Visual positioning method and system for tire mold automatic drilling robot

Abstract

The invention provides a visual positioning method and system for an automatic drilling robot of a tire mold, and the method comprises the steps: calibrating the internal and external parameters of a camera, and obtaining the target surface identification of the tire mold and the workpiece coordinates of key points; based on the image feature template, identifying a pixel coordinate of the target surface identifier and an actual physical size corresponding to a single pixel; according to the internal reference of the camera and the workpiece coordinates and pixel coordinates of the target surface identifier, the initial conversion relation of the workpiece relative to the robot coordinate system is solved; pixel coordinates of the key points are obtained based on the initial conversion relation; calculating actual coordinates of a key point robot coordinate system according to the pixel coordinates of the key point, the photographing pose and the internal reference of the camera; and matching the three-dimensional coordinates of the key points with coordinates of a robot coordinate system through a singular value decomposition algorithm, and solving an accurate conversion relation of the workpiece relative to the robot coordinate system. A target surface and key points are recognized through template matching and machine vision, and the accurate conversion relation between a workpiece and a robot coordinate system is calculated, so that efficient positioning of the tire mold exhaust hole is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of robot control, and particularly to a visual positioning method and system for an automatic drilling robot of a tire mold. Background Art

[0002] A tire mold is a mold used for vulcanizing and forming various tires. During the tire vulcanization process, in order to discharge the air in the mold cavity, air holes need to be opened on the tread of the tire mold. Such air holes are numerous, irregularly distributed in position, and the axes of the air holes need to be perpendicular to the tread where they are located, with a large range of direction changes. Currently, for the processing of air holes in tire molds, manual drilling is mostly used. Manual processing has a large degree of randomness, and it is difficult to ensure processing consistency; high-intensity operations are accompanied by mechanical vibrations and noise pollution, and long-term engagement in this position is likely to cause occupational health risks, including but not limited to auditory damage, musculoskeletal strain, etc.

[0003] The invention patent with the application number 202410961287.8 discloses a robot system for automatic drilling of tire molds and its control method, which uses a contact probe to locate the exhaust holes by detecting the outer contour of the tire mold. The invention patent with the application number 202310685202.3 discloses a special full-automatic drilling machine for tire molds, which realizes the positioning of exhaust holes through a 3D scanner.

[0004] The above two positioning methods both save manpower, reduce the labor intensity of operators, and improve work efficiency. However, the contact probe uses a contact-type point-by-point measurement method to achieve positioning, which has disadvantages such as contact, low speed, and high cost; the 3D scanner needs to be equipped with special CAM software, and the point cloud data generated by its scanning needs to solve each exhaust hole one by one through algorithms, with a large amount of calculation, high requirements for hardware performance, and being greatly affected by noise, and the equipment cost is high. Summary of the Invention

[0005] In order to solve the above problems, the present invention proposes a visual positioning method and system for an automatic drilling robot of a tire mold, which respectively identify the target surface and key points through template matching and machine vision, reduce the influence of noise, and realize the positioning of the exhaust holes of the tire mold by calculating the precise conversion relationship between the workpiece and the robot coordinate system, thereby saving time and improving the positioning efficiency.

[0006] To achieve the above object, the present invention adopts the following technical solutions: [[ID=2s]]On the first aspect, the present invention provides a visual positioning method for an automatic drilling robot of a tire mold, including: Calibrating the internal and external parameters of the camera, making an image feature template and training the YOLO model; Extracting the workpiece coordinates of the target surface mark and key points on the tire mold; Based on the image feature template, identify the target surface of the tire mold, and obtain the pixel coordinates of the identification and the actual physical size corresponding to a single pixel; According to the camera internal parameters, the workpiece coordinates and pixel coordinates of the target surface identification, solve the initial conversion relationship of the workpiece relative to the robot coordinate system; Based on the initial conversion relationship and the workpiece coordinates of the key points, calculate the photographing pose of the key points in the robot coordinate system, and generate a camera movement instruction; drive the camera to the photographing pose for shooting, and obtain the pixel coordinates of the key points based on the pre-trained YOLO model; According to the pixel coordinates of the key points, the photographing pose and the camera internal parameters, calculate the actual coordinates of the key points relative to the robot coordinate system; Match the workpiece coordinates of the key points with the coordinates of the robot coordinate system through the singular value decomposition algorithm, and solve the accurate conversion relationship of the workpiece relative to the robot coordinate system.

