Online three-dimensional imaging full-size processing and detection integrated method and system

Abstract

The invention discloses an online three-dimensional imaging full-size processing and detection integrated method and system. Performing global three-dimensional data scanning on the surface of the workpiece to obtain an original three-dimensional point cloud; performing data processing on the original three-dimensional point cloud, and extracting contour point cloud data; generating a three-dimensional ideal model by using the contour point cloud data, and further generating a processing path; variable impedance force control is conducted on the grinding and polishing device, and grinding of the workpiece is completed according to the machining path; the workpiece is scanned again after being ground by the grinding and polishing device, and a solid model is generated again; comparing the entity model with the ideal model, calculating an error between the two models, and marking an area exceeding an error allowable range; and the processing path is adjusted according to the area exceeding the error allowable range obtained by the on-line detection module and the error, so that the shape deviation of the solid model is corrected. By integrating three-dimensional imaging, efficient data processing, dynamic force control and a closed-loop feedback mechanism, the technical breakthrough of a processing and detection integrated system is realized.

Description

Technical Field

[0001] This invention relates to the field of industrial automation inspection and intelligent manufacturing, and to an integrated method and system for online three-dimensional imaging full-size processing and inspection. Background Technology

[0002] As industrial manufacturing evolves towards higher precision and efficiency, traditional inspection methods are no longer sufficient to meet the demands of modern industrial production. Traditional offline inspection requires sampling or full-scale inspection of workpieces after processing, typically relying on manual labor or independent inspection equipment. This method separates the inspection process from the production process, increasing the production cycle, and is particularly problematic for complex shapes or large batches of workpieces, failing to achieve rapid response. Furthermore, it cannot monitor dimensional deviations and surface defects that may occur during processing in real time, leading to significant fluctuations in processing quality.

[0003] In recent years, the development of 3D imaging technology has provided new solutions for precision machining and inspection. Technologies such as laser scanning and structured light imaging can quickly and non-contactly acquire 3D morphological data of workpieces. However, the large amounts of 3D data acquired in real time require rapid processing and analysis, but most current systems struggle to achieve efficient data processing capabilities. Existing inspection systems are mostly static, lacking the ability to coordinate with machining equipment to adjust the machining path. Current technologies often separate 3D imaging from machining equipment, making integrated design and efficient collaboration difficult. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing an integrated online three-dimensional imaging full-size processing and inspection method and system, which aims to solve the problem of not being able to monitor and automatically adjust the processing trajectory online.

[0005] In a first aspect, the present invention provides an integrated online three-dimensional imaging full-size processing and inspection method, the method comprising:

[0006] Step S1: Use a laser scanner to perform a global three-dimensional data scan on the surface of the workpiece located on the processing line to obtain the original three-dimensional point cloud;

[0007] Step S2: Process the original 3D point cloud and extract the contour point cloud data;

[0008] Step S3: Ideal trajectory generation

[0009] Generate a workable 3D ideal model from contour point cloud data;

[0010] Based on the three-dimensional ideal model, computer-aided manufacturing software is used to generate the processing path; the processing path is then converted into the robot's working trajectory to guide the robot's work.

[0011] Step S4: Perform variable resistance control on the grinding and polishing device;

[0012] Step S5: Online Detection

[0013] The workpiece after grinding and polishing is re-scanned using a line laser scanner, and the point cloud data of the workpiece after grinding is converted into a solid model. The solid model is compared with the ideal model, the error between the two is calculated, and the area exceeding the allowable error range is marked. The processing path is adjusted according to the area exceeding the allowable error range and the error, thereby correcting the shape deviation of the solid model.

[0014] Preferably, step S2 specifically involves:

[0015] S21. Remove noise points from the original 3D point cloud;

[0016] S22. A voxel grid-based curvature-weighted downsampling method is used to downsample the point cloud after noise points have been removed.

[0017] S23. Normal vector estimation is performed on the downsampled point cloud. The normal vector direction of each point is calculated by using a local surface fitting method based on radius neighborhood search. Then, the angle between the point and the normal vector in its neighborhood is compared. If the angle exceeds a set threshold, the point is marked as a candidate boundary point. Finally, connectivity analysis is performed on the candidate boundary points through density filtering to extract a complete set of geometric contour points for subsequent trajectory generation.

