A collaborative three-dimensional measurement method and device using AGV and robotic arm

By combining real-time vibration monitoring and quality assessment with a dynamic reshooting strategy, the problems of error and low efficiency caused by vibration in collaborative 3D measurement between AGVs and robotic arms have been solved, achieving efficient and accurate 3D measurement.

CN123103773APending Publication Date: 2026-11-17GRANDLEADINTELLIGENTSYSTDONGGUANCO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611597083.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-15
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In existing 3D measurement solutions involving AGVs and robotic arms, the vibration of the AGV when it stops leads to measurement errors and low efficiency. Furthermore, the delayed quality assessment results in repeated measurements and multiple rounds of re-taking, which reduces measurement efficiency and accuracy.

Method used

By using a real-time vibration monitoring and fine-tuning mechanism, combined with real-time quality assessment and dynamic reshooting strategy, vibration is monitored using an accelerometer, the AGV position or robotic arm posture is adjusted in real time, data quality is scored and reshooting positions are determined, ensuring data quality and reducing repeated measurements.

Benefits of technology

It effectively suppresses vibration interference during the measurement process, improves measurement efficiency and data quality, reduces measurement errors and repetitive operations, and realizes an intelligent and automated measurement process.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123103773A_ABST
    Figure CN123103773A_ABST
Patent Text Reader

Abstract

This application discloses a collaborative three-dimensional measurement method and device using an AGV and a robotic arm. The method includes: controlling the AGV to stop after reaching a measurement point, collecting vibration signals and calculating the amplitude; if the amplitude is not greater than a preset amplitude threshold, controlling the robotic arm to sequentially move the measuring head to each measurement viewpoint included in the current measurement point for image capture according to the motion path; after capturing images at each measurement viewpoint, performing a quality score on the captured data; if the quality score is lower than a preset quality threshold, determining a re-capture position; determining whether the re-capture position is within the reachable range of the robotic arm at the current measurement point of the AGV; if so, controlling the robotic arm to reach the re-capture position and performing re-capture when the amplitude is not greater than the preset amplitude threshold; after the AGV has traversed all measurement points and the robotic arm has completed all measurement viewpoints, if the re-capture queue is not empty, marking the re-capture requirement in the re-capture queue as abnormal. This measurement method effectively improves measurement efficiency.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of industrial automation three-dimensional measurement technology, and in particular to a collaborative three-dimensional measurement method and equipment using AGV and robotic arm. Background Technology

[0002] Automated Guided Vehicles (AGVs) are battery-powered, driverless transport vehicles that navigate along pre-set paths using magnetic strips, tracks, lasers, or QR codes. In 3D photogrammetry of large workpieces, due to the limited field of view in a single frame, multiple photos need to be taken from various angles and stitched together. Currently, the mainstream automated measurement solution uses an AGV equipped with a robotic arm, with a 3D measuring head at the end of the arm. The AGV transports the robotic arm to various measurement stations, where the arm drives the measuring head to take multi-angle photos of the workpiece. AGV-equipped robotic arm 3D measurement saves significant manual labor and improves measurement efficiency. Furthermore, the workpiece remains stationary throughout the measurement process, eliminating vibrations and deformations caused by workpiece rotation and reducing measurement errors. However, vibrations may occur when the AGV stops at measurement points, and the quality of the finished product captured by the AGV and robotic arm in existing technologies is usually evaluated only after all measurement points have been completed. If quality is found to be substandard, multiple rounds of re-photograms are required, leading to a significant decrease in overall measurement efficiency. Summary of the Invention

[0003] Based on this, embodiments of this application provide a collaborative three-dimensional measurement method and device using an AGV and a robotic arm to improve measurement efficiency.

