Visual calibration method and system for Delta structure parallel 3D printer and medium

Abstract

The invention discloses a visual calibration method and system for a Delta structure parallel 3D printer and a medium, and the method comprises the steps that data collection is carried out based on preset sampling point coordinates to obtain an observation data set, and the observation data set comprises guide rail encoder readings and calibration plate coordinates under a camera coordinate system; coordinate transformation is carried out on each observation point in the observation data set to obtain a coordinate list of the center point of the movable disc under world coordinates, and coordinate transformation comprises calibration plate correction and camera offset correction; and performing coordinate calculation based on the coordinate list to obtain a movable disc coordinate of the movable disc control point in the world coordinate system and a guide rail coordinate of the guide rail control point in the world coordinate system. According to the method, high-precision joint identification and compensation of the Delta parallel 3D printer multi-source errors are achieved through the self-defined 22-parameter full-link error model in combination with the nonlinear optimization and stop stable sampling process, and the high-speed and high-precision printing requirements are effectively met.

Description

Technical Field

[0001] This invention relates to the field of vision processing, and more specifically, to a vision calibration method, system, and medium for a Delta structure parallel 3D printer. Background Technology

[0002] The high-precision printing of Delta-structure parallel 3D printers relies heavily on the accurate measurement and control of the end effector (nozzle) position. Traditional calibration techniques primarily focus on compensating for some mechanical geometric errors (such as rod length and base radius), calibrating through simplified models and limited parameters. However, in practical industrial applications, the sources of printer errors are complex and diverse, exceeding the capabilities of traditional methods.

[0003] The existing technology has the following drawbacks:

[0004] 1. Incomplete coverage of error sources: Traditional solutions only focus on some mechanical errors and fail to systematically incorporate and jointly compensate for vision system errors (such as camera installation offset and calibration plate placement deviation) and sensor residual errors, resulting in the accumulation of multiple error sources and making it difficult for the overall positioning accuracy to break through "±0.05mm".

[0005] 2. The mathematical model is oversimplified, and the parameter model has a low dimension. It ignores key factors such as the rotational deviation of the guide rail around the Z-axis and the spatial pose error of the calibration plate, resulting in a significant gap between the mathematical model and physical reality.

[0006] 3. The identification algorithm is not stable enough. Some of the optimization algorithms used in the scheme (such as gradient descent) are prone to getting trapped in local optima. Furthermore, the physical constraints of the parameters are not clearly defined, resulting in poor algorithm robustness and low practical value in engineering.

[0007] 4. The data acquisition process was not standardized. The standard procedure of "stopping and taking a picture" was not strictly followed when collecting data. The observed data had a lot of noise due to the vibration of the mechanism, which seriously affected the final accuracy of parameter identification. Summary of the Invention

[0008] The purpose of this invention is to provide a visual calibration method, system, and medium for Delta parallel 3D printers, which solves the technical problems of low positioning accuracy and inability to compensate for the accumulation of multi-source errors caused by incomplete error models and neglect of visual and sensor errors in existing Delta parallel 3D printer calibration methods.

[0009] The first aspect of this invention provides a visual calibration method for a Delta structure parallel 3D printer, comprising the following steps:

[0010] The observation dataset is obtained by collecting data based on the preset sampling point coordinates. The observation dataset includes the guide rail encoder readings and the calibration plate coordinates in the camera coordinate system.

[0011] A coordinate transformation is performed on each observation point in the observation dataset to obtain a list of coordinates of the center point of the moving disk in world coordinates. The coordinate transformation includes calibration plate correction and camera offset correction.

[0012] Based on the coordinate list, coordinate calculations are performed to obtain the coordinates of the moving disk control point in the world coordinate system and the coordinates of the guide rail control point in the world coordinate system.

[0013] Based on the coordinates of the moving disk and the coordinates of the guide rail, combined with the preset link length, a residual objective function is constructed to obtain the objective function value;

[0014] Based on the objective function value and the preset error parameters, a nonlinear optimization iterative solution is performed to obtain the objective parameter vector, and the objective parameter vector is injected into the kinematic model of the 3D printer to complete the parameter configuration.

[0015] In this scheme, the process of acquiring the observation dataset based on preset sampling point coordinates specifically includes:

[0016] The sampling points correspond to the planned three-dimensional mesh sampling points within the Delta workspace, covering the range x∈[-140,140]mm, y∈[-140,140]mm, z∈[-350,-5]mm, wherein the sampling points are not collinear;

[0017] The nozzle is controlled to move to each of the sampling points, and after stopping and waiting for a preset time, the encoder is read and the calibration plate image is captured.

[0018] The calibration plate coordinates are obtained by extracting the image pixel coordinates of the feature points of the calibration plate from the calibration plate image and correcting them. At the same time, the observation dataset is obtained by extracting the guide rail encoder readings after reading the encoder.

[0019] In this scheme, the step of performing coordinate transformation on each observation point in the observation dataset to obtain a list of coordinates of the center point of the moving disk in world coordinates specifically includes:

[0020] The formula for correcting calibration plate rotation and center offset is as follows:

[0021] ;

[0022] ;

[0023] Among them, (x obs , y obs(x) represents the coordinates of the calibration plate being observed. bd , y bd () represents the corrected coordinates of the calibration plate. For the calibration plate rotation matrix, This is the calibration plate rotation correction value. , This is the correction value for the center offset of the calibration plate;

[0024] Camera offset correction is calculated as follows:

[0025] x p =x bd -c x , y p =y bd -c y , z p =z obs -c z ;

[0026] Where (cx, cy, cz) is the camera optical center offset, (x p ,y p ,z p (x) represents the coordinates of the center point of the moving disk in world coordinates, and the coordinate list is (x) p ,y p ,z p )_k, where k is the sampling point number.

[0027] In this solution, the coordinate calculation based on the coordinate list specifically includes:

[0028] The coordinates of the moving disk control points in the world coordinate system are calculated using the following formula:

[0029] ;

[0030] in, , , For the control point of the moving disk, Let x be the radius of the moving disk. p ,y p ,z p () represents the coordinates of the center point of the moving disk in world coordinates;

[0031] The coordinates of the guide rail control points in the world coordinate system are calculated using the following formula:

[0032] ;

[0033] ;

[0034] in, For encoder reading conversion distance, The inverse of the rotation matrix, For the translation of the guide rail base, , , For base translation at different guide rail control points, Let be the radius of the guide rail distribution circle.