[0018] More preferably, step 21 specifically involves:

[0019] S211. Establishing Point Cloud Topology

[0020] A KD-tree is constructed from the original 3D point cloud, and the k-neighborhood search is used to find the k points closest to the sample to establish the topological relationship of the point cloud.

[0021] S212. A dual filtering method combining radius filtering and statistical filtering is used to remove discrete points in the point cloud.

[0022] Preferably, step S4 specifically involves:

[0023] Based on the contour point cloud data processed in step S2, the actual position x of the grinding and polishing device and the force tracking error e are compared. f A dynamic relationship for impedance control is established between them; where e f = f r -f, f r f and f are the expected contact force and the actual contact force, respectively;

[0024] Assuming the flexibility position x of the grinding and polishing device c= x and the plane from which the steady-state force tracking error is obtained are considered as the environment, and the stiffness of the environment is k. e The actual contact force f is expressed as

[0025] (1)

[0026] Where x e Refers to the undeformed environmental location;

[0027] According to formula (1), the flexibility position x can be determined. c Further:

[0028] (2)

[0029] Assume f r It is a constant. Substituting formula (2) into the dynamic relationship of impedance control and considering x c = x, then the steady-state force tracking error is:

[0030] (3)

[0031] From formula (3), it can be seen that to make the steady-state error equal to zero, the reference position is:

[0032] (4)

[0033] The obtained reference position is combined with the dynamic relationship of impedance control, and the velocity and acceleration are numerically discretized using first-order and second-order difference formulas to construct a discrete control law suitable for controller implementation.

[0034] After the processing path obtained in step S3 is sent to the grinding and polishing device, the grinding and polishing device proceeds according to the reference position x. r move.

[0035] More preferably, step S4 specifically involves:

[0036] In the dynamic relationship of impedance control, the impedance parameter is dynamically adjusted through fuzzy PID logic.

[0037] More preferably, the general form of the PID controller is as follows:

[0038] (5)

[0039] In the formula, e(t) represents the force tracking error at the current moment. , , These represent the proportional, integral, and derivative gains, respectively; u(t) is the output of the PID controller at the current moment.

[0040] By using fuzzy PID controllers, the proportional, integral, and derivative gains are dynamically adjusted based on real-time force errors using fuzzy rules, thus forming a variable impedance controller.

[0041] Preferably, in step S5, if there is a negative deviation in the area exceeding the allowable error range, the density of trajectory points in that area is increased by interpolation of the processing path; if there is a positive deviation in the area exceeding the allowable error range, compensation is made by reducing the actual contact force.

[0042] Secondly, the present invention provides an integrated online three-dimensional imaging full-size processing and inspection system, comprising:

[0043] The data acquisition module is responsible for performing a global three-dimensional data scan on the surface of the workpiece located on the processing line using a laser scanner to obtain the original three-dimensional point cloud.

[0044] The data processing and contour extraction module is responsible for processing the original 3D point cloud data and extracting the contour point cloud data.

[0045] The ideal trajectory generation module is responsible for generating a workable 3D ideal model from the contour point cloud data; based on the 3D ideal model, computer-aided manufacturing software is used to generate the machining path.

[0046] The impedance control module is responsible for controlling the variable impedance force of the grinding and polishing device and completing the grinding of the workpiece according to the processing path.

[0047] The online inspection module is responsible for rescanning the workpiece after it has been polished by the grinding and polishing device using a line laser scanner, converting the point cloud data of the polished workpiece into a solid model, comparing the solid model with the ideal model, calculating the error between the two, and marking the areas that exceed the allowable error range.

[0048] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to perform the method described thereon.

[0049] Fourthly, the present invention provides a computing device, including a memory and a processor, wherein the memory stores executable code, and the processor executes the executable code to implement the method described above.

[0050] The beneficial effects of the present invention include at least the following:

[0051] This invention employs a laser scanner to perform global 3D scanning of workpieces in a processing production line. Combining dual filtering with a voxel mesh-curvature weighted downsampling algorithm, it significantly preserves contour features while reducing the amount of point cloud data. This invention proposes a variable impedance control model, which dynamically adjusts damping and stiffness parameters using fuzzy PID control to address environmental stiffness uncertainties, enhance compliance, avoid surface damage caused by overload, and control force tracking steady-state error. This invention constructs a closed-loop "scan-process-processing-re-scan" chain. Through modular design (data acquisition, contour extraction, trajectory generation, impedance control, and online detection), it achieves seamless hardware and software collaboration, reducing manual intervention and shortening the overall processing cycle.