[0004] Therefore, in a first aspect, embodiments of this application provide a collaborative three-dimensional measurement method using an AGV and a robotic arm, comprising: controlling the AGV to stop after reaching the measurement point, collecting vibration signals through an accelerometer and calculating the amplitude; if the amplitude is greater than a preset amplitude threshold and continues for more than a preset time, controlling the AGV to perform position fine-tuning or retracting the robotic arm to lower the center of gravity, then re-collecting vibration signals and determining whether the amplitude exceeds the limit; if the amplitude is not greater than the preset amplitude threshold, controlling the robotic arm to sequentially move the measuring head to each measurement viewpoint included in the current measurement point according to the motion path for taking pictures; after each completion... After each measurement viewpoint is captured, the captured data is scored for quality. If the quality score is lower than a preset quality threshold, a reshoot location is determined. It is then determined whether the reshoot location is within the reach of the robotic arm at the current measurement point of the AGV. If yes, the robotic arm is controlled to reach the reshoot location, and a reshoot is performed when the amplitude is not greater than a preset amplitude threshold. If no, the reshoot requirement is recorded in the reshoot queue, and the reshoot is performed at the next measurement point of the AGV. When the AGV has traversed all measurement points and the robotic arm has completed all measurement viewpoints, if the reshoot queue is not empty, the reshoot requirement in the reshoot queue is marked as abnormal.

[0005] In one embodiment, the AGV stops after reaching the measurement point, and the vibration signal is collected by an accelerometer to calculate the amplitude, including: The AGV stops after reaching the measurement point, and vibration signals are collected in real time by a triaxial accelerometer installed on the AGV. The vibration signals include acceleration components in three directions. , where t is time; Calculate the vibration amplitude A(t):

[0006] The vibration amplitude is smoothed by applying a moving average filter. :

[0007] Where N is the length of the sliding window. The sampling period.

[0008] In one embodiment, the metrics for quality scoring include, but are not limited to, point cloud density. Point cloud noise and coverage integrity ; The quality score Q is a weighted average of all indicators:

[0009] in, The weights of each indicator are given.

[0010] In one embodiment, if the quality score is lower than a preset quality threshold, the location for reshooting is determined, including: If the quality score is lower than the preset quality threshold, first identify the type of quality defect: if the point cloud density is lower than the density threshold, the quality defect is insufficient point cloud density; if the point cloud noise is higher than the noise threshold, the quality defect is excessive point cloud noise; if the coverage integrity is lower than the integrity threshold, the quality defect is incomplete coverage. The location for reshoots was then determined based on the quality defects.

[0011] In one embodiment, if the quality defect is insufficient point cloud density, the measuring head is moved a first distance towards the workpiece, starting from the current measuring head position, while keeping the current measuring direction unchanged, to obtain the re-shooting position.

[0012] In one embodiment, if the quality defect is excessive point cloud noise, the measuring head is moved a second distance away from the workpiece, starting from the current measuring head position, to obtain the re-shooting position. Then, the measuring head orientation is adjusted to reduce the angle between the measuring head optical axis and the workpiece surface normal vector.

[0013] In one embodiment, if the quality defect is incomplete coverage, the measuring head is moved a third distance away from the workpiece, starting from the current measuring head position, while keeping the current measuring direction unchanged, to obtain the replacement position.

[0014] In one embodiment, when there is more than one quality defect type, the reshoot location is determined by prioritizing the quality defect type ranked first, in the order of excessive point cloud noise, incomplete coverage, and insufficient point cloud density.

[0015] In one embodiment, the AGV and robotic arm collaborative three-dimensional measurement method further includes: after each re-shoot, the quality score of the re-shoot data is scored again; if the quality score of the re-shoot data is lower than a preset quality threshold, the re-shoot position is re-determined and re-shoot is performed until the quality score of the re-shoot data is not lower than the preset quality threshold or the maximum number of re-shoots is reached.

[0016] In one embodiment, the AGV and robotic arm collaborative three-dimensional measurement method further includes: After the robotic arm has completed all the measurement viewpoints of the current measurement site, if the reshoot queue is not empty, the reshoot requirements in the reshoot queue are merged with the measurement viewpoints of the next measurement site into a set. The AGV is controlled to move to the next measurement point, and the vibration signal is collected by the accelerometer and the amplitude is calculated. If the amplitude is not greater than the preset amplitude threshold, the robotic arm's measuring head is first controlled to traverse and process the reshooting requirements in the set; if the reshooting requirements in the set are completed, the corresponding reshooting requirements are deleted from the reshooting queue; otherwise, the corresponding reshooting requirements are retained in the reshooting queue. The robotic arm's measuring head is then moved to various measuring viewpoints within the assembly to take pictures and measurements.