[0007] Preferably, the target surface identification is nine regular circles with the same spacing and the same size arranged in a 3×3 form pre-made at the center position of the lower edge of the tire mold; The image feature template is used to identify the target surface identification, and the template includes nine regular circles with the same spacing and the same size arranged in a 3×3 form.

[0008] Preferably, the key points include the center point of the exhaust hole to be drilled and eight boundary points, and the nine key points are arranged in a 3×3 form.

[0009] Preferably, the actual physical size corresponding to a single pixel is specifically:

[0010] where D and d are the actual distance and pixel distance of any regular circle in the target surface identification, respectively.

[0011] Preferably, according to the camera internal parameters, the workpiece coordinates and pixel coordinates of the target surface identification, use the SQPnP algorithm to solve the initial conversion relationship of the workpiece relative to the robot coordinate system.

[0012] Preferably, the calculation of the photographing pose of the key points in the robot coordinate system based on the initial conversion relationship and the workpiece coordinates of the key points specifically includes: Based on the initial conversion relationship and the workpiece coordinates of the key points, calculate the initial position of the key points in the robot coordinate system; Move the initial position backward along the normal vector of the key point by a distance to obtain the photographing pose; where is the camera focal length, is the x coordinate of a single pixel, is the actual physical size corresponding to a single pixel.

[0013] Preferably, calculating the actual coordinates of the key points relative to the robot coordinate system according to the pixel coordinates of the key points, the photographing pose, and the camera internal parameters specifically includes: ,

[0014] where is the backward distance of the photographing pose, and the actual coordinates of the key points relative to the robot coordinate system are , the photographing pose is ), is the pixel coordinate of the key point; ( ) is the image center coordinate, and is the camera internal parameter; is the actual physical size corresponding to a single pixel.

[0015] In a second aspect, the present invention provides a vision positioning system for an automatic drilling robot of a tire mold, including: a calibration module for calibrating the camera internal parameters and external parameters, making an image feature template, and training a YOLO model; a workpiece coordinate extraction module for extracting the workpiece coordinates of the target surface markings and key points on the tire mold; ]>a template matching module for identifying the target surface of the tire mold based on the image feature template, and obtaining the pixel coordinates of the markings and the actual physical size corresponding to a single pixel; an initial solution module for solving the initial conversion relationship of the workpiece relative to the robot coordinate system according to the camera internal parameters, the workpiece coordinates of the target surface markings, and the pixel coordinates; a dynamic positioning module for calculating the photographing pose of the key points in the robot coordinate system based on the initial conversion relationship and the workpiece coordinates of the key points, generating a camera movement instruction; driving the camera to the photographing pose for shooting, and obtaining the pixel coordinates of the key points based on the pre-trained YOLO model; a coordinate calculation module for calculating the actual coordinates of the key points relative to the robot coordinate system according to the pixel coordinates of the key points, the photographing pose, and the camera internal parameters; an accurate solution module for matching the workpiece coordinates of the key points with the coordinates of the robot coordinate system through the singular value decomposition algorithm, and solving the accurate conversion relationship of the workpiece relative to the robot coordinate system.

[0016] In a third aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the steps in the vision positioning method for an automatic drilling robot of a tire mold described in the first aspect.

