Background Technology

[0002] Converter stations are stations built in high-voltage direct current (HVDC) transmission systems to convert AC to DC power and meet the power system's requirements for safety, stability, and power quality. Valve halls, enclosed structures housing converter valves, are the core equipment of converter stations, and their operational status directly affects the stable operation of the entire power system. With the development of power line inspection, conventional manual inspections are gradually being replaced by intelligent inspection devices. The precise positioning of these devices is crucial for achieving autonomous inspection and fault location.

[0003] Currently, conventional inspection methods include RFID tags and track positioning. Among them, RFID tag positioning requires the pre-deployment of a large number of tags in the environment, which results in high equipment costs and susceptibility to signal interference; track positioning is limited by fixed paths, restricting the movement range of the inspection device and failing to meet the flexible inspection needs in complex environments. Summary of the Invention

[0004] To overcome the above-mentioned shortcomings of the prior art, the present invention provides an intelligent positioning system for valve hall inspection devices.

[0005] The technical solution adopted by this invention to solve its technical problem is: an intelligent positioning system for a valve hall inspection device, comprising: a vision sensor and an industrial control computer, wherein the vision sensor and the industrial control computer are installed on the inspection device, and the vision sensor is electrically connected to the industrial control computer; the industrial control computer is equipped with an image feature extraction module, an image feature matching module, a data processing module, and an output module; the image feature matching module is electrically connected to the image feature extraction module and the data processing module respectively; the data processing module is electrically connected to the output module.

[0006] As a further improvement of the present invention: the image feature extraction module is used to extract feature points of the environmental image; the image feature matching module is used to match feature points of two adjacent frames of images to obtain two sets of matching points; the data processing module is used to calculate the pose transformation between two adjacent frames of images based on the matching point sets; the output module is used to output the cumulative pose information of the inspection device; wherein, the data processing module obtains the real-time pose of the inspection device relative to the initial position by iteratively calculating and accumulating the pose transformation of multiple frames of images. The image feature extraction module and the image feature matching module work together efficiently to accurately obtain the correspondence between feature points of adjacent frames; the data processing module achieves high-precision positioning by iteratively calculating and accumulating the pose transformation; the output module provides real-time feedback of the cumulative pose information. The cooperation of the image feature extraction module, the image feature matching module, the data processing module, and the output module ensures that the inspection device can still autonomously construct a trajectory based on environmental features even in an environment without preset markers, significantly improving the flexibility and accuracy of positioning.

[0007] As a further improvement of the present invention, the vision sensor adopts a depth camera. By using a depth camera with the vision sensor, color images and depth images can be acquired simultaneously. The depth information provides three-dimensional spatial data, upgrading the positioning from two-dimensional to three-dimensional, and significantly improving accuracy. Moreover, under complex lighting conditions in the valve hall, the depth information can reduce feature mismatches, providing a reliable basis for subsequent pose calculations and adapting to the needs of inspection and positioning.

[0008] As a further improvement of the present invention, the visual sensor adopts a Kinect depth camera. By using a Kinect depth camera for the visual sensor, color images and depth images can be quickly output, adapting to the dynamic movement requirements of the inspection device. Its cost is lower than that of professional industrial depth cameras, and data acquisition can be completed without additional components, simplifying system deployment. Even in the enclosed space of the valve hall, it can still accurately capture the depth information of objects, improve the stability of feature matching, and ensure positioning accuracy.

[0009] As a further improvement of the present invention: the feature points extracted by the image feature extraction module are SIFT, SURF, or ORB feature points. SIFT and SURF feature points are highly resistant to changes in lighting and scale, making them suitable for complex lighting environments in valve halls; ORB feature points have a fast calculation speed, meeting the real-time requirements of inspection; and the ability to flexibly switch between SIFT, SURF, and ORB feature points allows for the selection of the optimal solution based on the scene, ensuring stable feature extraction even in areas with rich textures or variable lighting.

[0010] As a further improvement of the present invention: the industrial control computer adopts the MIO-5272U-U6A1E industrial control computer. By using the MIO-5272U-U6A1E industrial control computer, it has the advantages of strong computing performance and resistance to electromagnetic interference, can quickly process tasks such as image feature extraction and matching, is suitable for strong electromagnetic environments, has rich interfaces, and can smoothly connect to depth cameras.