[0017] Secondly, embodiments of this application also provide an electronic device, which includes a processor and a memory; wherein the memory stores a computer program, the computer program being adapted to be loaded by the processor and executed by the processor to provide the AGV and robotic arm collaborative three-dimensional measurement method described above.

[0018] This application provides a collaborative three-dimensional measurement method for AGV and robotic arm. It suppresses vibration interference during the measurement process through real-time vibration monitoring and fine-tuning mechanisms, and ensures data quality by combining real-time quality assessment and dynamic reshooting strategies. This avoids repeated measurements caused by vibration and delayed quality assessment in traditional solutions, reduces measurement errors caused by vibration when the AGV stops, and avoids multiple reshoots due to unqualified quality, thus effectively improving measurement efficiency. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a perspective view of the AGV and robotic arm assembly in the embodiments of this application.

[0021] Figure 2 This is a flowchart of the collaborative three-dimensional measurement method of AGV and robotic arm in an embodiment of this application.

[0022] Figure 3 This is a schematic diagram of the electronic device frame according to an embodiment of this application. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0024] It should be understood that the described embodiments are merely some, not all, of the embodiments in this application. All other embodiments obtained by those skilled in the art based on the embodiments in this application without inventive effort are within the scope of protection of this application.

[0025] In the following description, when referring to the accompanying drawings, the same numbers in different drawings denote the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0026] In the description of this application, it should be understood that the terms "first," "second," "third," etc., are used only to distinguish similar objects and are not necessarily used to describe a specific order or sequence, nor should they be construed as indicating or implying relative importance. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances. Furthermore, in the description of this application, unless otherwise stated, "multiple" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0027] In collaborative 3D measurement using AGVs and robotic arms, the quality assessment of measurement data is typically performed manually after all measurement points have been completed. This lag in assessment means that when the measurement data quality is found to be unsatisfactory, some or all measurement viewpoints need to be re-captured, leading to decreased measurement efficiency. Furthermore, because quality assessment is not performed promptly after each individual measurement viewpoint, the system cannot take immediate remedial action when problems occur, increasing unnecessary repetitive operations.

[0028] For example, in 3D photogrammetry of a large aircraft wing, an AGV (Automated Guided Vehicle) is controlled to move to multiple preset measurement points, and a 3D measurement head at the end of the robotic arm takes multi-angle images of the wing surface. Specifically, at a certain measurement point, after the robotic arm completes the imaging of all measurement viewpoints, the AGV moves to the next position. After all points have been measured, manual evaluation reveals that the point cloud data for a certain viewpoint has excessive noise, requiring re-imaging. Therefore, since this viewpoint is located at a point the AGV has already left, the AGV must be rescheduled to return to its original position for re-imaging, which prolongs the measurement cycle and increases the complexity of system scheduling.

[0029] If these issues are not addressed, subsequent processes will require frequent re-taking, extending the overall measurement time. Furthermore, multiple re-taking sessions may introduce additional systematic errors, affecting the accuracy and reliability of the measurement results. In automated measurement systems, this inefficient evaluation mechanism reduces the system's autonomous decision-making ability and increases the frequency of human intervention, thereby diminishing the automation advantages of AGV-robotic arm collaborative measurement.

[0030] Therefore, embodiments of this application provide a collaborative three-dimensional measurement method using an AGV and a robotic arm, such as... Figure 2 As shown, it includes: controlling the AGV to stop after reaching the measurement point, collecting vibration signals through an accelerometer and calculating the amplitude; if the amplitude is greater than a preset amplitude threshold and continues for more than a preset time, controlling the AGV to perform position fine-tuning or retracting the robotic arm to lower the center of gravity, then re-collecting vibration signals and determining whether the amplitude exceeds the limit; if the amplitude is not greater than the preset amplitude threshold, controlling the robotic arm to move the measuring head sequentially to each measurement viewpoint included in the current measurement point for shooting; after completing the shooting of each measurement viewpoint, scoring the quality of the captured data; if the quality score is lower than the preset quality threshold, determining the reshooting position; determining whether the reshooting position is within the reachable range of the robotic arm of the current measurement point of the AGV: if yes, controlling the robotic arm to reach the reshooting position and performing reshooting when the amplitude is not greater than the preset amplitude threshold; if no, recording the reshooting requirement in the reshooting queue and performing reshooting at the next measurement point of the AGV; when the AGV has traveled to all measurement points and the robotic arm has completed all measurement viewpoints, if the reshooting queue is not empty, marking the reshooting requirement in the reshooting queue as abnormal.