[0035] In this scheme, the step of constructing a residual objective function based on the coordinates of the moving disk and the guide rail, combined with a preset link length, to obtain the objective function value specifically includes:

[0036] Obtain the length of the connecting rod, wherein the connecting rod is used to connect the control point of the moving disk and the control point of the guide rail;

[0037] The formula for calculating the residual vector of a single sampling point is as follows:

[0038] ;

[0039] in, For a single sampling point, For the Euclidean norm, , , For the control point of the moving disk, , , Here, k represents the guide rail control point, and k is the sampling point number. , and The length of the link;

[0040] The overall objective function value for the entire observation dataset is calculated using the following formula:

[0041] ;

[0042] Wherein, the total objective function value This corresponds to the sum of squares of the residuals of all sampling points, where P is the preset parameter vector, N is the total number of sampling points, and k is the sampling point index. This is the residual vector for a single sampling point.

[0043] In this scheme, the target parameter vector is obtained by iteratively optimizing the preset parameter vector based on the preset algorithm. The target parameter vector is then written into the inverse kinematics calculation module of the 3D printer to replace the original preset parameter vector, thereby completing the parameter configuration. The target parameter vector includes 22 parameter values.

[0044] A second aspect of the present invention also provides a visual calibration system for a Delta structure parallel 3D printer, comprising a memory and a processor. The memory includes a visual calibration method program for the Delta structure parallel 3D printer, which, when executed by the processor, performs the following steps:

[0045] The observation dataset is obtained by collecting data based on the preset sampling point coordinates. The observation dataset includes the guide rail encoder readings and the calibration plate coordinates in the camera coordinate system.

[0046] A coordinate transformation is performed on each observation point in the observation dataset to obtain a list of coordinates of the center point of the moving disk in world coordinates. The coordinate transformation includes calibration plate correction and camera offset correction.

[0047] Based on the coordinate list, coordinate calculations are performed to obtain the coordinates of the moving disk control point in the world coordinate system and the coordinates of the guide rail control point in the world coordinate system.

[0048] Based on the coordinates of the moving disk and the coordinates of the guide rail, combined with the preset link length, a residual objective function is constructed to obtain the objective function value;

[0049] Based on the objective function value and the preset error parameters, a nonlinear optimization iterative solution is performed to obtain the objective parameter vector, and the objective parameter vector is injected into the kinematic model of the 3D printer to complete the parameter configuration.

[0050] In this scheme, the process of acquiring the observation dataset based on preset sampling point coordinates specifically includes:

[0051] The sampling points correspond to the planned three-dimensional mesh sampling points within the Delta workspace, covering the range x∈[-140,140]mm, y∈[-140,140]mm, z∈[-350,-5]mm, wherein the sampling points are not collinear;

[0052] The nozzle is controlled to move to each of the sampling points, and after stopping and waiting for a preset time, the encoder is read and the calibration plate image is captured.

[0053] The calibration plate coordinates are obtained by extracting the image pixel coordinates of the feature points of the calibration plate from the calibration plate image and correcting them. At the same time, the observation dataset is obtained by extracting the guide rail encoder readings after reading the encoder.

[0054] In this scheme, the step of performing coordinate transformation on each observation point in the observation dataset to obtain a list of coordinates of the center point of the moving disk in world coordinates specifically includes:

[0055] The formula for correcting calibration plate rotation and center offset is as follows:

[0056] ;

[0057] ;

[0058] Among them, (x obs , y obs (x) represents the coordinates of the calibration plate being observed. bd , y bd () represents the corrected coordinates of the calibration plate. For the calibration plate rotation matrix, This is the calibration plate rotation correction value. , This is the correction value for the center offset of the calibration plate;

[0059] Camera offset correction is calculated as follows:

[0060] x p =x bd -c x , y p =y bd -c y , z p =z obs -c z ;

[0061] Where (cx, cy, cz) is the camera optical center offset, (x p ,y p ,z p (x) represents the coordinates of the center point of the moving disk in world coordinates, and the coordinate list is (x) p ,y p ,z p )_k, where k is the sampling point number.

[0062] In this solution, the coordinate calculation based on the coordinate list specifically includes:

[0063] The coordinates of the moving disk control points in the world coordinate system are calculated using the following formula:

[0064] ;

[0065] in, , , For the control point of the moving disk, Let x be the radius of the moving disk. p ,y p ,z p () represents the coordinates of the center point of the moving disk in world coordinates;

[0066] The coordinates of the guide rail control points in the world coordinate system are calculated using the following formula:

[0067] ;

[0068] ;

[0069] in, For encoder reading conversion distance, The inverse of the rotation matrix, For the translation of the guide rail base, , , For base translation at different guide rail control points, Let be the radius of the guide rail distribution circle.

[0070] In this scheme, the step of constructing a residual objective function based on the coordinates of the moving disk and the guide rail, combined with a preset link length, to obtain the objective function value specifically includes:

[0071] Obtain the length of the connecting rod, wherein the connecting rod is used to connect the control point of the moving disk and the control point of the guide rail;

[0072] The formula for calculating the residual vector of a single sampling point is as follows:

[0073] ;

[0074] in, For a single sampling point, For the Euclidean norm, , , For the control point of the moving disk, , , Here, k represents the guide rail control point, and k is the sampling point number. , and The length of the link;

[0075] The overall objective function value for the entire observation dataset is calculated using the following formula:

[0076] ;

[0077] Wherein, the total objective function value This corresponds to the sum of squares of the residuals of all sampling points, where P is the preset parameter vector, N is the total number of sampling points, and k is the sampling point index. This is the residual vector for a single sampling point.

[0078] In this scheme, the target parameter vector is obtained by iteratively optimizing the preset parameter vector based on the preset algorithm. The target parameter vector is then written into the inverse kinematics calculation module of the 3D printer to replace the original preset parameter vector, thereby completing the parameter configuration. The target parameter vector includes 22 parameter values.

[0079] A third aspect of the present invention provides a computer-readable storage medium comprising a machine program for a visual calibration method for a Delta structure parallel 3D printer, wherein when executed by a processor, the visual calibration method program for a Delta structure parallel 3D printer implements the steps of a visual calibration method for a Delta structure parallel 3D printer as described in any of the preceding claims.

[0080] This invention discloses a visual calibration method, system, and medium for Delta-structure parallel 3D printers. By constructing a 22-parameter end-to-end error model encompassing mechanical, visual, and sensor parameters, and combining nonlinear optimization and stabilization sampling processes, it achieves high-precision joint identification and compensation of multi-source errors in Delta parallel 3D printers, improving end-effector positioning accuracy to within ±0.03mm. This effectively supports the demands of high-speed, high-precision printing. Specific beneficial effects are as follows:

[0081] 1. Improved parameter comprehensiveness: 22 parameters cover the entire error chain of mechanics, vision, and sensors, solving the problem of "error omission" in traditional solutions;

[0082] 2. Improved mathematical rigor: Based on Euler angle rotation matrix and coordinate transformation theory, the residual model strictly satisfies the link length constraint, without any simplified approximations;

[0083] 3. Improved engineering practicality, fast algorithm convergence speed (<200 iterations), parameter constraints that meet industrial scenarios, and can be directly integrated into the 3D printer control system;

[0084] 4. Accuracy is guaranteed. By stabilizing the sampling process, data noise is reduced, the parameter identification deviation is small, and the actual printing accuracy reaches "±0.03mm". Attached Figure Description

[0085] Figure 1 A flowchart of a visual calibration method for a Delta structure parallel 3D printer according to the present invention is shown;

[0086] Figure 2 This invention illustrates a structural schematic diagram of a Delta structure parallel 3D printer.