[0052] In summary, this invention overcomes key technical bottlenecks in online detection, feature preservation, dynamic force control, and system integration by integrating three-dimensional imaging, efficient data processing, dynamic force control, and closed-loop feedback mechanisms, providing a high-precision and highly adaptable integrated solution for intelligent manufacturing. Attached Figure Description

[0053] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] The terms "comprising" and "having," and any variations thereof, used in the embodiments of this invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the steps or units listed, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to such processes, methods, products, or devices.

[0056] This embodiment provides an integrated online 3D imaging full-size processing and inspection method. See attached document. Figure 1 ,include:

[0057] Step S1: Leaf point cloud data acquisition

[0058] The original three-dimensional point cloud is obtained by scanning the surface of the workpiece with a laser scanner to perform global three-dimensional data scanning.

[0059] As an example, the workpiece is located in the first stage of the processing line;

[0060] Step S2: Data Processing and Contour Extraction

[0061] S21. Preprocess the original 3D point cloud to remove noise points; specifically:

[0062] S211. Establishing Point Cloud Topology

[0063] A KD-tree (also known as a K-dimensional tree) is constructed from the original 3D point cloud. The k-neighborhood search is used to find the k points closest to the sample to establish the topological relationship of the point cloud.

[0064] The process of constructing a point cloud KD-tree is as follows:

[0065] (1) Statistically analyze the distribution of point cloud data in three dimensions, find the dimension with the widest point cloud distribution based on variance, and select that dimension as the first dimension for segmentation.

[0066] (2) Sort the point cloud data according to the dimension selected in step (1), and select the point with the median of the sorting result as the root node.

[0067] (3) Assign points less than the median to the left subtree and points greater than the median to the right subtree.

[0068] (4) Iterate through the above steps for the left and right subtrees respectively until the last tree has only one point.

[0069] S212. A dual filtering method combining radius filtering and statistical filtering is used to remove discrete points in the point cloud.

[0070] This invention proposes a dual filtering method combining radius filtering and statistical filtering for discrete point removal. Based on statistical filtering, it adds a judgment of radius neighborhood points. Specifically, before calculating the average distance from a sampling point to all points in its k-neighborhood, radius filtering is used to judge and delete discrete noise points. Then, the remaining point cloud is processed using statistical filtering. The details are as follows:

[0071] Step S2121, Radius Filtering:

[0072] Perform a radius neighborhood search on the KD-tree; count the number of radius neighborhood points of all sampled points in the point cloud; if the number of neighborhood points is greater than a set threshold, retain the point; otherwise, identify the corresponding sampled point as a discrete point and delete it.

[0073] Step S2122

[0074] Assume the position of the i-th sampling point retained after radius filtering is p. i (x i y i , zi ), calculate the average distance s from the point to other points in its k-neighborhood. i :

[0075] (1)

[0076] Assuming the total number of points in the retained point cloud data is n, calculate the total number of sampling points s. i The mean μ and standard deviation :

[0077] (2)

[0078] (3)

[0079] The point cloud after radius filtering is filtered based on the calculated variance, standard deviation, and a set scaling factor r. Points are selected when the average distance s from a point to all points in its k-neighborhood is reached. i If the following formula is satisfied, the point is retained; otherwise, the point is identified as a discrete point and deleted.

[0080] (4)

[0081] S22. A voxel grid-based curvature-weighted downsampling method is used to downsample the point cloud after noise points have been removed.

[0082] Because the collected point cloud data has a high density, it will affect the computational efficiency during subsequent point cloud processing. Therefore, it is necessary to process the point cloud data to reduce the density of the point cloud while preserving the macroscopic geometric features of the point cloud as much as possible. This processing is called point cloud downsampling.

[0083] This invention proposes a method based on a combination of voxel grid and curvature-weighted downsampling. Specifically:

[0084] The point cloud with discrete points removed is downsampled using the voxel grid method, which divides the point cloud into small voxels, and the points in each voxel are simplified into a representative point.

[0085] For a point within a small voxel, calculate the curvature of each point;

[0086] Points with curvature greater than the threshold (i.e., areas where the object's surface changes significantly) are assigned a larger downsampling weight, while sampling of areas with smaller curvature is reduced.