[0031] For ease of understanding, some technical terms in the embodiments are explained below: An Automated Guided Vehicle (AGV) is an unmanned transport vehicle that travels along a specific path using a preset navigation method to transport equipment or workpieces to a designated location. In this embodiment, the AGV is used to carry a robotic arm, the end of which is equipped with a measuring head, which is then moved to different measuring points.

[0032] A robotic arm is a programmable, multi-degree-of-freedom automated mechanical device, typically equipped with a 3D measuring head at its end effector. In this embodiment, the robotic arm is used to carry the 3D measuring head (such as a GOM measuring head) and move it to various measuring viewpoints according to a preset motion path for data acquisition.

[0033] A measurement point refers to a specific location where the AGV needs to stop when performing a 3D measurement task. At each measurement point, the robotic arm will extend and take pictures of the workpiece from multiple angles.

[0034] An accelerometer is a device that measures the acceleration of an object. In this embodiment, an accelerometer is used to acquire vibration signals in real time after the AGV stops, in order to assess the stability of the measurement environment.

[0035] A measuring head is a device mounted at the end of a robotic arm to acquire three-dimensional data, such as a 3D scanner or a photogrammetric camera. Driven by the robotic arm, the measuring head captures images of different areas of the workpiece.

[0036] The measurement viewpoint refers to the specific spatial position and posture that the measuring head needs to move to at each measurement point in order to photograph the workpiece from a specific angle and obtain complete measurement data.

[0037] A quality score is a quantitative assessment of the quality of the acquired 3D data. This score can be calculated comprehensively based on multiple indicators such as point cloud density, noise, and coverage integrity to determine whether the data meets measurement requirements.

[0038] The reachable range of a robotic arm refers to all the spatial areas that its end effector can reach from the current AGV measurement point. Determining whether the re-shooting location falls within this range determines whether the re-shooting task can be executed immediately.

[0039] The retake queue is a storage structure used to record retake requests that cannot be executed immediately. When a retake location is outside the current reachable range of the robotic arm, the retake request is recorded in the retake queue for processing at subsequent measurement points.

[0040] The AGV and robotic arm devices involved in the embodiments are as follows: Figure 1As shown, a robotic arm 12 is fixedly connected to the top of the AGV 11, and a GOM measuring head 13 is fixed at the end of the robotic arm 12. The AGV 11 drives the robotic arm 12 to move to the corresponding measuring position, and the robotic arm 12 drives the GOM measuring head 13 to move in space to reach the corresponding measuring viewpoint.

[0041] In the AGV and robotic arm collaborative 3D measurement method of this embodiment, the AGV is first controlled to stop after reaching a predetermined measurement point. To assess the stability of the measurement environment, an accelerometer mounted on the AGV collects its own vibration signals, and the current amplitude is calculated based on the collected signals. For example, a simple peak detection method or a root mean square (RMS) calculation method can be used to obtain the amplitude. In one implementation, the accelerometer can be a single-axis sensor, measuring vibration in only one direction; in another implementation, a multi-axis sensor can be used to perform comprehensive amplitude calculation based on data from multiple axes.

[0042] Secondly, the calculated amplitude is assessed. If the amplitude exceeds a preset threshold and this excessive state persists for more than a preset time, it indicates that the current environmental vibration is too great and unsuitable for high-precision 3D measurement. In this case, the AGV is fine-tuned, for example, by moving slightly forward, backward, or left and right to find a more stable stopping point. Alternatively, the robotic arm can be retracted downwards to lower the system's center of gravity and reduce swaying. After these adjustments, the vibration signal is re-collected via the accelerometer, and the amplitude is reassessed until it meets the requirements. If the amplitude fails to meet the requirements after a maximum of three adjustments, an error is reported, terminating the 3D measurement, indicating a test error, and requiring manual inspection.