[0017] Fourthly, the present invention provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the steps in the vision positioning method for an automatic drilling robot of a tire mold described in the first aspect.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: By calibrating the camera, making templates, and training the YOLO model, the present invention provides a basis for accurate positioning. Among them, the positioning of the exhaust holes of the tire mold by a monocular camera reduces costs; the recognition of key points by the YOLO deep learning model effectively eliminates the influence of noise and improves the recognition accuracy. Further, by extracting the target surface identification and the coordinates of key points, and combining the image feature template to recognize the target surface, key information can be quickly obtained. Solving the initial transformation relationship, then calculating the photographing pose and obtaining the pixel coordinates of key points, and further obtaining the actual coordinates. Finally, the accurate transformation relationship is obtained through the singular value decomposition algorithm, realizing high-precision positioning and improving the positioning efficiency.

[0019] Compared with manual drilling, the method eliminates the randomness of manual operation, ensures processing consistency, and also avoids health risks. Compared with the contact probe and 3D scanner positioning methods, it does not need to contact the mold, has a fast measurement speed, low cost, small calculation amount, low requirement for hardware performance, and is less affected by noise, effectively solving the deficiencies of the existing positioning methods, providing efficient and accurate positioning support for the automatic drilling of tire molds, and promoting the development of the tire mold processing industry towards automation and intelligence.

[0020] The advantages of the additional aspects of the present invention will be partially given in the following description, partially become obvious from the following description, or be understood through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The specification drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute a limitation to the present invention.

[0022] Figure 1 It is the main flowchart of a vision positioning method for an automatic drilling robot of a tire mold provided by an embodiment of the present invention; Figure 2 It is a schematic diagram of a tire mold provided by an embodiment of the present invention; Figure 3 It is a schematic diagram of the target surface of a tire mold provided by an embodiment of the present invention; Figure 4 It is a schematic diagram of recognizing the target surface of a tire mold based on an image feature template provided by an embodiment of the present invention; Figure 5The actual coordinates of the key points relative to the robot coordinate system provided by the embodiments of the present invention Schematic diagram for calculation. Detailed implementation manners

[0023] The present invention will be further described below in conjunction with the drawings and embodiments.

[0024] Embodiment 1 As Figure 1 shown, this embodiment discloses a vision positioning method for an automatic drilling robot of a tire mold, including the following steps: S1: Calibrate the internal and external parameters of the camera, make an image feature template and train the YOLO model; S2: Extract the workpiece coordinates of the target surface marks and key points on the tire mold; S3: Based on the image feature template, identify the target surface of the tire mold, and obtain the pixel coordinates of the marks and the actual physical size corresponding to a single pixel; S4: According to the internal parameters of the camera, the workpiece coordinates and pixel coordinates of the target surface marks, solve the initial conversion relationship of the workpiece relative to the robot coordinate system; S5: Based on the initial conversion relationship and the workpiece coordinates of the key points, calculate the photographing pose of the key points in the robot coordinate system, generate a camera movement instruction; drive the camera to the photographing pose for shooting, and obtain the pixel coordinates of the key points based on the pre-trained YOLO model; S6: According to the pixel coordinates of the key points, the photographing pose and the internal parameters of the camera, calculate the actual coordinates of the key points relative to the robot coordinate system; S7: Match the workpiece coordinates of the key points with the coordinates of the robot coordinate system through the singular value decomposition algorithm, and solve the accurate conversion relationship of the workpiece relative to the robot coordinate system.

[0025] Next, in combination with Figure 1 , a vision positioning method for an automatic drilling robot of a tire mold disclosed in this embodiment will be described in detail.

[0026] S1. Calibrate the internal and external parameters of the camera, make an image feature template and train the YOLO model.

[0027] Specifically, the camera used is a monocular camera, and the internal and external parameters of the camera are calibrated by the Zhang Zhengyou calibration method. The calibration results include the camera focal length F , the size of a single pixel( ), the image center coordinates( ) and the conversion relationship from the camera to the end flange of the robot and other parameters.

[0028] Furthermore, make an image feature template and train the YOLO model.

[0029] The image feature template is made based on the shape of the target surface coordinates and is used to identify the target surface at the center of the lower edge of the tire. The target surface is machined by a numerically controlled machine tool, and has high precision, making it suitable for template matching with low cost and high efficiency.