[0087] By combining these two methods, both computational efficiency and important features of the object's surface are preserved.

[0088] S23. Normal vector estimation is performed on the downsampled point cloud. The normal vector direction of each point is calculated by using a local surface fitting method based on radius neighborhood search. Then, the angle between the point and the normal vector in its neighborhood is compared. If the angle exceeds a set threshold, the point is marked as a boundary candidate point. Finally, the connectivity analysis of the boundary candidate points is performed through density filtering to extract the complete geometric contour point set for subsequent trajectory generation.

[0089] The process for determining whether a point is a contour point is as follows:

[0090] (1) Use KD-trees to spatially index point cloud data to accelerate domain search efficiency. For any point... An efficient radius neighborhood search can be achieved using a KD-tree to obtain the set of its neighboring points.

[0091] (5)

[0092] in For point cloud datasets; For the threshold;

[0093] (2) Calculation using the local surface fitting method The normal vector at point n = (A, B, C).

[0094] (3) According to The calculation of the normal vector n is performed after The tangent plane of set Q projects the points in set Q onto the tangent plane, denoted as set N.

[0095] (4) Choose any point in set N. ,by With the origin of the coordinate system as the origin and the normal vector n as the w-axis, Let the unit vector be the u-axis and w×u be the v-axis, and construct a local reference coordinate system, denoted as ( , u, v, w).

[0096] (5) Calculate separately Divide from set N Other points P k vector P k The clockwise angle θ between the local coordinate system and the u-axis n Let Ω be the set Ω = (θ1, θ2, …, θ n According to the formula below, the set Ω' = (θ'1 , θ'2 , …, θ') is obtained by subtracting each pair of adjacent included angles. n-1 ).

[0097] (6)

[0098] (6) Find the maximum value in the set Ω'. If the maximum value is greater than the set threshold, then the point is an edge point.

[0099] (7) Perform the above steps on other points in the point cloud in sequence to find all edge points and generate an edge point cloud.

[0100] Step S3: Ideal trajectory generation

[0101] Import the contour point cloud data into CloudCompare to generate a workable 3D ideal model;

[0102] Based on the three-dimensional ideal model, computer-aided manufacturing (CAM) software is used to generate the machining path; the machining path is then converted into the robot's machining trajectory through the robot toolbox to guide the robot's work.

[0103] Step S4: Novel Variable Impedance Control

[0104] Impedance control has been widely used in robotic machining due to its simplicity, flexibility, and stable performance. However, in practical applications, the accuracy of dynamic models for grinding and polishing devices is often limited, making it difficult to support torque control modes. Therefore, position controllers are typically used to adjust mechanical impedance by changing the reference trajectory, i.e., position-based impedance control.

[0105] Based on the contour point cloud data processed in step S2, impedance control establishes a dynamic relationship e between the actual position x of the grinding and polishing equipment and the force tracking error. f = f r -f, where f r Let f be the desired contact force and the actual contact force, respectively, which can be expressed as:

[0106] (7)

[0107] in , These are the first and second derivatives of the actual position x, representing the velocity and acceleration of the end effector, respectively; x r That is the desired position.

[0108] In the formula m f b f and k f These are the mass term, damping term, and stiffness term, as well as x. r This indicates the reference position. Assuming the grinding and polishing device has good position tracking accuracy, the compliance position x can be obtained. c :x c = x, where x c It is a compliance position.

[0109] Assume the plane from which the steady-state force tracking error is obtained is considered the environment, and the stiffness of the environment is k. e The actual contact force can be expressed as

[0110] (8)

[0111] Where x e Refers to the undeformed environmental location;

[0112] According to formula (8), the flexibility position x can be determined. c Further:

[0113] (9)

[0114] If the environment is not flat, it is difficult to obtain steady-state force tracking error. Similar to existing practices, the environment here is considered as a plane from which steady-state force tracking error is obtained. Meanwhile, it is assumed that f... r It is a constant. Substituting equation (9) into equation (7) and considering x c In the equation x, the steady-state force tracking error is:

[0115] (10)

[0116] As can be seen from equation (10), there are two methods to make the steady-state error equal to zero. One is to set the stiffness k. f Alternatively, you can set the reference position to zero.