[0011] As a further improvement of the present invention, the image feature matching module employs the Fast Approximate Nearest Neighbor (FLANN) algorithm to match adjacent image frames. Compared to traditional algorithms, the FLANN algorithm, through its index-based fast search for matching points, significantly improves the matching speed between adjacent image frames, adapting to dynamic inspection scenarios. Furthermore, even in complex environments like valve halls, it can accurately select matching point pairs, providing reliable data for subsequent pose calculations, balancing efficiency and accuracy. By using ORB feature points and the FLANN matching algorithm, the computational load is reduced, ensuring real-time output of positioning results and meeting the dynamic positioning requirements of the inspection device, thus possessing the advantage of strong real-time performance.

[0012] As a further improvement of the present invention: the extraction of feature points from the environmental image specifically includes: the image feature extraction module processes the color image acquired by the depth camera, selects pixel point P, and selects 16 pixels on a circle with P as the center and a radius of 3 pixels; if there are N consecutive pixels with pixel values ​​greater than I+a or less than Ia, then P is determined to be an ORB feature point, and finally a feature point set is generated, where I is the pixel value of point P and a is the threshold. By using ORB feature points, selecting 16 pixels on a circle with a fixed radius, and combining the pixel value threshold judgment, the rules are clear and the computational load is small, resulting in fast extraction speed; the judgment condition of N consecutive points can effectively filter noise and ensure the stability of feature points; it adapts to the color image of the depth camera, providing reliable features for subsequent matching.

[0013] As a further improvement of the present invention: the data processing module iteratively calculates the position and pose changes of the camera between two sets of matching points using the ICP algorithm; the position and pose changes are continuously accumulated to obtain the real-time pose of the camera relative to its initial position. By iteratively calculating the two sets of matching points using the ICP algorithm through the data processing module, the camera pose changes between adjacent frames can be calculated with high precision, providing accurate data for positioning; the continuous accumulation mechanism avoids reliance on a single reference point, can construct a complete motion trajectory, and enables the inspection device to achieve autonomous positioning in an environment without preset markers, significantly improving the system's flexibility and adaptability.

[0014] As a further improvement of the present invention: the data processing module iteratively calculates the two sets of matching points using the ICP algorithm to obtain the position and pose changes of the camera between the two frames of images, specifically including:

[0015] 1) Construct the centroid-free coordinates of the matching point set:

[0016]

[0017] Where, p i and p i '' represents the set of matching points corresponding to two adjacent frames, and q represents the set of matching points. i and q i ′ are the centroid coordinates of the matching point set, respectively;

[0018] 2) Calculate the rotation matrix R and the translation vector t:

[0019] R = UV T

[0020]

[0021] Where U and V are matrices The SVD decomposition results.

[0022] Compared with the prior art, the beneficial effects of the present invention are:

[0023] 1. The intelligent positioning system of the present invention enables the inspection device to be free from the restriction of fixed routes, and the inspection device can complete its own positioning no matter where it is located; by using a depth camera to replace expensive positioning equipment, the hardware cost is reduced; by combining the depth camera with feature matching and ICP algorithm, the positioning accuracy of the intelligent positioning system is improved.

[0024] 2. The intelligent positioning system of the valve hall inspection device of this invention integrates a depth camera, feature extraction, and ICP algorithm to achieve high-precision autonomous positioning. By using a Kinect depth camera to simultaneously acquire color and depth maps, it provides three-dimensional spatial data for positioning, improving positioning accuracy by 40% compared to traditional monocular vision solutions. The MIO-5272U-U6A1E industrial control computer has anti-electromagnetic interference capabilities, improving data processing stability in the strong electromagnetic environment of the valve hall. Its compact size and efficient heat dissipation design are suitable for the limited space of the inspection device. The ICP algorithm iteratively optimizes the matching point set, resulting in small pose calculation errors. By continuously accumulating pose changes, the system can construct a complete motion trajectory, achieving fully autonomous positioning without the need for preset markers. Through hardware collaboration between the depth camera and the industrial control computer, combined with feature extraction, matching, and pose calculation, the system response time is shortened, and inspection efficiency is improved. Attached Figure Description

[0025] Figure 1 This is a structural block diagram of the present invention.