[0043] Furthermore, if the amplitude is not greater than a preset amplitude threshold, indicating a stable environment, the robotic arm is controlled to move the measuring head sequentially to each measurement viewpoint within the current measurement location for image capture, following a preset motion path. The robotic arm's motion path can be generated in advance through teach programming or offline planning. In one implementation, the measuring head can use a GOM 3D laser scanner for data acquisition.

[0044] After each measurement viewpoint is captured, the acquired 3D data is scored for quality. This quality scoring aims to assess the usability and accuracy of the data. For example, preliminary automated scoring is performed based on statistical indicators such as the number of point clouds and average point spacing. If the quality score is lower than a preset quality threshold, the current captured data is considered unqualified, and a reshoot location needs to be determined. The reshoot location can be set as a small offset from the current measurement head position, such as moving a short distance in a fixed direction. If the quality score is not lower than the preset quality threshold, it is determined whether this is the last measurement viewpoint for the current measurement site. If not, the robotic arm controls the measurement head to move to the next measurement viewpoint for 3D measurement. If the last measurement viewpoint for the current measurement site has already been captured and measured, it is determined whether this is the last measurement site. If not, the AGV moves to the next measurement site for capturing and measuring.

[0045] Subsequently, it is determined whether the identified re-shooting position is within the reachable range of the robotic arm at the AGV's current measurement point. The reachable range of the robotic arm can be calculated using its kinematic model. If the re-shooting position is within this range, the robotic arm is controlled to move to the re-shooting position, and the re-shoot is performed when the amplitude does not exceed a preset amplitude threshold. This means that vibration detection is still required before re-shooting to ensure environmental stability during the re-shooting process. If the re-shooting position is not within the current reachable range of the robotic arm, the re-shooting request is recorded in the re-shooting queue, waiting to be processed when the AGV moves to the next measurement point. The re-shooting queue can be a list that records the measurement viewpoint information required for the re-shooting (such as the three-dimensional coordinates of the measurement viewpoint, camera direction, etc.).

[0046] Finally, after the AGV has completed the movement of all preset measurement points and the robotic arm has completed the shooting of all measurement viewpoints, the reshoot queue is checked. If there are still unprocessed reshoot requests in the reshoot queue, it indicates that these reshoot tasks have not been successfully completed. These reshoot requests are marked as abnormal to prompt operators to perform manual intervention and analysis.

[0047] The overall technical solution of the embodiment significantly improves the efficiency and data quality of AGV and robotic arm collaborative 3D measurement through multiple quality control and dynamic adjustment mechanisms before, during and after measurement, reduces the frequency of manual intervention and rework, realizes the intelligence and automation of the measurement process, and has high practical value and technological advancement.

[0048] Therefore, the AGV and robotic arm collaborative three-dimensional measurement method provided in this application suppresses vibration interference during the measurement process through real-time vibration monitoring and fine-tuning mechanisms, and ensures data quality by combining real-time quality assessment and dynamic reshooting strategies. This avoids repeated measurements caused by vibration and delayed quality assessment in traditional solutions, reduces measurement errors caused by vibration when the AGV stops, and avoids multiple reshoots due to unqualified quality, thus effectively improving measurement efficiency.

[0049] In one embodiment, the AGV stops after reaching the measurement point, and the vibration signal is collected by an accelerometer to calculate the amplitude, including: The AGV stops after reaching the measurement point, and vibration signals are collected in real time by a triaxial accelerometer installed on the AGV. The vibration signals include acceleration components in three directions. , where t is time; Calculate the vibration amplitude A(t):

[0050] The vibration amplitude is smoothed by applying a moving average filter. :

[0051] Where N is the length of the sliding window. The sampling period.