[0117] (11)

[0118] Because zero stiffness is neither natural nor intuitive. Furthermore, stiffness is the most important impedance parameter affecting task performance. Therefore, k... f It may not be unreasonable to set it to zero, so equation (11) was naturally adopted. According to equation (11), if the environmental parameters are inaccurate or change, steady-state force error will occur.

[0119] By combining the aforementioned reference position (Equation 11) with the impedance control equation (Equation 7), and using first-order and second-order difference formulas to numerically discretize the velocity and acceleration, a discrete control law suitable for controller implementation can be constructed. After the processing path obtained in step S3 is sent to the grinding and polishing device, the device can proceed according to this reference position x. r Move to ensure force tracking effect.

[0120] S42. Adjusting the damping and stiffness parameters of the target impedance online using the variable impedance method.

[0121] In impedance control, the selection of impedance parameters is a crucial factor affecting transient performance and steady-state force tracking error. Therefore, this invention proposes a variable impedance method to adjust the damping and stiffness parameters of the target impedance online. By using real-time force error and fuzzy PID control, the impedance parameters can be adjusted. Thus, the generated compliance position trajectory can adapt to changes in environmental parameters, thereby achieving accurate force tracking. Specific details are described below.

[0122] First, the designed variable impedance control is further expressed by formula (7) as follows:

[0123] (12)

[0124] In the formula b f (t) and k f (t) represents the time-varying damping and variable stiffness parameters, respectively.

[0125] Then, fuzzy PID control design is performed. The key to fuzzy PID control design is to dynamically adjust the PID parameter b through fuzzy logic. f (t) and k f (t). The general form of a PID controller is as follows:

[0126] (13)

[0127] In the formula The force tracking error at the current moment, , , These represent proportional, integral, and differential gains, respectively.

[0128] A six-dimensional force sensor collects six-dimensional force / torque information from the end effector during processing in real time, which is used as the actual contact force. This actual contact force is then compared with the expected contact force to obtain the force error. This error signal is input to a fuzzy logic module, which dynamically adjusts the system's stiffness and damping coefficients according to preset fuzzy rules.

[0129] With fuzzy PID, the adjustment of proportional, integral, and derivative gains is no longer fixed, but is dynamically adjusted by fuzzy rules based on real-time force error, thus forming a new variable impedance controller.

[0130] The fuzzy control rules are designed as follows:

[0131] In fuzzy PID control, fuzzy rules determine how the PID gain is adjusted. Here is a basic example of a fuzzy rule:

[0132] Input variable: Force error e(t)

[0133] Output variable: Damping coefficient bf (t) and stiffness coefficient k f (t)

[0134] Example of fuzzy control rules:

[0135] If force error If the PID proportional gain is large, then... The integral gain should be increased. and differential gain It should be moderate. The corresponding damping coefficient b is... f (t) should be moderate and stiffness coefficient k f (t) should be increased.

[0136] If force error If the PID is small, then the proportional gain of the PID controller is small. The integral gain should be reduced. and differential gain It should be moderate. The corresponding damping coefficient b is... f (t) should be moderate and stiffness coefficient k f (t) should be reduced.

[0137] By using such fuzzy rules, the damping coefficient b of the force-position relationship can be dynamically adjusted in impedance control. f (t) and stiffness coefficient k f (t).

[0138] S5, Online Detection

[0139] The workpiece after grinding and polishing is re-scanned using a line laser scanner, and the point cloud data of the workpiece is converted into a solid model. An ICP iterative algorithm is used to precisely align the solid model with a standard CAD model. PolyWorks software is used to calculate the point-to-point position between the two, generating an error heatmap. Areas exceeding the allowable error range are marked, and their coordinates and deviations are exported as input data for the next round of machining path optimization. If a region has a negative deviation (incomplete machining), the trajectory point density in that region is increased through path interpolation; if a region has a positive deviation (over-machining), compensation is achieved by reducing the contact force.