[0026] Figure 2 This is a structural block diagram of the industrial control computer of the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. It should be noted that the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products, or devices.

[0028] Please see Figure 1-2 A smart positioning system for a valve hall inspection device includes: a vision sensor and an industrial control computer, wherein the vision sensor and the industrial control computer are installed on the inspection device, and the vision sensor is electrically connected to the industrial control computer; the industrial control computer is equipped with an image feature extraction module, an image feature matching module, a data processing module, and an output module; the image feature matching module is electrically connected to the image feature extraction module and the data processing module respectively; the data processing module is electrically connected to the output module.

[0029] The inspection device achieves positioning by collecting environmental features through visual sensors. It can autonomously locate itself at any position and is suitable for flexible inspection in complex environments.

[0030] The intelligent positioning system of this valve hall inspection device is not limited by fixed routes. It relies on visual sensors to collect environmental images and does not require preset tracks or tags. The inspection device can be located anywhere in the valve hall, adapting to complex environments. At the same time, the intelligent positioning system of this valve hall inspection device uses a combination of visual sensors and industrial control computers, eliminating the need for expensive equipment and reducing costs. In addition, the image feature extraction module, image feature matching module, data processing module, and output module of the industrial control computer work together efficiently to output positioning results in real time, with strong real-time performance.

[0031] In some implementations, the image feature extraction module is used to extract feature points from the environmental image; the image feature matching module is used to match feature points of two adjacent frames of images to obtain two sets of matching points; the data processing module is used to calculate the pose transformation between two adjacent frames of images based on the matching point sets; and the output module is used to output the cumulative pose information of the inspection device; wherein, the data processing module obtains the real-time pose of the inspection device relative to its initial position by iteratively calculating and accumulating the pose transformation of multiple frames of images.

[0032] The system acquires environmental images using a visual sensor and transmits them to an image feature extraction module to extract feature points. Then, the image feature matching module performs image matching on two adjacent frames to obtain two sets of matched points. Through the data processing module, the position and pose changes of the camera between the two sets of points are iteratively calculated. By accumulating the poses between the calculated consecutive images and outputting the accumulated poses, the current position and pose of the camera relative to the origin are obtained.

[0033] The intelligent positioning system of the present invention enables the inspection device to be free from the restriction of fixed routes, and the inspection device can complete its own positioning no matter where it is.

[0034] In some implementations, the vision sensor is a depth camera.

[0035] By employing depth cameras with visual sensors, richer data dimensions are achieved. Compared to monocular cameras that can only acquire two-dimensional images, depth cameras can simultaneously capture color images and depth maps. The depth map contains depth information for each pixel, providing three-dimensional spatial data for positioning and significantly improving positioning accuracy. At the same time, depth cameras are more adaptable to different environments. In complex lighting or occluded scenes, depth information can help distinguish the spatial location of objects and reduce feature mismatches. In addition, by using depth cameras to replace expensive positioning equipment (such as LiDAR), hardware costs can be reduced.

[0036] In some implementations, the visual sensor is a Kinect depth camera.

[0037] The Kinect depth camera, utilizing a visual sensor, offers advantages such as high cost-effectiveness, strong real-time performance, good environmental adaptability, and high integration. Compared to professional industrial depth cameras, the Kinect depth camera is more cost-effective and can stably output color images and high-precision depth maps, balancing economy and data quality. Secondly, the hardware-accelerated depth computing capabilities of the Kinect depth camera can quickly acquire 3D environmental data, adapting to the dynamic movement of inspection devices. In addition, the Kinect depth camera has good environmental adaptability, accurately capturing object depth information even under complex lighting conditions in the enclosed space of the valve hall, reducing feature matching errors. At the same time, the Kinect depth camera has high integration, completing data acquisition without additional components, simplifying system deployment and better meeting the compact installation requirements of valve hall inspection.

[0038] In some implementations, the feature points extracted by the image feature extraction module are ORB, SIFT, or SURF feature points.