[0052] The triaxial accelerometer is configured to detect instantaneous acceleration changes of the AGV along three mutually orthogonal axes in three-dimensional space. In specific implementations, the triaxial accelerometer can be a microelectromechanical system (MEMS) piezoresistive accelerometer, a high-precision capacitive accelerometer, or a piezoelectric multi-axis sensor. It is typically rigidly mounted at the chassis center of gravity of the AGV or at the connecting flange of the robotic arm base to detect multi-degree-of-freedom swaying without omission.

[0053] The vibration amplitude A(t) is the magnitude of the acceleration components in the three directions, which are combined into a scalar representing the total spatial vibration intensity. Moving average filtering is a time-domain smoothing algorithm used to dynamically filter out occasional high-frequency pulse interference, thereby outputting a continuous, stable, and truly reflective smooth vibration amplitude that reflects the overall residual vibration decay process of the system. In its implementation, this moving average filtering can achieve first-in-first-out (FIFO) update calculations by allocating a circular FIFO buffer of length N in memory; alternatively, a recursive moving average algorithm can be used, introducing new sampled values ​​and subtracting historical slip-out values ​​to reduce the processor's computational load and achieve real-time smoothing with extremely low latency.

[0054] The above technical solution, through the spatial modulus calculation of the three-axis acceleration vector, comprehensively covers the composite swaying components of the AGV in each degree of freedom in space, avoiding the omission of vibration assessment due to incomplete unidirectional monitoring; at the same time, by combining moving average filtering to perform time-domain denoising on the synthesized vibration amplitude, the influence of occasional pulse interference on the amplitude over-limit judgment is effectively eliminated, which greatly improves the accuracy and stability of amplitude calculation, and provides a reliable vibration monitoring basis for the accurate determination of the subsequent robotic arm operation timing.

[0055] In one embodiment, the metrics for quality scoring include, but are not limited to, point cloud density. Point cloud noise and coverage integrity ; The quality score Q is a weighted average of all indicators:

[0056] in, The weights of each indicator are given.

[0057] Specifically, point cloud density Point cloud density is used to measure the number of effective 3D points collected within a unit spatial area, reflecting the ability of measurement data to express the microscopic geometric details of the workpiece surface. In specific implementations, point cloud density can be obtained by statistically analyzing the number of points contained per unit surface area or per unit spatial volume element within the measurement area, or it can be characterized by calculating the reciprocal of the average Euclidean distance between adjacent points.

[0058] Point cloud noise This is used to evaluate the degree of distribution of outliers or measurement errors in a measurement point cloud that deviate from the actual workpiece surface. In specific implementations, point cloud noise can be quantified by calculating the average vertical distance or standard deviation of each point cloud to its neighborhood fitting plane, or by statistically determining the proportion of outliers in the total point cloud or the dispersion index.

[0059] Coverage integrity This is used to evaluate the complete coverage of the effective point cloud acquired from the current measurement viewpoint on the preset target measurement area or the geometric features to be measured on the workpiece. In specific implementations, coverage integrity can be evaluated by projecting the acquired point cloud onto the surface of the workpiece's CAD design model and calculating the hit ratio of the model's effective mesh cells, or by calculating the overlap ratio between the point cloud's convex hull and the preset theoretical contour region.

[0060] Indicator weights This is used to balance the importance of point cloud density, point cloud noise, and coverage integrity under different measurement tasks or different workpiece characteristics. Weight values ​​are typically configured as normalized positive values ​​to ensure that the overall quality score is uniformly measured within a standardized scoring scale. In specific implementations, the indicator weights... The values ​​are 0.3, 0.4, and 0.3 respectively.

[0061] The above technical solution integrates structured and quantitative point cloud quality features, avoiding evaluation biases caused by relying on a single quality indicator. This allows for a comprehensive and objective reflection of the actual quality of the 3D measurement data. Furthermore, the deep integration of multi-indicator weighted comprehensive scoring with the motion control logic of the AGV and robotic arm significantly improves the accuracy of triggering re-shoot decisions, ensuring the targeted and effective nature of subsequent re-shoot actions. This effectively avoids invalid re-shoots or missed detections, thereby guaranteeing the high precision and reliability of the overall 3D measurement operation.