[0030] The YOLO model is used to identify key points. Its training includes image acquisition of the exhaust holes of the tire mold, annotation of the exhaust holes, and training of the YOLO model weights. Among them, the image acquisition of the exhaust holes of the tire mold is completed by taking pictures of the exhaust holes of different models and different regions with a calibrated camera. The annotation of the exhaust holes is completed by Labelme software. In this embodiment, the YOLOv8l-seg model is preferably used for the YOLO model. Since the machining positions of the exhaust holes are cast and have irregular shapes, a more suitable YOLO deep learning model is selected.

[0031] In this embodiment, the internal and external parameters of the camera are calibrated by the Zhang Zhengyou calibration method, which can accurately obtain the parameters of the camera and provide accurate basic data for subsequent image-based measurement and positioning. The made image feature template can quickly identify the target surface of the tire mold, improving the efficiency and accuracy of target surface recognition. Training the YOLO model targets the irregular shape of the exhaust holes, effectively identifying key points, reducing recognition errors caused by complex shapes, laying a foundation for accurately determining the positions of the exhaust holes, and overall improving the performance and reliability of the visual positioning system.

[0032] S2. Extract the workpiece coordinates of the target surface markings and key points on the tire mold 。

[0033] As Figure 2 shown, the tire mold includes a target surface, exhaust holes, and key points located at the center of the lower edge. The target surface marking coordinates and key point coordinates , including the coordinates of nine circles with circular features on the target surface, and the coordinates of nine key points composed of the center point and eight boundary points of the exhaust holes to be drilled, arranged in a 3×3 form. Among them, the extraction of the coordinates is carried out by CAM software.

[0034] As Figure 3 shown, the target surface is composed of nine circles in a 3×3 form. In this embodiment, the horizontal interval is 12.5 mm and the vertical interval is 10 mm.

[0035] In this embodiment, the workpiece coordinates of the target surface identification and key points on the tire mold are accurately extracted, providing key data support for calculating the position and attitude of the mold in the robot coordinate system. These coordinate information are the basis for establishing the motion relationship between the mold and the robot. By obtaining accurate coordinates, the robot can operate on the mold more precisely. Moreover, the coordinate point distribution in the form of 3×3 facilitates subsequent calculations and analyses, improving the accuracy and stability of positioning, and ensuring that the robot can quickly find the target position during the automatic drilling process.

[0036] In S3, based on the image feature template, the target surface of the tire mold is identified, and the pixel coordinates of the identification and the actual physical size corresponding to a single pixel are obtained.

[0037] The shape of the target surface is as Figure 3 shown, and the acquisition of the target surface image is realized by taking pictures with a camera.

[0038] The image feature template matching is performed by a shape-based template matching method. After template matching, the pixel coordinates of nine circular identifications and the actual physical size corresponding to a single pixel are obtained.

[0039] Specifically, as Figure 4 shown, based on the image feature template, the target surface of the tire mold is identified. After matching the circles, the pixel coordinates of the centers of the circles are obtained. Arbitrarily select two identified circles, and . The pixel distance between the two circles , the actual distance between the two circles , then :

[0040] where D and d are the actual distance and pixel distance of any positive circle in the target surface identification respectively.

[0041] In this embodiment, by identifying the target surface based on the image feature template, the position of the target surface can be quickly and accurately found, and the pixel coordinates of the identification are obtained. Calculating the actual physical size corresponding to a single pixel establishes the connection between the image pixels and the actual physical size, and converts the image information into a physical quantity with practical significance. This enables the subsequent accurate calculation of the position and size of the target surface in the actual space, provides key data for solving the conversion relationship between the workpiece and the robot coordinate system, greatly improves the positioning accuracy, and reduces the influence of positioning errors on the subsequent drilling operation.

[0042] In S4, according to the internal parameters of the camera, the workpiece coordinates and pixel coordinates of the target surface identification, the initial conversion relationship of the workpiece relative to the robot coordinate system 1; Specifically, the robot is an industrial six-degree-of-freedom robot; the solution method is a pose estimation algorithm including PnP, DLT, EPnP, SQPnP. Preferably, the SQPnP algorithm is used for solution.