[0087] Figure 3The diagram illustrates a Cartesian coordinate system for defining the Delta structure in a visual calibration method for a Delta structure parallel 3D printer according to the present invention.

[0088] Figure 4 A detailed diagram of coordinate system parameters for a visual calibration method for a Delta structure parallel 3D printer according to the present invention is shown.

[0089] Figure 5 A block diagram of a vision calibration system for a Delta structure parallel 3D printer according to the present invention is shown. Detailed Implementation

[0090] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0091] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0092] Figure 1 A flowchart of a visual calibration method for a Delta structure parallel 3D printer according to this application is shown.

[0093] like Figure 1 As shown, this application discloses a visual calibration method for a Delta structure parallel 3D printer, including the following steps:

[0094] S102, data acquisition is performed based on preset sampling point coordinates to obtain an observation dataset, wherein the observation dataset includes guide rail encoder readings and calibration plate coordinates in the camera coordinate system;

[0095] S104, Perform coordinate transformation on each observation point in the observation dataset to obtain a list of coordinates of the center point of the moving disk in world coordinates, wherein the coordinate transformation includes calibration plate correction and camera offset correction.

[0096] S106, Based on the coordinate list, perform coordinate calculation to obtain the coordinates of the moving disk control point in the world coordinate system and the coordinates of the guide rail control point in the world coordinate system.

[0097] S108, Based on the coordinates of the moving disk and the coordinates of the guide rail, combined with the preset link length, a residual objective function is constructed to obtain the objective function value;

[0098] S110, based on the objective function value and the preset error parameters, a nonlinear optimization iterative solution is performed to obtain the objective parameter vector, and the objective parameter vector is injected into the kinematic model of the 3D printer to complete the parameter configuration.

[0099] It should be noted that, in this embodiment, the present invention relates to the field of 3D printing and machine vision interdisciplinary technology, specifically to high-precision visual calibration and multi-source error calibration of Delta structure parallel 3D printers, proposing a technical solution based on joint identification of "22" parameters, wherein, for example... Figure 2 As shown, this is a schematic diagram of a Delta-structure parallel 3D printer. The preset error parameter details of the "22" parameters described in this invention include "3" error parameters used in the first calibration and "19" error parameters used in the second calibration. Table 1 shows the details of the "3" error parameters, and Table 2 shows the details of the "19" error parameters. Constructing a complete model containing "22" error parameters covers the entire chain of mechanical, visual, and sensor errors, enabling joint identification of multi-source errors. Furthermore, as shown... Figure 3 As shown, the Delta structure defines a Cartesian coordinate system (X-axis forward, Y-axis right, Z-axis down). Further, as... Figure 4 As shown, the diagram displays a detailed list of coordinate system parameters. The origin of the coordinate system is located at the center of the base. The guide rails are distributed on a circle with radius R in the XY plane (B1 is along the positive direction of the X-axis, B2 and B3 are at ±120° to the X-axis, and B1, B2 and B3 are the guide rail control points). The moving disk is parallel to the XY plane with radius r. A1, A2 and A3 are the control points for the three equal divisions of the circumference of the moving disk. Points A and B are connected by a link of length L1 / L2 / L3. This invention integrates mechanical errors, vision system errors and sensor errors to achieve a positioning accuracy of "≤±0.03mm" for the end effector (printer), making it suitable for high-speed printing scenarios.

[0100] Table 1. Detailed list of error parameters for "3"

[0101]

[0102] Table 2. Detailed list of error parameters for "19"

[0103]

[0104] According to an embodiment of the present invention, the step of obtaining an observation dataset by collecting data based on preset sampling point coordinates specifically includes:

[0105] The sampling points correspond to the planned three-dimensional mesh sampling points within the Delta workspace, covering the range x∈[-140,140]mm, y∈[-140,140]mm, z∈[-350,-5]mm, wherein the sampling points are not collinear;

[0106] The nozzle is controlled to move to each of the sampling points, and after stopping and waiting for a preset time, the encoder is read and the calibration plate image is captured.

[0107] The calibration plate coordinates are obtained by extracting the image pixel coordinates of the feature points of the calibration plate from the calibration plate image and correcting them. At the same time, the observation dataset is obtained by extracting the guide rail encoder readings after reading the encoder.

[0108] It should be noted that, in this embodiment, a three-dimensional mesh sampling point is planned within the Delta workspace to cover the effective printing area, where x∈[-140,140] mm, y∈[-140,140] mm, and z∈[-350,-5] mm. Specifically, an "8×8×8" mesh is used to generate "512" non-collinear points. The steps for stopping the sampling operation are as follows:

[0109] ① Drive the nozzle to move to the target sampling point at the XY coordinates, and descend along the Z-axis to the preset height;

[0110] ② Stop the closed-loop stepper motor and wait 50ms to ensure the mechanism has come to a complete stop (position jitter ≤ 0.001mm).

[0111] ③ Read the encoder readings b1 / b2 / b3 at three points B, and take the average of "10" readings to reduce noise;

[0112] ④ Trigger the camera to capture an image of the calibration board and extract the observation coordinates (x, y) of the feature points. obs , y obs , z obs );

[0113] ⑤ Verify data: If If the deviation from the theoretical length is >0.1mm, re-acquire the data at that point;

[0114] Among these, standardizing the "stationary sampling" process can reduce noise in the observation data, ensure the reliability of the identification results, and ultimately form the observation dataset: {(b1,b2,b3, x obs , y obs , z obs ) k | k=1,2,...,N}, where k is the sampling point number and N is the total number of sampling points.

[0115] According to an embodiment of the present invention, the step of performing coordinate transformation on each observation point in the observation dataset to obtain a list of coordinates of the center point of the moving disk in world coordinates specifically includes:

[0116] The formula for correcting calibration plate rotation and center offset is as follows:

[0117] ;

[0118] ;

[0119] Among them, (x obs , y obs (x) represents the coordinates of the calibration plate being observed. bd , y bd () represents the corrected coordinates of the calibration plate. For the calibration plate rotation matrix, This is the calibration plate rotation correction value. , This is the correction value for the center offset of the calibration plate;

[0120] Camera offset correction is calculated as follows:

[0121] x p =x bd -c x , y p =y bd -c y , z p =z obs -c z ;

[0122] Where (cx, cy, cz) is the camera optical center offset, (x p ,y p ,z p (x) represents the coordinates of the center point of the moving disk in world coordinates, and the coordinate list is (x) p ,y p ,z p )_k, where k is the sampling point number.