[0062] In one embodiment, if the quality score is lower than a preset quality threshold, the location for reshooting is determined, including: If the quality score is lower than the preset quality threshold, first identify the type of quality defect: if the point cloud density is lower than the density threshold, the quality defect is insufficient point cloud density; if the point cloud noise is higher than the noise threshold, the quality defect is excessive point cloud noise; if the coverage integrity is lower than the integrity threshold, the quality defect is incomplete coverage. The location for reshoots was then determined based on the quality defects.

[0063] Regarding the identification of quality defect types, insufficient point cloud density indicates that the number of point cloud points collected per unit area or space is too low. This can be determined by adjusting the point cloud density obtained from the current measurement viewpoint. The calculated value is directly compared with a preset density threshold. For example, the point density or laser line projection density in a local geometric area of ​​the workpiece is statistically analyzed. When the value does not reach the basic density threshold necessary to ensure the reconstruction of geometric features, it can be determined that there is a density defect.

[0064] Excessive point cloud noise indicates that outliers, spikes, or random perturbations in the point cloud deviate from the true surface and exceed the allowable range. This can be determined by calculating the local surface fitting residuals, Gaussian curvature variance, or outlier proportions of the point cloud, and then applying the calculated point cloud noise... If the noise threshold is exceeded, the defect is classified as excessive noise.

[0065] Incomplete coverage indicates the presence of blind spots or missing areas within the expected measurement field of view due to factors such as occlusion or limited viewing angle. This can be determined by comparing the boundary contour or convex hull of the actual measured point cloud with the theoretical geometric range of the target workpiece. If the projection field of view of the point cloud in the area to be measured accounts for a certain percentage (coverage integrity), then the measurement is considered complete. If the content is below the integrity threshold, it is defined as incomplete coverage.

[0066] The above technical solution refines the quality scoring threshold judgment into defect type diagnosis for different indicator dimensions, avoiding blind and untargeted repeated shooting or random search for reshoot poses when measurement data is abnormal. This ensures that the subsequently determined reshoot position can accurately target and eliminate specific quality problems, significantly improving the success rate of reshoot and measurement efficiency, and guaranteeing the integrity and accuracy of 3D model reconstruction during the collaborative measurement process of AGV and robotic arm.

[0067] In one embodiment, if the quality defect is insufficient point cloud density, the measuring head is moved a first distance towards the workpiece, starting from the current measuring head position, while keeping the current measuring direction unchanged, to obtain the re-shooting position.

[0068] Specifically, the first distance refers to the translational amount by which the measuring head moves closer to the workpiece surface along the current optical axis. As a feasible implementation, the first distance can be dynamically calculated based on the current depth range of the optical sensor mounted on the measuring head and a preset point cloud density compensation step size. Alternatively, the first distance can be pre-configured as a fixed safety step distance, set within the safe threshold allowed by the measuring head's working distance. Maintaining the current measurement direction means that during position adjustment, the measuring head's attitude angle and line of sight remain consistent with the spatial orientation at the original measurement viewpoint. This can be manifested by changing only the translational component of the measuring head in the spatial coordinate system while keeping the rotation matrix unchanged, or by adjusting the inverse kinematics of each joint of the robotic arm to allow the end-effector measuring head to perform a pure translational motion along the original line of sight vector.

[0069] The above solution shortens the acquisition distance by directional control of the measuring head towards the workpiece surface when the quality defect is determined to be insufficient point cloud density. Since the field of view of the optical measuring equipment corresponds to a smaller physical coverage area at closer object distances, the number of projected or acquired sampling points per unit surface area significantly increases, thereby fundamentally improving the density of the local geometric point cloud. Simultaneously, by maintaining the current measurement direction, the optimal observation normal and line-of-sight angle planned from the original measurement viewpoint are locked, avoiding point cloud distortion or surface feature occlusion caused by posture deflection. Based on this, the re-shooting position is organically coordinated with the aforementioned robotic arm accessibility and vibration state judgments, ensuring that while the measuring head approaches the workpiece to acquire a high-density point cloud, it can complete high-precision re-shooting operations under conditions of stable robotic arm dynamics and vibration convergence. Through this technical solution, adaptive and accurate pose replanning can be achieved for defect types with insufficient point cloud density. Shortening the object distance effectively improves the sampling resolution and point cloud density of the target area, and maintaining the measurement direction avoids introducing additional projection distortion and occlusion errors, significantly improving the detail reproduction and data integrity of 3D measurements.