[0039] The feature points extracted by the image feature extraction module are ORB, SIFT, or SURF feature points, which can be adapted to different scenarios. ORB is fast and has strong real-time performance, making it suitable for the dynamic movement of inspection devices. SIFT and SURF have excellent resistance to illumination and scale changes, and can cope with complex lighting or distance changes in the valve hall. ORB, SIFT, or SURF feature points can all extract stable features from the texture and structure of the valve hall equipment, reducing false matching and achieving high matching accuracy.

[0040] In some implementations, the industrial computer is an MIO-5272U-U6A1E industrial computer.

[0041] The MIO-5272U-U6A1E industrial PC is used. The MIO-5272U-U6A1E boasts strong computing performance; its high-efficiency processor can quickly support computational tasks such as image feature extraction, matching, and ICP algorithms, ensuring real-time processing of positioning data. Simultaneously, the MIO-5272U-U6A1E exhibits excellent stability and strong anti-electromagnetic interference capabilities, adapting to strong electromagnetic environments and preventing data processing interruptions. The MIO-5272U-U6A1E has high integration, stably supporting modules such as image feature extraction, and offers rich interfaces for seamless connection to depth cameras, simplifying system connectivity. Its compact size is suitable for installations in limited space on inspection devices, while its excellent heat dissipation ensures continuous operation during extended inspection periods.

[0042] In some implementations, the image feature matching module uses the Fast Approximate Nearest Neighbor (FLANN) algorithm to match two adjacent frames of images, resulting in two sets of matching points.

[0043] The image feature matching module employs the FLANN algorithm, which boasts advantages such as fast matching speed, high matching accuracy, and strong adaptability. By constructing an index, it quickly searches for approximate nearest neighbor feature points, significantly reducing the amount of computation and adapting to the real-time matching requirements of the inspection device during dynamic movement. In the complex environment of the valve hall, it can effectively filter out true matching point pairs between adjacent frames, reducing false matching. It can still process images with a large number of feature points efficiently, and even if the equipment features in the valve hall are dense, it can stably output the matching point set, providing reliable data support for subsequent pose calculations.

[0044] By employing ORB feature points and FLANN matching algorithms, the computational load is reduced, ensuring real-time output of positioning results and meeting the dynamic positioning requirements of inspection devices, thus offering the advantage of strong real-time performance.

[0045] In some implementations, the extraction of feature points from the environmental image specifically includes: the image feature extraction module processes the color image acquired by the depth camera, selects pixel P, and selects 16 pixels on a circle with P as the center and a radius of 3 pixels; if there are N consecutive pixels with pixel values ​​greater than I+a or less than Ia, then P is determined to be an ORB feature point, and finally a feature point set is generated, where I is the pixel value of point P and a is the threshold.

[0046] In some implementations, the data processing module iteratively calculates the two sets of matching points using the ICP algorithm to obtain the position and pose changes of the camera between the two frames; the position and pose changes are continuously accumulated to obtain the real-time pose of the camera relative to its initial position.

[0047] By combining depth cameras with feature matching and ICP algorithms, the positioning accuracy of the intelligent positioning system is improved. Depth cameras can simplify the calculation logic, and 3D data can be directly used for ICP algorithms to solve for pose without the need for additional depth estimation, thereby improving the real-time positioning performance and making it more suitable for the dynamic positioning needs of inspection devices.

[0048] In some implementations, the data processing module iteratively calculates the position and pose changes of the camera between the two sets of matching points using the ICP algorithm, specifically including:

[0049] 1) Construct the centroid-free coordinates of the matching point set:

[0050]

[0051] Where, p i and p i '' represents the set of matching points corresponding to two adjacent frames, and q represents the set of matching points. i and q i ′ are the centroid coordinates of the matching point set, respectively;

[0052] 2) Calculate the rotation matrix R and the translation vector t:

[0053] R = UV T

[0054]

[0055] Where U and V are matrices The SVD decomposition results.

[0056] The intelligent positioning system of this valve hall inspection device has high positioning accuracy. Through image feature extraction (such as ORB feature points), matching and ICP algorithm processing, it can accurately calculate pose changes and accumulate the real-time position.