[0043] It should be understood that the process of using the algorithm for solution can be achieved by those skilled in the art.

[0044] In S5, based on the initial transformation relationship and the key-point workpiece coordinates, calculate the photographing pose of the key points in the robot coordinate system, and generate a camera movement instruction; drive the camera to the photographing pose for shooting, and obtain the pixel coordinates of the key points based on the pre-trained YOLO model. Specifically, it includes: S501: Calculate the initial position of the key points in the robot coordinate system based on the initial transformation relationship and the key-point workpiece coordinates ) S502: Move backward along the normal vector of the key point ( by the distance to obtain the photographing pose , and generate a camera movement instruction; S503: Based on the camera movement instruction, drive the camera to the photographing pose for shooting, and obtain the pixel coordinates of the key points based on the pre-trained YOLO model.

[0045] In S501, first, the key points are selected to conform to the spatial analysis principle, and nine key points are selected at the edge and center of the entire exhaust hole area.

[0046] Specifically, the calculation method is as follows: 1

[0047] Among them, is the transformation relationship from the robot end flange to the robot coordinate system (known when the robot leaves the factory), is the transformation relationship from the camera to the end flange (obtained by camera calibration), is the transformation relationship from the object to the camera coordinate system; is the initial transformation relationship solved in S4 1; is the key-point workpiece coordinate.

[0048] Furthermore, in S502, calculate the photographing pose ) Specifically, the calculation method of the photographing pose is to move backward along the normal vector of the exhaust hole by the distance.

[0049] A suitable photographing pose enables the camera to clearly capture key points, avoid image blurring or information loss caused by improper perspectives, and ensure accurate recognition by the YOLO model. In this embodiment, the pose is determined by precisely calculating the backward movement distance and other methods, which can reduce the coordinate conversion error, improve the accuracy of the conversion from pixel coordinates to actual physical coordinates, and further provide reliable data for solving the precise conversion relationship between the workpiece and the robot coordinate system, ensuring the precise positioning of the exhaust hole by the robot and improving the accuracy and consistency of drilling.

[0050] Further, in S503, the robot drives the camera to , and takes pictures and recognizes each key point P n one by one; Specifically, the key point recognition is performed by the YOLO model, and the recognition result outputs the pixel coordinates of the key points .

[0051] It should be understood that only one key point can be recognized in step S5 at a time. Therefore, it needs to be executed 9 times to obtain the pixel coordinates of all key points.

[0052] In S6, calculate the coordinates of the key points in the robot coordinate system ); Specifically, combined with Figure 5 , the calculation method of the coordinates of the key points in the robot coordinate system is: , where ; In this embodiment, the actual coordinates are calculated based on the key point pixel coordinates, the photographing pose, and the camera internal parameters, and the pixel information in the image is converted into the actual space coordinates in the robot coordinate system. Through the precise coordinate conversion formula and making full use of the various parameters obtained previously, the accurate mapping from the image space to the actual physical space is achieved. The actual coordinates of the key points obtained in this step can accurately reflect the position of the exhaust hole in the robot working space, providing key data for finally determining the precise conversion relationship between the mold and the robot coordinate system, and ensuring the accuracy and reliability of the positioning.

[0053] In S7, calculate the precise conversion relationship between the workpiece and the robot coordinate system through the key points 2; Specifically, the precise conversion relationship [[ID=4I]]2 forms point pairs by the key point workpiece coordinates P n and their corresponding points , and is solved by the singular value decomposition algorithm. This process can be achieved by those skilled in the art.

[0054] In this embodiment, the singular value decomposition algorithm is used to match the key point coordinates and solve the precise transformation relationship, which can make full use of a large amount of data obtained previously, eliminate error accumulation, and obtain a more accurate position and attitude relationship between the mold and the robot coordinate system. This precise transformation relationship enables the robot to accurately locate the position of the exhaust hole on the tire mold, greatly improving the positioning accuracy. Based on the precise transformation relationship, the robot can perform drilling operations more accurately, improving the accuracy and consistency of drilling, reducing the scrap rate, and enhancing the quality and efficiency of tire mold processing.