[0140] Table 1 Measurement Error Report

[0141] point Actual X Measure X Deviation X Actual Y Measuring Y DeviationY Actual Z Measurement Z Deviation Z 1 -25.618 -25.582 0.036 11.744 11.728 -0.016 7.099 7.161 0.062 2 -2.123 -2.114 0.009 11.584 11.566 -0.018 3.547 3.42 -0.127 3 31.208 31.525 0.317 12.394 12.489 0.095 21.578 21.701 0.123 4 -17.149 -17.132 0.017 12.603 12.625 0.022 26.223 26.28 0.057 5 -11.412 -11.44 -0.028 13.278 13.309 0.031 41.249 41.131 -0.118 6 8.532 8.45 -0.082 14.333 14.411 -0.078 64.744 64.707 -0.037 7 -16.329 -16.393 -0.064 14.713 14.782 0.069 73.214 73.292 0.078 8 -12.431 -12.411 0.02 13.086 13.171 0.085 100.257 100.255 -0.002 9 18.37 18.469 0.099 1.859 1.884 0.025 93.428 93.475 0.047 10 -10.865 -10.919 -0.054 14.026 14.096 0.07 57.914 58.179 0.265 11 2.835 3.056 0.221 -0.753 -0.568 0.185 154.63 154.594 -0.036 12 12.38 12.526 0.146 -9.501 -9.355 0.146 203.264 203.411 0.147 13 -2.842 -2.562 0.28 5.533 5.478 -0.055 204.895 204.815 -0.08 14 -6.127 -5.925 0.202 8.683 8.707 0.024 172.382 172.293 -0.089 15 5.267 5.379 0.112 -1.062 -0.757 0.305 129.496 129.493 -0.003 16 -7.775 -7.649 0.126 9.378 9.458 0.08 139.054 139.073 0.019 17 17.274 17.407 0.133 -2.517 -2.362 0.155 130.586 130.513 -0.073 18 -0.997 -0.83 0.167 3.657 3.747 0.09 98.077 98.085 0.008 19 -0.717 -0.49 0.227 2.778 2.878 0.1 115.563 115.553 -0.01 20 7.166 7.15 -0.016 13.548 13.618 0.07 47.259 47.357 0.098 21 -10.865 -10.854 0.011 12.112 12.116 0.004 15.295 15.293 -0.002 22 -26.711 -26.773 -0.062 12.235 12.238 0.003 18.027 18.061 0.034 23 22.193 22.539 0.346 -0.78 -1.688 -0.908 114.195 114.064 -0.131 24 -6.931 -6.728 0.203 11.184 11.204 0.02 187.953 187.878 -0.075 25 25.471 25.391 -0.08 14.627 14.742 0.115 71.301 71.273 -0.028

[0142] This invention provides an electronic device, specifically, the electronic device includes a memory and a processor, the memory stores executable code, and when the processor executes the executable code, it implements the method described in any of the embodiments.

[0143] The memory may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk drive. Communication between this system network element and at least one other network element is achieved through at least one communication interface (which can be wired or wireless), such as the Internet, wide area network, local area network, or metropolitan area network.

[0144] The bus can be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc.

[0145] The memory is used to store programs. After receiving an execution instruction, the processor executes the program. The method executed by the device for defining the flow process disclosed in any of the foregoing embodiments of the present invention can be applied to the processor or implemented by the processor.

[0146] The processor may be an integrated circuit chip with signal processing capabilities. In implementation, the steps of the above methods can be completed by integrated logic circuits in the processor's hardware or by software instructions. The processor can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in the memory, and the processor reads the information in the memory and, in conjunction with its hardware, completes the steps of the above method.

[0147] The computer program product of the readable storage medium provided in the embodiments of the present invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the foregoing method embodiments. For specific implementation, please refer to the foregoing method embodiments, which will not be repeated here.

[0148] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0149] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An integrated online three-dimensional imaging full-size processing and inspection method, characterized in that, The method includes: Step S1: Use a laser scanner to perform a global three-dimensional data scan on the surface of the workpiece located on the processing line to obtain the original three-dimensional point cloud; Step S2: Process the original 3D point cloud and extract the contour point cloud data; Step S3: Ideal trajectory generation: Generate a workable 3D ideal model from contour point cloud data; Based on the three-dimensional ideal model, computer-aided manufacturing software is used to generate the machining path; the machining path is then converted into the robot's machining trajectory to guide the robot's work. Step S4: Perform variable resistance control on the grinding and polishing device; Step S5, Online Detection: The workpiece after grinding and polishing is re-scanned using a line laser scanner, and the point cloud data of the workpiece after grinding is converted into a solid model. The solid model is compared with the ideal model, the error between the two is calculated, and the area exceeding the allowable error range is marked. The processing path is adjusted according to the area exceeding the allowable error range and the error, thereby correcting the shape deviation of the solid model.