[0057] Working principle of the invention:

[0058] The visual sensor acquires environmental images and transmits them to the image feature extraction module to extract ORB feature points. Then, the image feature matching module performs image matching on two adjacent frames to obtain two sets of matched points. Through the data processing module, the position and pose changes of the camera between the two sets of points are calculated iteratively. By accumulating the poses between the calculated consecutive images and outputting the accumulated poses, the current position and pose of the camera relative to the origin are obtained.

[0059] Implementation Case 1:

[0060] An intelligent positioning system for a valve hall inspection device includes: a vision sensor and an industrial control computer. The vision sensor and industrial control computer are mounted on the inspection device, and the vision sensor is electrically connected to the industrial control computer. The industrial control computer is equipped with an image feature extraction module, an image feature matching module, a data processing module, and an output module. The image feature matching module is electrically connected to both the image feature extraction module and the data processing module. The data processing module is electrically connected to the output module. Positioning is achieved by collecting environmental features through the vision sensor. The inspection device can autonomously locate itself at any position, making it suitable for flexible inspections in complex environments.

[0061] The image feature extraction module is used to extract feature points from the environmental image; the image feature matching module is used to match feature points of two adjacent frames of images to obtain two sets of matching points; the data processing module is used to calculate the pose transformation between two adjacent frames of images based on the matching point sets; the output module is used to output the cumulative pose information of the inspection device; wherein, the data processing module obtains the real-time pose of the inspection device relative to the initial position by iteratively calculating and accumulating the pose transformation of multiple frames of images.

[0062] The vision sensor employs a depth camera. Compared to a monocular camera, the image captured by a depth camera includes both a color image and a depth map. The depth map contains the depth information of corresponding pixels in the color image. By using a depth camera instead of expensive positioning devices (such as LiDAR), hardware costs are reduced.

[0063] The image feature extraction module extracts ORB feature points. The industrial control computer used is an MIO-5272U-U6A1E. The image feature matching module uses the Fast Approximate Nearest Neighbor (FLANN) algorithm to match adjacent image frames, obtaining two sets of matching points. By employing ORB feature points and the FLANN matching algorithm, the computational load is reduced, ensuring real-time output of positioning results and meeting the dynamic positioning requirements of the inspection device, thus offering the advantage of strong real-time performance.

[0064] The extraction of feature points from the environmental image specifically includes: the image feature extraction module processes the color image acquired by the depth camera, selects pixel point P, and selects 16 pixel points on a circle with P as the center and a radius of 3 pixels; if there are N consecutive pixel points with pixel values ​​greater than I+a or less than Ia, then P is determined to be an ORB feature point, and finally a feature point set is generated, where I is the pixel value of point P and a is the threshold.

[0065] The data processing module iteratively calculates the position and pose changes of the camera between the two sets of matching points using the ICP algorithm. These position and pose changes are then continuously accumulated to obtain the camera's real-time pose relative to its initial position. By combining a depth camera with feature matching and the ICP algorithm, the positioning accuracy of the intelligent positioning system is improved.

[0066] The data processing module iteratively calculates the two sets of matching points using the ICP algorithm to obtain the changes in the camera's position and pose between the two frames, specifically including:

[0067] 1) Construct the centroid-free coordinates of the matching point set:

[0068]

[0069] Where, p i and p i '' represents the set of matching points corresponding to two adjacent frames, and q represents the set of matching points. i and q i ′ are the centroid coordinates of the matching point set, respectively;

[0070] 2) Calculate the rotation matrix R and the translation vector t:

[0071] R = UV T

[0072]

[0073] Where U and V are matrices The SVD decomposition results.

[0074] The system acquires environmental images using a visual sensor and transmits them to an image feature extraction module to extract feature points. Then, the image feature matching module performs image matching on two adjacent frames to obtain two sets of matched points. Through the data processing module, the position and pose changes of the camera between the two sets of points are iteratively calculated. By accumulating the poses between the calculated consecutive images and outputting the accumulated poses, the current position and pose of the camera relative to the origin are obtained.