[0055] In view of the deficiencies of the existing positioning methods for the exhaust holes of tire molds, this specific embodiment proposes an efficient and low-cost vision positioning method. Compared with the positioning method using a contact probe, it avoids the drawbacks of contact measurement. By visually identifying and calculating to determine the workpiece pose, it does not directly contact the mold, reducing equipment wear, increasing the measurement speed, and lowering costs.

[0056] Compared with the method using a 3D scanner, this embodiment does not require a dedicated CAM software. It uses the YOLO model to screen key points, combines the image feature template and camera parameters to solve the transformation relationship, has a small amount of calculation, low requirements for hardware performance, and is relatively less affected by noise, significantly reducing the equipment cost. At the same time, this embodiment can accurately position. By performing multiple calculations and using the singular value decomposition algorithm to solve the precise transformation relationship, it ensures the accuracy of positioning, provides a reliable positioning basis for automatic drilling of tire molds, and effectively improves the overall efficiency and quality of the drilling work.

[0057] Embodiment 2 This embodiment provides a vision positioning system for an automatic drilling robot of a tire mold, including: A calibration module for calibrating the internal and external parameters of the camera, making an image feature template, and training the YOLO model; A workpiece coordinate extraction module for extracting the workpiece coordinates of the target surface identification and key points on the tire mold; A template matching module for identifying the target surface of the tire mold based on the image feature template, obtaining the pixel coordinates of the identification and the actual physical size corresponding to a single pixel; An initial solution module for solving the initial transformation relationship of the workpiece relative to the robot coordinate system according to the internal parameters of the camera, the workpiece coordinates and pixel coordinates of the target surface identification; A dynamic positioning module for calculating the photographing pose of the key points in the robot coordinate system based on the initial transformation relationship and the workpiece coordinates of the key points, generating a camera movement instruction; driving the camera to the photographing pose for shooting, and obtaining the pixel coordinates of the key points based on the pre-trained YOLO model; A coordinate calculation module for calculating the actual coordinates of the key points relative to the robot coordinate system according to the pixel coordinates of the key points, the photographing pose, and the internal parameters of the camera. An exact solution module, which is used to match the workpiece coordinates of key points with the robot coordinate system coordinates through the singular value decomposition algorithm, and solve the exact conversion relationship of the workpiece relative to the robot coordinate system.

[0058] Embodiment III This embodiment provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the steps in a visual positioning method for an automatic drilling robot of a tire mold as described in Embodiment I above.

[0059] Embodiment IV This embodiment provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the steps in a visual positioning method for an automatic drilling robot of a tire mold as described in Embodiment I above.

[0060] The steps or modules involved in Embodiments II to IV above correspond to those in Embodiment I. For specific implementation manners, reference may be made to the relevant description part of Embodiment I. The term "computer-readable storage medium" should be understood to include a single medium or multiple media containing one or more instruction sets; it should also be understood to include any medium that can store, encode, or carry an instruction set for execution by a processor and cause the processor to execute any method in the present invention.

[0061] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention may have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A vision positioning method for an automatic drilling robot of a tire mold, characterized in that Including: Calibrating the internal and external parameters of the camera, making an image feature template and training the YOLO model; Extracting the workpiece coordinates of the target surface markings and key points on the tire mold; Based on the image feature template, identifying the target surface of the tire mold, obtaining the pixel coordinates of the markings and the actual physical size corresponding to a single pixel; According to the camera internal parameters, the workpiece coordinates and pixel coordinates of the target surface markings, solving the initial transformation relationship of the workpiece relative to the robot coordinate system; Based on the initial transformation relationship and the workpiece coordinates of the key points, calculating the photographing pose of the key points in the robot coordinate system, generating a camera movement instruction; driving the camera to the photographing pose for shooting, and obtaining the pixel coordinates of the key points based on the pre-trained YOLO model; According to the pixel coordinates of the key points, the photographing pose and the camera internal parameters, calculating the actual coordinates of the key points relative to the robot coordinate system; Matching the workpiece coordinates of the key points with the coordinates in the robot coordinate system through the singular value decomposition algorithm, and solving the precise transformation relationship of the workpiece relative to the robot coordinate system.