[0123] It should be noted that, in this embodiment, the guide rail attitude error is quantized using the ZYX Euler angle rotation matrix, where R is the rotation matrix of any guide rail i. i Defined as a composite matrix of rotations about the Z-axis (ψᵢ), Y-axis (θᵢ), and X-axis (φᵢ), the formula is as follows:

[0124] ;

[0125] Among them, the rotation matrix around the X-axis (φ) ᵢ (for rotation angle)

[0126] ;

[0127] Rotation matrix around the Y-axis (θᵢ is the rotation angle):

[0128] ;

[0129] Rotation matrix around the Z-axis (ψᵢ is the rotation angle, B2: ψ2=120°+Δψ2; B3: ψ3=240°+Δψ3):

[0130] ;

[0131] Furthermore, the coordinates of the feature points on the calibration board need to be transformed to the world coordinate system of the base to eliminate the error between the camera and the calibration board. The specific steps are as follows:

[0132] The formula for correcting calibration plate rotation and center offset is as follows:

[0133] ;

[0134] ;

[0135] Among them, (x obs , y obs (x) represents the coordinates of the calibration plate being observed. Since the Z-axis is not rotated in this embodiment, the Z-axis coordinates are not displayed. Similarly, if the Z-axis is to be rotated, the following steps can be taken; the specific process will not be elaborated upon. Furthermore, (x) bd , y bd () represents the corrected coordinates of the calibration plate. For the calibration plate rotation matrix, It is its inverse matrix (the inverse of an orthogonal matrix is ​​equal to its transpose). This is the calibration plate rotation correction value. , This is the correction value for the center offset of the calibration plate;

[0136] Specifically, three positioning screws are installed at 120° intervals on the calibration plate mounting position on the printer base plate (the distribution logic is consistent with the guide rail control points B1, B2, and B3, with a machining accuracy of ≤0.01mm, and the center points of the three screws are strictly aligned with the origin of the world coordinate system). These screws are individually calibrated and fixed using visual recognition. , and The "3" parameters, as known quantities for subsequent identification, require the following explanation: When applying these parameters, the screw heads should be designed with identifiable visual features such as cross-shaped grooves / circular protrusions. The camera should be temporarily fixed above the base (to prevent movement with the rotating disc) to capture images containing the "3" positioning screws and the calibration plate. This sampling should be repeated "20 times." The visual coordinates of the "3" positioning screws in each sample should be extracted, and the screw center point coordinates (x0, y0) – the visual observation value of the world coordinate system origin – should be calculated using the least squares method. Additionally, preset feature points on the calibration plate (such as checkerboard corner points) should be extracted, and the visual coordinates (xboard, yboard) of the calibration plate center should be calculated. The bd value for each sample should also be calculated. x =xboard-x0, bdᵧ =yboard-y0, ψbd (angle between the characteristic direction of the calibration board and the direction of the screw connection), take the average of 20 sampling results as the final fixed value (bd is required) x / bdᵧ Deviation ≤ 0.03mm, ψ bd (If the deviation is ≤0.2°), it will not be used as an optimization variable in the subsequent second calibration.

[0137] Camera offset correction is calculated as follows:

[0138] x p =x bd -c x , y p =y bd -c y , z p =z obs -c z ;

[0139] Where (cx, cy, cz) is the camera optical center offset, (x p ,y p ,z p (x) represents the coordinates of the center point of the moving disk in world coordinates, and the coordinate list is (x) p ,y p ,z p )_k, where k is the sampling point number.

[0140] According to an embodiment of the present invention, the coordinate calculation based on the coordinate list specifically includes:

[0141] The coordinates of the moving disk control points in the world coordinate system are calculated using the following formula:

[0142] ;

[0143] in, , , For the control point of the moving disk, Let x be the radius of the moving disk. p ,y p ,z p () represents the coordinates of the center point of the moving disk in world coordinates;

[0144] The coordinates of the guide rail control points in the world coordinate system are calculated using the following formula:

[0145] ;

[0146] ;

[0147] in, For encoder reading conversion distance, The inverse of the rotation matrix, For the translation of the guide rail base, , , For base translation at different guide rail control points, Let be the radius of the guide rail distribution circle.

[0148] It should be noted that, in this embodiment, the coordinates of point A (the control point of the moving disk, in the same direction as the guide rail control point B) are calculated as follows:

[0149] ;

[0150] Among them, the guide rail control point B moves only along the Z-axis in the local coordinate system of the guide rail, with coordinates (0, 0, b). i + db i )(b i (The distance for encoder reading conversion), conversion to the world coordinate system requires a rotation matrix. (Inverse rotation matrix) and translation T of the guide rail base i :

[0151] ;

[0152] The guide rail base translation T i for:

[0153] ;

[0154] in, , , The base translation for different guide rail control points corresponds to the guide rail control points. , , The base was translated.

[0155] According to an embodiment of the present invention, the step of constructing a residual objective function based on the coordinates of the moving disk and the coordinates of the guide rail, combined with a preset link length, to obtain the objective function value specifically includes:

[0156] Obtain the length of the connecting rod, wherein the connecting rod is used to connect the control point of the moving disk and the control point of the guide rail;

[0157] The formula for calculating the residual vector of a single sampling point is as follows:

[0158] ;

[0159] in, For a single sampling point, For the Euclidean norm, , , For the control point of the moving disk, , , Here, k represents the guide rail control point, and k is the sampling point number. , and The length of the link;

[0160] The overall objective function value for the entire observation dataset is calculated using the following formula:

[0161] ;

[0162] Wherein, the total objective function value This corresponds to the sum of squares of the residuals of all sampling points, where P is the preset parameter vector, N is the total number of sampling points, and k is the sampling point index. This is the residual vector for a single sampling point.

[0163] It should be noted that, in this embodiment, the core objective of identification is to minimize the "deviation between the square of the actual link length and the square of the theoretical length". The residual vector of a single sampling point is defined as:

[0164] ;

[0165] in, For a single sampling point, For the Euclidean norm, , , For the control point of the moving disk, , , Here, k represents the guide rail control point, and k is the sampling point number. , and Given the link length, further, calculate the overall objective function value for the entire observation dataset. The overall objective function value is the sum of squares of the residuals at all sampling points, calculated as follows:

[0166] ;

[0167] Wherein, the total objective function value This corresponds to the sum of squares of the residuals of all sampling points, where P is the preset parameter vector, N is the total number of sampling points, and k is the sampling point index. The residual vector is a single sampling point. The preset parameter vector P is a "22" parameter vector, and N is the number of sampling points (in practical applications, it is recommended that N≥512 to ensure the degree of freedom of parameter identification). In this embodiment, a rigorous residual function is established based on Euler angle rotation matrix and coordinate transformation theory, which can ensure that the mathematical model is consistent with the kinematics of the actual mechanism.