[0075] Implementation Case 2:

[0076] like Figure 1-2As shown, an intelligent positioning system for a valve hall inspection device includes: a vision sensor and an industrial control computer (ICC). The vision sensor and ICC are mounted on the inspection device, and the vision sensor is electrically connected to the ICC. The ICC is equipped with an image feature extraction module, an image feature matching module, a data processing module, and an output module. The image feature matching module is electrically connected to both the image feature extraction module and the data processing module. The data processing module is electrically connected to the output module. The vision sensor is a Kinect depth camera, and the ICC is an MIO-5272U-U6A1E IC. The feature points extracted by the image feature extraction module are ORB feature points.

[0077] The workflow of this implementation case is as follows:

[0078] When the inspection device moves, the vision sensor acquires environmental images with depth information and transmits the environmental images to the image feature extraction module.

[0079] The image feature extraction module extracts ORB feature points from the environmental image. The acquisition method is as follows: Select pixel P in the image, assuming its value is I. Set a threshold a, and select 16 pixels on a circle with a radius of 3 centered at pixel P. If N consecutive pixels among these 16 pixels have pixel values ​​greater than I+a or less than Ia, then the current pixel P is identified as a feature point. Perform the above operation on each pixel to extract the feature points from the image.

[0080] The image feature matching module uses the Fast Approximate Nearest Neighbor (FLANN) algorithm to perform image matching between two adjacent frames, resulting in two sets of matched points.

[0081] Since a Kinect depth camera is used, the matched point pairs consist of two 3D coordinates. The ICP algorithm is used in the data processing module to solve the motion estimation problem between the two matched points. The solution is the motion relationship of the camera between the two frames. The ICP algorithm solution formula is as follows:

[0082] Define the error of the i-th 3D matching point as:

[0083] e i =p i -(Rp′ i +t)

[0084] In the formula, e i To solve for the error, p i Let ' be a point from the previous image, and p i Let R be the matching point corresponding to the adjacent image, R be the rotation matrix, and t be the translation matrix.

[0085] Constructing the least squares problem:

[0086]

[0087] Define matrix:

[0088]

[0089] Constructing the centroid-free coordinates of the matching point set:

[0090]

[0091] Where, p i and p i '' represents the set of matching points corresponding to two adjacent frames, and q represents the set of matching points. i and q i ′ are the centroid coordinates of the matching point set, respectively;

[0092] Perform SVD decomposition on the W matrix:

[0093] W=U∑V T

[0094] The rotation matrix R and translation vector t are obtained by solving:

[0095] R = UV T

[0096]

[0097] The calculated R is the rotation matrix, which represents the rotation of the camera relative to the three coordinate axes between two frames, and t is the translation matrix, which represents the translation of the camera in the X, Y, and Z directions.

[0098] By accumulating the rotation matrix R and the translation vector t, we can obtain the coordinates ts of the current camera, i.e. the inspection device, relative to the starting coordinate system. If we plot ts at each moment, we can obtain the movement trajectory of the camera from the starting position, i.e. the movement trajectory of the inspection robot.

[0099] The resulting movement trajectory is output to the computer of the inspection device and the converter station through the output module, so that the inspection device and the staff can obtain the location information of the inspection device in real time.

[0100] The main functions of this invention are:

[0101] The intelligent positioning system of this invention enables the inspection device to be unrestricted by fixed routes, allowing it to locate itself from any position. By using a depth camera instead of expensive positioning equipment, hardware costs are reduced; by combining the depth camera with feature matching and the ICP algorithm, the positioning accuracy of the intelligent positioning system is improved. The intelligent positioning system for the inspection device of this invention has the advantages of low cost, accurate positioning, and freedom from fixed route restrictions. The intelligent positioning system for the valve hall inspection device of this invention integrates a depth camera, feature extraction, and the ICP algorithm to achieve high-precision autonomous positioning; by using a Kinect depth camera to simultaneously acquire color and depth maps, it provides three-dimensional spatial data for positioning, improving positioning accuracy by 40% compared to traditional monocular vision solutions. The MIO-5272U-U6A1E industrial control computer has anti-electromagnetic interference capabilities, improving data processing stability in the strong electromagnetic environment of the valve hall; its compact size and efficient heat dissipation design are suitable for the limited space of the inspection device. The ICP algorithm iteratively optimizes the matching point set, resulting in small pose calculation errors. By continuously accumulating pose changes, the system can construct a complete motion trajectory, achieving fully autonomous positioning without the need for preset markers. By combining the hardware collaboration between the depth camera and the industrial control computer with a software closed loop of feature extraction, matching, and pose calculation, the system has a short response time and improves inspection efficiency.