2. The vision positioning method for an automatic drilling robot of a tire mold according to claim 1, characterized in that The target surface markings are nine equally spaced and equally sized regular circles arranged in a 3×3 form pre-made at the center position of the lower edge of the tire mold; The image feature template is used to identify the target surface markings, and the template includes nine equally spaced and equally sized regular circles arranged in a 3×3 form.

3. The vision positioning method for an automatic drilling robot of a tire mold according to claim 1, characterized in that, The key points include the center point of the exhaust hole to be drilled and eight boundary points, and the nine key points are arranged in a 3×3 form.

4. The vision positioning method for an automatic drilling robot of a tire mold according to claim 2, characterized in that, The actual physical size corresponding to a single pixel is specifically: Where D and d are the actual distance and pixel distance of any regular circle in the target surface markings respectively.

5. The vision positioning method for an automatic drilling robot of a tire mold according to claim 1, characterized in that, According to the camera internal parameters, the workpiece coordinates and pixel coordinates of the target surface markings, using the SQPnP algorithm to solve the initial transformation relationship of the workpiece relative to the robot coordinate system.

6. The vision positioning method for an automatic drilling robot of a tire mold according to claim 1, characterized in that, The calculating the photographing pose of the key points in the robot coordinate system based on the initial transformation relationship and the workpiece coordinates of the key points specifically includes: Calculating the initial position of the key points in the robot coordinate system based on the initial transformation relationship and the workpiece coordinates of the key points; Shift the initial position backward along the normal vector of the key point by a distance to obtain the photographing pose; where is the camera focal length, is the x coordinate of a single pixel, is the actual physical size corresponding to a single pixel.

7. The vision positioning method for an automatic drilling robot of a tire mold according to claim 1, characterized in that, The calculating the actual coordinates of the key points relative to the robot coordinate system according to the pixel coordinates of the key points, the photographing pose and the camera internal parameters specifically includes: ， Among them, is the distance of the post-shift of the photographing pose, and the actual coordinates of the key point relative to the robot coordinate system are , and the photographing pose is ); is the pixel coordinate of the key point; ( ) is the image center coordinate, which is the camera internal parameter; is the actual physical size corresponding to a single pixel.

8. A vision positioning system for an automatic drilling robot of a tire mold, characterized in that, Including: A calibration module for calibrating the internal and external parameters of the camera, making an image feature template and training the YOLO model; A workpiece coordinate extraction module for extracting the workpiece coordinates of the target surface markings and key points on the tire mold; A template matching module for identifying the target surface of the tire mold based on the image feature template, obtaining the pixel coordinates of the markings and the actual physical size corresponding to a single pixel; An initial solution module for solving the initial transformation relationship of the workpiece relative to the robot coordinate system according to the camera internal parameters, the workpiece coordinates and pixel coordinates of the target surface markings; A dynamic positioning module for calculating the photographing pose of the key points in the robot coordinate system based on the initial transformation relationship and the workpiece coordinates of the key points, generating a camera movement instruction; driving the camera to the photographing pose for shooting, and obtaining the pixel coordinates of the key points based on the pre-trained YOLO model; A coordinate calculation module for calculating the actual coordinates of the key points relative to the robot coordinate system according to the pixel coordinates of the key points, the photographing pose and the camera internal parameters; An exact solution module is used to match the workpiece coordinates of key points with the coordinates of the robot coordinate system through the singular value decomposition algorithm, and solve the exact conversion relationship of the workpiece relative to the robot coordinate system.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by a processor, it implements the steps in a visual positioning method for an automatic drilling robot for tire molds as described in any one of claims 1-7.

10. A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in a visual positioning method for an automatic drilling robot for tire molds as described in any one of claims 1-7.

Images