[0168] According to an embodiment of the present invention, the target parameter vector is obtained by iteratively optimizing the preset parameter vector based on a preset algorithm. The target parameter vector is then written into the inverse kinematics calculation module of the 3D printer to replace the original preset parameter vector, thereby completing the parameter configuration. The target parameter vector includes 22 parameter values.

[0169] It should be noted that, in this embodiment, the values ​​of the "22" parameters are described in Table 1. The preset algorithm used in this embodiment includes the Levenberg-Marquardt algorithm, which is a numerical optimization algorithm widely used to solve nonlinear least squares problems. Its purpose is to find a set of parameters that minimizes the sum of squares of the differences between the model's predicted values ​​and the actual observed values. Since this embodiment applies the algorithm without changing its flow, the specific process of the algorithm will not be described in detail in this embodiment. The necessary key settings are as follows: the initial values ​​of the parameters are based on the design nominal values ​​(e.g., R=R0, L...). i =L0, attitude angle = 0°, visual offset = 0), to avoid the initial value deviating too far from the optimal solution; upper and lower bounds of parameters: set constraints based on the engineering error range (refer to Table 1) to prevent parameters from converging to physically unreasonable values; convergence criterion: function tolerance: the change in the objective function between adjacent iterations |J k+1 - J k |<1e-8; Step tolerance: Change in parameter vector Maximum number of iterations: 10,000 (the actual number of convergence iterations is usually <200).

[0170] It is worth mentioning that, in order to verify the effectiveness of the present invention, an example is given, and the actual parameters set are shown in Table 23 to simulate industrial scenario errors.

[0171] Table 3. Detailed List of Simulated Industrial Scene Settings

[0172] After optimization and convergence using the methods described in the above embodiments, the deviations between the identified values ​​and the true values ​​of each parameter are as follows:

[0173] Guide rail attitude angle deviation: <0.01°;

[0174] Geometric dimensional deviations: R / r / L deviation < 0.02mm, dbᵢ deviation < 0.01mm;

[0175] Visual parameter deviation: cx / cy / cz deviation < 0.1mm, bd X / bdᵧ deviation < 0.05mm, ψ_bd deviation < 0.05°;

[0176] Total residual sum of squares: converges to <1e-6, verifying the stability of the algorithm.

[0177] Furthermore, the identified "22" parameters were substituted into the inverse kinematics of the Delta mechanism to correct the nozzle motion trajectory. The actual printing test results were as follows: end positioning accuracy: ±0.03mm; trajectory tracking error: ≤0.05mm; and suitable printing speed: 2000mm / s (maximum acceleration 30000mm / s²), which meets the requirements of high-speed printing.

[0178] Figure 5 A block diagram of a vision calibration system for a Delta structure parallel 3D printer according to the present invention is shown.

[0179] like Figure 5 As shown, this invention discloses a visual calibration system for a Delta structure parallel 3D printer, including a memory and a processor. The memory includes a visual calibration method program for the Delta structure parallel 3D printer. When the processor executes the visual calibration method program for the Delta structure parallel 3D printer, it performs the following steps:

[0180] The observation dataset is obtained by collecting data based on the preset sampling point coordinates. The observation dataset includes the guide rail encoder readings and the calibration plate coordinates in the camera coordinate system.

[0181] A coordinate transformation is performed on each observation point in the observation dataset to obtain a list of coordinates of the center point of the moving disk in world coordinates. The coordinate transformation includes calibration plate correction and camera offset correction.

[0182] Based on the coordinate list, coordinate calculations are performed to obtain the coordinates of the moving disk control point in the world coordinate system and the coordinates of the guide rail control point in the world coordinate system.

[0183] Based on the coordinates of the moving disk and the coordinates of the guide rail, combined with the preset link length, a residual objective function is constructed to obtain the objective function value;

[0184] Based on the objective function value and the preset error parameters, a nonlinear optimization iterative solution is performed to obtain the objective parameter vector, and the objective parameter vector is injected into the kinematic model of the 3D printer to complete the parameter configuration.

[0185] It should be noted that, in this embodiment, the present invention relates to the field of 3D printing and machine vision interdisciplinary technology, specifically to high-precision visual calibration and multi-source error calibration of Delta structure parallel 3D printers, proposing a technical solution based on joint identification of "22" parameters, wherein, for example... Figure 2 As shown, this is a schematic diagram of a Delta-structure parallel 3D printer. The preset error parameter details of the "22" parameters described in this invention include "3" error parameters used in the first calibration and "19" error parameters used in the second calibration. Table 1 shows the details of the "3" error parameters, and Table 2 shows the details of the "19" error parameters. Constructing a complete model containing "22" error parameters covers the entire chain of mechanical, visual, and sensor errors, enabling joint identification of multi-source errors. Furthermore, as shown... Figure 3 As shown, the Delta structure defines a Cartesian coordinate system (X-axis forward, Y-axis right, Z-axis down). Further, as... Figure 4 As shown, the diagram displays a detailed list of coordinate system parameters. The origin of the coordinate system is located at the center of the base. The guide rails are distributed on a circle with radius R in the XY plane (B1 is along the positive direction of the X-axis, B2 and B3 are at ±120° to the X-axis, and B1, B2 and B3 are the guide rail control points). The moving disk is parallel to the XY plane with radius r. A1, A2 and A3 are the control points for the three equal divisions of the circumference of the moving disk. Points A and B are connected by a link of length L1 / L2 / L3. This invention integrates mechanical errors, vision system errors and sensor errors to achieve a positioning accuracy of "≤±0.03mm" for the end effector (printer), making it suitable for high-speed printing scenarios.

[0186] According to an embodiment of the present invention, the step of obtaining an observation dataset by collecting data based on preset sampling point coordinates specifically includes:

[0187] The sampling points correspond to the planned three-dimensional mesh sampling points within the Delta workspace, covering the range x∈[-140,140]mm, y∈[-140,140]mm, z∈[-350,-5]mm, wherein the sampling points are not collinear;

[0188] The nozzle is controlled to move to each of the sampling points, and after stopping and waiting for a preset time, the encoder is read and the calibration plate image is captured.