2. The method according to claim 1, characterized in that, Step S2 is as follows: S21. Remove noise points from the original 3D point cloud; S22. A voxel grid-based curvature-weighted downsampling method is used to downsample the point cloud after noise points have been removed. S23. Normal vector estimation is performed on the downsampled point cloud. The normal vector direction of each point is calculated by using a local surface fitting method based on radius neighborhood search. Then, the angle between the point and the normal vector in its neighborhood is compared. If the angle exceeds a set threshold, the point is marked as a candidate boundary point. Finally, connectivity analysis is performed on the candidate boundary points through density filtering to extract a complete set of geometric contour points for subsequent trajectory generation.

3. The method according to claim 2, characterized in that, Step 21 specifically involves: S211. Establishing point cloud topology: A KD-tree is constructed from the original 3D point cloud, and the k-neighborhood search is used to find the k points closest to the sample to establish the topological relationship of the point cloud. S212. A dual filtering method combining radius filtering and statistical filtering is used to remove discrete points in the point cloud.

4. The method according to claim 1, characterized in that, Step S4 is as follows: Based on the contour point cloud data processed in step S2, the actual position x of the grinding and polishing device and the force tracking error e are compared. f A dynamic relationship for impedance control is established between them; where e f = f r - f, f r f and f are the expected contact force and the actual contact force, respectively; Assuming the flexibility position x of the grinding and polishing device c = x and the plane from which the steady-state force tracking error is obtained are considered as the environment, and the stiffness of the environment is k. e The actual contact force f is expressed as （1） Where x e Refers to the undeformed environmental location; According to formula (1), the flexibility position x can be determined. c Further: （2） Assume f r It is a constant. Substituting formula (2) into the dynamic relationship of impedance control and considering x c = x, then the steady-state force tracking error is: （3） From formula (3), it can be seen that to make the steady-state error equal to zero, the reference position is: (4) The obtained reference position is combined with the dynamic relationship of impedance control, and the velocity and acceleration are numerically discretized using first-order and second-order difference formulas to construct a discrete control law suitable for controller implementation. After the processing path obtained in step S3 is sent to the grinding and polishing device, the grinding and polishing device proceeds according to the reference position x. r move.

5. The method according to claim 4, characterized in that, Step S4 is as follows: In the dynamic relationship of impedance control, the impedance parameter is dynamically adjusted through fuzzy PID logic.

6. The method according to claim 5, characterized in that, The general form of a PID controller is as follows: (5) In the formula, e(t) represents the force tracking error at the current moment. , , These represent the proportional, integral, and derivative gains, respectively; u(t) is the output of the PID controller at the current moment. By using fuzzy PID controllers, the proportional, integral, and derivative gains are dynamically adjusted based on real-time force errors using fuzzy rules, thus forming a variable impedance controller.

7. The method according to claim 1, characterized in that, If there is a negative deviation in the area exceeding the allowable error range in step S5, the density of trajectory points in that area is increased by interpolation of the processing path; if there is a positive deviation in the area exceeding the allowable error range, compensation is made by reducing the actual contact force.

8. An integrated online three-dimensional imaging full-size processing and inspection system for implementing the method of any one of claims 1-7, characterized in that, include: The data acquisition module is responsible for performing a global three-dimensional data scan on the surface of the workpiece located on the processing line using a laser scanner to obtain the original three-dimensional point cloud. The data processing and contour extraction module is responsible for processing the original 3D point cloud data and extracting the contour point cloud data. The ideal trajectory generation module is responsible for generating a workable 3D ideal model from the contour point cloud data; based on the 3D ideal model, computer-aided manufacturing software is used to generate the machining path. The impedance control module is responsible for controlling the variable impedance force of the grinding and polishing device and completing the grinding of the workpiece according to the processing path. The online inspection module is responsible for rescanning the workpiece after it has been polished by the grinding and polishing device using a line laser scanner, converting the point cloud data of the polished workpiece into a solid model, comparing the solid model with the ideal model, calculating the error between the two, and marking the areas that exceed the allowable error range. The correction module is responsible for adjusting the processing path based on the areas exceeding the allowable error range obtained from the online detection module, thereby correcting the shape deviations of the solid model.

9. A computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to perform the method of any one of claims 1-7.

10. A computing device comprising a memory and a processor, wherein the memory stores executable code, and the processor, when executing the executable code, implements the method of any one of claims 1-7.

Images