[0102] The functions described herein may be implemented in hardware, software executed by a processor, firmware, or any combination thereof. If implemented in software executed by a processor, the functions may be stored as one or more instructions or code on or transmitted via a computer-readable medium. Other embodiments and implementations are within the scope and spirit of this invention and the appended claims. For example, due to the nature of software, the functions described above may be implemented using software executed by a processor, hardware, firmware, hardwired, or any combination thereof. Furthermore, the functional units may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit.

[0103] In the implementation examples provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection can be through some interfaces; the indirect coupling or communication connection between units or modules can be electrical or other forms.

[0104] The units described as separate components may or may not be physically separate. Similarly, the components of the control device may or may not be physical units; they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0105] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0106] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. An intelligent positioning system for a valve hall inspection device, characterized in that: include: A vision sensor and an industrial control computer are mounted on an inspection device. The vision sensor is electrically connected to the industrial control computer. The industrial control computer is equipped with an image feature extraction module, an image feature matching module, a data processing module, and an output module. The image feature matching module is electrically connected to both the image feature extraction module and the data processing module. The data processing module is electrically connected to the output module.

2. The intelligent positioning system for a valve hall inspection device according to claim 1, characterized in that: The image feature extraction module is used to extract feature points from the environmental image; the image feature matching module is used to match feature points of two adjacent frames of images to obtain two sets of matching points; the data processing module is used to calculate the pose transformation between two adjacent frames of images based on the matching point sets; the output module is used to output the cumulative pose information of the inspection device; wherein, the data processing module obtains the real-time pose of the inspection device relative to the initial position by iteratively calculating and accumulating the pose transformation of multiple frames of images.

3. The intelligent positioning system for a valve hall inspection device according to claim 1, characterized in that: The visual sensor is a depth camera.

4. The intelligent positioning system for a valve hall inspection device according to claim 3, characterized in that: The visual sensor used is a Kinect depth camera.

5. The intelligent positioning system for a valve hall inspection device according to claim 2, characterized in that: The image feature extraction module extracts SIFT, SURF, or ORB feature points.

6. The intelligent positioning system for a valve hall inspection device according to claim 1, characterized in that: The industrial control computer used is the MIO-5272U-U6A1E industrial control computer.

7. The intelligent positioning system for a valve hall inspection device according to claim 2, characterized in that: The image feature matching module uses a fast approximate nearest neighbor algorithm to match two adjacent image frames.

8. The intelligent positioning system for a valve hall inspection device according to claim 2, characterized in that: The extraction of feature points from the environmental image specifically includes: the image feature extraction module processes the color image acquired by the depth camera, selects pixel point P, and selects 16 pixel points on a circle with P as the center and a radius of 3 pixels; if there are N consecutive pixel points with pixel values ​​greater than I+a or less than Ia, then P is determined to be an ORB feature point, and finally a feature point set is generated, where I is the pixel value of point P and a is the threshold.

9. The intelligent positioning system for a valve hall inspection device according to claim 2, characterized in that: The data processing module iteratively calculates the two sets of matching points using the ICP algorithm to obtain the position and pose changes of the camera between the two frames; the position and pose changes are continuously accumulated to obtain the real-time pose of the camera relative to the initial position.

10. The intelligent positioning system for a valve hall inspection device according to claim 9, characterized in that: The data processing module iteratively calculates the two sets of matching points using the ICP algorithm to obtain the changes in the camera's position and pose between the two frames, specifically including: 1) Construct the centroid-free coordinates of the matching point set: Where, p i and p i '' represents the set of matching points corresponding to two adjacent frames, and q represents the set of matching points. i and q i ′ are the centroid coordinates of the matching point set, respectively; 2) Calculate the rotation matrix R and the translation vector t: R=UV T Where U and V are matrices The SVD decomposition results.

Images