[0189] The calibration plate coordinates are obtained by extracting the image pixel coordinates of the feature points of the calibration plate from the calibration plate image and correcting them. At the same time, the observation dataset is obtained by extracting the guide rail encoder readings after reading the encoder.

[0190] It should be noted that, in this embodiment, a three-dimensional mesh sampling point is planned within the Delta workspace to cover the effective printing area, where x∈[-140,140] mm, y∈[-140,140] mm, and z∈[-350,-5] mm. Specifically, an "8×8×8" mesh is used to generate "512" non-collinear points. The steps for stopping the sampling operation are as follows:

[0191] ① Drive the nozzle to move to the target sampling point at the XY coordinates, and descend along the Z-axis to the preset height;

[0192] ② Stop the closed-loop stepper motor and wait 50ms to ensure the mechanism has come to a complete stop (position jitter ≤ 0.001mm).

[0193] ③ Read the encoder readings b1 / b2 / b3 at three points B, and take the average of "10" readings to reduce noise;

[0194] ④ Trigger the camera to capture an image of the calibration board and extract the observation coordinates (x, y) of the feature points. obs , y obs , z obs );

[0195] ⑤ Verify data: If If the deviation from the theoretical length is >0.1mm, re-acquire the data at that point;

[0196] Among these, standardizing the "stationary sampling" process can reduce noise in the observation data, ensure the reliability of the identification results, and ultimately form the observation dataset: {(b1,b2,b3, x obs , y obs , z obs ) k | k=1,2,...,N}, where k is the sampling point number and N is the total number of sampling points.

[0197] According to an embodiment of the present invention, the step of performing coordinate transformation on each observation point in the observation dataset to obtain a list of coordinates of the center point of the moving disk in world coordinates specifically includes:

[0198] The formula for correcting calibration plate rotation and center offset is as follows:

[0199] ;

[0200] ;

[0201] Among them, (x obs , y obs (x) represents the coordinates of the calibration plate being observed. bd , y bd () represents the corrected coordinates of the calibration plate. For the calibration plate rotation matrix, This is the calibration plate rotation correction value. , This is the correction value for the center offset of the calibration plate;

[0202] Camera offset correction is calculated as follows:

[0203] x p =x bd -c x , y p =y bd -c y , z p =z obs -c z ;

[0204] Where (cx, cy, cz) is the camera optical center offset, (x p ,y p ,z p (x) represents the coordinates of the center point of the moving disk in world coordinates, and the coordinate list is (x) p ,y p ,z p )_k, where k is the sampling point number.

[0205] It should be noted that, in this embodiment, the guide rail attitude error is quantized using the ZYX Euler angle rotation matrix, where R is the rotation matrix of any guide rail i. i Defined as a composite matrix of rotations about the Z-axis (ψᵢ), Y-axis (θᵢ), and X-axis (φᵢ), the formula is as follows:

[0206] ;

[0207] Among them, the rotation matrix around the X-axis (φ) ᵢ (for rotation angle)

[0208] ;

[0209] Rotation matrix around the Y-axis (θᵢ is the rotation angle):

[0210] ;

[0211] Rotation matrix around the Z-axis (ψᵢ is the rotation angle, B2: ψ2=120°+Δψ2; B3: ψ3=240°+Δψ3):

[0212] ;

[0213] Furthermore, the coordinates of the feature points on the calibration board need to be transformed to the world coordinate system of the base to eliminate the error between the camera and the calibration board. The specific steps are as follows:

[0214] The formula for correcting calibration plate rotation and center offset is as follows:

[0215] ;

[0216] ;

[0217] Among them, (x obs , y obs (x) represents the coordinates of the calibration plate being observed. Since the Z-axis is not rotated in this embodiment, the Z-axis coordinates are not displayed. Similarly, if the Z-axis is to be rotated, the following steps can be taken; the specific process will not be elaborated upon. Furthermore, (x) bd , y bd () represents the corrected coordinates of the calibration plate. For the calibration plate rotation matrix, It is its inverse matrix (the inverse of an orthogonal matrix is ​​equal to its transpose). This is the calibration plate rotation correction value. , This is the correction value for the center offset of the calibration plate;

[0218] Specifically, three positioning screws are installed at 120° intervals on the calibration plate mounting position on the printer base plate (the distribution logic is consistent with the guide rail control points B1, B2, and B3, with a machining accuracy of ≤0.01mm, and the center points of the three screws are strictly aligned with the origin of the world coordinate system). These screws are individually calibrated and fixed using visual recognition. , and The "3" parameters, as known quantities for subsequent identification, require the following explanation: When applying these parameters, the screw heads should be designed with identifiable visual features such as cross-shaped grooves / circular protrusions. The camera should be temporarily fixed above the base (to prevent movement with the rotating disc) to capture images containing the "3" positioning screws and the calibration plate. This sampling should be repeated "20 times." The visual coordinates of the "3" positioning screws in each sample should be extracted, and the screw center point coordinates (x0, y0) – the visual observation value of the world coordinate system origin – should be calculated using the least squares method. Additionally, preset feature points on the calibration plate (such as checkerboard corner points) should be extracted, and the visual coordinates (xboard, yboard) of the calibration plate center should be calculated. The bd value for each sample should also be calculated. X=xboard-x0, bdᵧ =yboard-y0, ψbd (angle between the characteristic direction of the calibration board and the direction of the screw connection), take the average of 20 sampling results as the final fixed value (bd is required) X / bdᵧ Deviation ≤ 0.03mm, ψ bd (If the deviation is ≤0.2°), it will not be used as an optimization variable in the subsequent second calibration.

[0219] Camera offset correction is calculated as follows:

[0220] x p =x bd -c x , y p =y bd -c y , z p =z obs -c z ;

[0221] Where (cx, cy, cz) is the camera optical center offset, (x p ,y p ,z p (x) represents the coordinates of the center point of the moving disk in world coordinates, and the coordinate list is (x) p ,y p ,z p )_k, where k is the sampling point number.

[0222] According to an embodiment of the present invention, the coordinate calculation based on the coordinate list specifically includes:

[0223] The coordinates of the moving disk control points in the world coordinate system are calculated using the following formula:

[0224] ;

[0225] in, , , For the control point of the moving disk, Let x be the radius of the moving disk. p ,y p ,z p () represents the coordinates of the center point of the moving disk in world coordinates;

[0226] The coordinates of the guide rail control points in the world coordinate system are calculated using the following formula:

[0227] ;

[0228] ;

[0229] in, For encoder reading conversion distance, The inverse of the rotation matrix, For the translation of the guide rail base, , , For base translation at different guide rail control points, Let be the radius of the guide rail distribution circle.

[0230] It should be noted that, in this embodiment, the coordinates of point A (the control point of the moving disk, in the same direction as the guide rail control point B) are calculated as follows:

[0231] ;

[0232] Among them, the guide rail control point B moves only along the Z-axis in the local coordinate system of the guide rail, with coordinates (0, 0, b). i + db i )(b i (The distance for encoder reading conversion), conversion to the world coordinate system requires a rotation matrix. (Inverse rotation matrix) and translation T of the guide rail base i :

[0233] ;

[0234] The guide rail base translation T i for:

[0235] ;

[0236] in, , , The base translation for different guide rail control points corresponds to the guide rail control points. , , The base was translated.

[0237] According to an embodiment of the present invention, the step of constructing a residual objective function based on the coordinates of the moving disk and the coordinates of the guide rail, combined with a preset link length, to obtain the objective function value specifically includes:

[0238] Obtain the length of the connecting rod, wherein the connecting rod is used to connect the control point of the moving disk and the control point of the guide rail;

[0239] The formula for calculating the residual vector of a single sampling point is as follows:

[0240] ;

[0241] in, For a single sampling point, For the Euclidean norm, , , For the control point of the moving disk, , , Here, k represents the guide rail control point, and k is the sampling point number. , and The length of the link;

[0242] The overall objective function value for the entire observation dataset is calculated using the following formula:

[0243] ;

[0244] Wherein, the total objective function value This corresponds to the sum of squares of the residuals of all sampling points, where P is the preset parameter vector, N is the total number of sampling points, and k is the sampling point index. This is the residual vector for a single sampling point.

[0245] It should be noted that, in this embodiment, the core objective of identification is to minimize the "deviation between the square of the actual link length and the square of the theoretical length". The residual vector of a single sampling point is defined as:

[0246] ;

[0247] in, For a single sampling point, For the Euclidean norm, , , For the control point of the moving disk, , , Here, k represents the guide rail control point, and k is the sampling point number. , and Given the link length, further, calculate the overall objective function value for the entire observation dataset. The overall objective function value is the sum of squares of the residuals at all sampling points, calculated as follows:

[0248] ;

[0249] Wherein, the total objective function value This corresponds to the sum of squares of the residuals of all sampling points, where P is the preset parameter vector, N is the total number of sampling points, and k is the sampling point index. The residual vector is a single sampling point. The preset parameter vector P is a "22" parameter vector, and N is the number of sampling points (in practical applications, it is recommended that N≥512 to ensure the degree of freedom of parameter identification). In this embodiment, a rigorous residual function is established based on Euler angle rotation matrix and coordinate transformation theory, which can ensure that the mathematical model is consistent with the kinematics of the actual mechanism.

[0250] According to an embodiment of the present invention, the target parameter vector is obtained by iteratively optimizing the preset parameter vector based on a preset algorithm. The target parameter vector is then written into the inverse kinematics calculation module of the 3D printer to replace the original preset parameter vector, thereby completing the parameter configuration. The target parameter vector includes 22 parameter values.

[0251] It should be noted that, in this embodiment, the values ​​of the "22" parameters are described in Table 1. The preset algorithm used in this embodiment includes the Levenberg-Marquardt algorithm, which is a numerical optimization algorithm widely used to solve nonlinear least squares problems. Its purpose is to find a set of parameters that minimizes the sum of squares of the differences between the model's predicted values ​​and the actual observed values. Since this embodiment applies the algorithm without changing its flow, the specific process of the algorithm will not be described in detail in this embodiment. The necessary key settings are as follows: the initial values ​​of the parameters are based on the design nominal values ​​(e.g., R=R0, L...). i =L0, attitude angle = 0°, visual offset = 0), to avoid the initial value deviating too far from the optimal solution; upper and lower bounds of parameters: set constraints based on the engineering error range (refer to Table 1) to prevent parameters from converging to physically unreasonable values; convergence criterion: function tolerance: the change in the objective function between adjacent iterations |J k+1 - J k |<1e-8; Step tolerance: Change in parameter vector Maximum number of iterations: 10,000 (the actual number of convergence iterations is usually <200).

[0252] It is worth mentioning that, in order to verify the effectiveness of the present invention, an example is given, and the actual parameters set are shown in Table 3 to simulate industrial scenario errors.

[0253] After optimization and convergence using the methods described in the above embodiments, the deviations between the identified values ​​and the true values ​​of each parameter are as follows:

[0254] Guide rail attitude angle deviation: <0.01°;

[0255] Geometric dimensional deviations: R / r / L deviation < 0.02mm, dbᵢ deviation < 0.01mm;

[0256] Visual parameter deviation: cx / cy / cz deviation < 0.1mm, bd X / bdᵧ deviation < 0.05mm, ψ_bd deviation < 0.05°;

[0257] Total residual sum of squares: converges to <1e-6, verifying the stability of the algorithm.

[0258] Furthermore, the identified "22" parameters were substituted into the inverse kinematics of the Delta mechanism to correct the nozzle motion trajectory. The actual printing test results were as follows: end positioning accuracy: ±0.03mm; trajectory tracking error: ≤0.05mm; and suitable printing speed: 2000mm / s (maximum acceleration 30000mm / s²), which meets the requirements of high-speed printing.

[0259] A third aspect of the present invention provides a computer-readable storage medium including a visual calibration method program for a Delta structure parallel 3D printer, wherein when the visual calibration method program for a Delta structure parallel 3D printer is executed by a processor, it implements the steps of a visual calibration method for a Delta structure parallel 3D printer as described in any of the preceding claims.

[0260] This invention discloses a visual calibration method, system, and medium for Delta structure parallel 3D printers. By constructing a 22-parameter end-to-end error model covering mechanical, visual, and sensor components, and combining nonlinear optimization and stabilization sampling processes, it achieves high-precision joint identification and compensation of multi-source errors in Delta parallel 3D printers, improving end-point positioning accuracy to within ±0.03mm, effectively supporting the requirements of high-speed and high-precision printing.

[0261] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0262] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0263] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0264] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0265] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, 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 methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

Claims

1. A visual calibration method for a Delta structure parallel 3D printer, characterized in that, Includes the following steps: The observation dataset is obtained by collecting data based on the preset sampling point coordinates. The observation dataset includes the guide rail encoder readings and the calibration plate coordinates in the camera coordinate system. A coordinate transformation is performed on each observation point in the observation dataset to obtain a list of coordinates of the center point of the moving disk in world coordinates. The coordinate transformation includes calibration plate correction and camera offset correction. Based on the coordinate list, coordinate calculations are performed to obtain the coordinates of the moving disk control point in the world coordinate system and the coordinates of the guide rail control point in the world coordinate system. Based on the coordinates of the moving disk and the coordinates of the guide rail, combined with the preset link length, a residual objective function is constructed to obtain the objective function value; Based on the objective function value and the preset error parameters, a nonlinear optimization iterative solution is performed to obtain the objective parameter vector, and the objective parameter vector is injected into the kinematic model of the 3D printer to complete the parameter configuration.

2. The visual calibration method for a Delta structure parallel 3D printer according to claim 1, characterized in that, The observation dataset obtained by collecting data based on preset sampling point coordinates specifically includes: The sampling points correspond to the planned three-dimensional mesh sampling points within the Delta workspace, covering the range x∈[-140,140] mm, y∈[-140,140] mm, z∈[-350,-5] mm, wherein the sampling points are not collinear; The nozzle is controlled to move to each of the sampling points, and after stopping and waiting for a preset time, the encoder is read and the calibration plate image is captured. The calibration plate coordinates are obtained by extracting the image pixel coordinates of the feature points of the calibration plate from the calibration plate image and correcting them. At the same time, the observation dataset is obtained by extracting the guide rail encoder readings after reading the encoder.

3. The visual calibration method for a Delta structure parallel 3D printer according to claim 2, characterized in that, The process of performing coordinate transformation on each observation point in the observation dataset to obtain a list of coordinates of the center point of the moving disk in world coordinates specifically includes: The formula for correcting calibration plate rotation and center offset is as follows: ； ； Among them, (x obs , y obs (x) represents the coordinates of the calibration plate being observed. bd , y bd () represents the corrected coordinates of the calibration plate. For the calibration plate rotation matrix, This is the calibration plate rotation correction value. , This is the correction value for the center offset of the calibration plate; Camera offset correction is calculated as follows: x p =x bd -c x , y p =y bd -c y , z p =z obs -c z ; Where (cx, cy, cz) is the camera optical center offset, (x p ,y p ,z p (x) represents the coordinates of the center point of the moving disk in world coordinates, and the coordinate list is (x) p ,y p ,z p )_k, where k is the sampling point number.

4. The visual calibration method for a Delta structure parallel 3D printer according to claim 3, characterized in that, The coordinate calculation based on the coordinate list specifically includes: The coordinates of the moving disk control points in the world coordinate system are calculated using the following formula: ； in, , , For the control point of the moving disk, Let x be the radius of the moving disk. p ,y p ,z p () represents the coordinates of the center point of the moving disk in world coordinates; The coordinates of the guide rail control points in the world coordinate system are calculated using the following formula: ； ； in, For encoder reading conversion distance, The inverse of the rotation matrix, For the translation of the guide rail base, , , For base translation at different guide rail control points, Let be the radius of the guide rail distribution circle.

5. A visual calibration method for a Delta structure parallel 3D printer according to claim 4, characterized in that, The process of constructing a residual objective function based on the coordinates of the moving disk and the guide rail, combined with a preset link length, to obtain the objective function value specifically includes: Obtain the length of the connecting rod, wherein the connecting rod is used to connect the control point of the moving disk and the control point of the guide rail; The formula for calculating the residual vector of a single sampling point is as follows: ； in, For a single sampling point, For the Euclidean norm, , , For the control point of the moving disk, , , Here, k represents the guide rail control point, and k is the sampling point number. , and The length of the link; The overall objective function value for the entire observation dataset is calculated using the following formula: ； Wherein, the total objective function value This corresponds to the sum of squares of the residuals of all sampling points, where P is the preset parameter vector, N is the total number of sampling points, and k is the sampling point index. This is the residual vector for a single sampling point.

6. A visual calibration method for a Delta structure parallel 3D printer according to claim 5, characterized in that, The target parameter vector is obtained by iteratively optimizing the preset parameter vector based on the preset algorithm. The target parameter vector is then written into the inverse kinematics calculation module of the 3D printer to replace the original preset parameter vector, thereby completing the parameter configuration. The target parameter vector includes 22 parameter values.

7. A vision calibration system for a Delta structure parallel 3D printer, characterized in that, The system includes a memory and a processor. The memory contains a visual calibration method program for a Delta structure parallel 3D printer. When executed by the processor, the visual calibration method program for the Delta structure parallel 3D printer performs the following steps: The observation dataset is obtained by collecting data based on the preset sampling point coordinates. The observation dataset includes the guide rail encoder readings and the calibration plate coordinates in the camera coordinate system. A coordinate transformation is performed on each observation point in the observation dataset to obtain a list of coordinates of the center point of the moving disk in world coordinates. The coordinate transformation includes calibration plate correction and camera offset correction. Based on the coordinate list, coordinate calculations are performed to obtain the coordinates of the moving disk control point in the world coordinate system and the coordinates of the guide rail control point in the world coordinate system. Based on the coordinates of the moving disk and the coordinates of the guide rail, combined with the preset link length, a residual objective function is constructed to obtain the objective function value; Based on the objective function value and the preset error parameters, a nonlinear optimization iterative solution is performed to obtain the objective parameter vector, and the objective parameter vector is injected into the kinematic model of the 3D printer to complete the parameter configuration.

8. A vision calibration system for a Delta structure parallel 3D printer according to claim 7, characterized in that, The observation dataset obtained by collecting data based on preset sampling point coordinates specifically includes: The sampling points correspond to the planned three-dimensional mesh sampling points within the Delta workspace, covering the range x∈[-140,140] mm, y∈[-140,140] mm, z∈[-350,-5] mm, wherein the sampling points are not collinear; The nozzle is controlled to move to each of the sampling points, and after stopping and waiting for a preset time, the encoder is read and the calibration plate image is captured. The calibration plate coordinates are obtained by extracting the image pixel coordinates of the feature points of the calibration plate from the calibration plate image and correcting them. At the same time, the observation dataset is obtained by extracting the guide rail encoder readings after reading the encoder.

9. A vision calibration system for a Delta structure parallel 3D printer according to claim 8, characterized in that, The process of performing coordinate transformation on each observation point in the observation dataset to obtain a list of coordinates of the center point of the moving disk in world coordinates specifically includes: The formula for correcting calibration plate rotation and center offset is as follows: ； ； Among them, (x obs , y obs (x) represents the coordinates of the calibration plate being observed. bd , y bd () represents the corrected coordinates of the calibration plate. For the calibration plate rotation matrix, This is the calibration plate rotation correction value. , This is the correction value for the center offset of the calibration plate; Camera offset correction is calculated as follows: x p =x bd -c x , y p =y bd -c y , z p =z obs -c z ; Where (cx, cy, cz) is the camera optical center offset, (x p ,y p ,z p (x) represents the coordinates of the center point of the moving disk in world coordinates, and the coordinate list is (x) p ,y p ,z p )_k, where k is the sampling point number.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a visual calibration method program for a Delta structure parallel 3D printer, which, when executed by a processor, implements the steps of a visual calibration method for a Delta structure parallel 3D printer as described in any one of claims 1 to 6.

Images