[0070] In one embodiment, if the quality defect is excessive point cloud noise, the measuring head is moved a second distance away from the workpiece, starting from the current measuring head position, to obtain the re-shooting position. Then, the measuring head orientation is adjusted to reduce the angle between the measuring head optical axis and the workpiece surface normal vector.

[0071] Specifically, the second distance can be determined based on a preset fixed step size threshold, or it can be obtained through dynamic weighted calculation based on the magnitude of the current point cloud noise exceeding the noise threshold. The workpiece surface normal vector can be calculated by plane fitting of the local surface of the workpiece in the point cloud acquired from the current viewpoint, or it can be obtained by matching the theoretical normal vector of the corresponding measurement area in the preset 3D CAD model of the workpiece. Reducing the angle between the optical axis of the measuring head and the workpiece surface normal vector can be achieved by controlling the rotational degree of freedom of the robotic arm's end effector, making the measuring head's line of sight more directly aligned with the workpiece surface to be measured. In feasible implementations, the optical axis direction... Vector of workpiece surface The included angle decreases, meaning it becomes more perpendicular to the workpiece surface:

[0072] Where β is the orientation adjustment coefficient (0.3~0.8), used to significantly reduce the incident angle. If the noise mainly comes from oblique incidence, a larger value is taken; if the noise mainly comes from too close a distance, a smaller value is taken.

[0073] The above solution, through targeted analysis of the causes of excessive point cloud noise, firstly controls the measuring head to retreat a second distance away from the workpiece when excessive noise is identified. This increases the imaging distance between the measuring head and the workpiece surface, effectively alleviating local reflection saturation and overexposure caused by excessively close measurement distances. Furthermore, the orientation of the measuring head is corrected to reduce the angle between the optical axis and the normal vector of the workpiece surface. This allows the measuring light to be projected and received at an angle closer to perpendicular to the workpiece surface, significantly reducing diffuse reflection distortion and edge scattering clutter introduced during large-angle tilt measurements. The combination of these two methods fundamentally suppresses the noise generation mechanism in optical measurements. This technical solution precisely addresses the specific defect of excessive point cloud noise by implementing a combination of retreating avoidance and normal adjustment for attitude reconstruction. It significantly reduces the negative interference of workpiece surface reflection and tilt scattering on 3D imaging, effectively suppresses outlier noise, and greatly improves the signal-to-noise ratio of point cloud data and the effectiveness of single-shot re-enhancing.

[0074] In one embodiment, if the quality defect is incomplete coverage, the measuring head is moved a third distance away from the workpiece, starting from the current measuring head position, while keeping the current measuring direction unchanged, to obtain the replacement position.

[0075] The third distance, moving away from the workpiece, refers to the translation of the measuring head in the spatial coordinate system in a direction away from the surface of the workpiece, increasing the shooting distance. This can be achieved by linearly retracting the measuring head in the opposite direction of its current optical axis, or by translating and retracting in the opposite direction of the local normal vector of the measured area of ​​the workpiece. The third distance is determined based on the area of ​​missing coverage; the larger the missing area, the larger the third distance. The value of the third distance ranges from 20 to 80 mm. Increasing the distance expands the single-frame field of view of the measuring head, which helps to cover the original missing area.

[0076] Maintaining the current measurement direction means that during the translation adjustment process, the attitude angles (including pitch, yaw, and roll angles) of the measuring head in space remain constant, ensuring that the line of sight of the measuring head remains consistent with the original shooting viewpoint. This can be achieved by locking the attitude quaternion or rotation matrix of the end effector during inverse kinematics planning of the robotic arm, applying offset transformations only to the three-dimensional position coordinates; or by fixing the spatial pointing vector of the robotic arm's end effector measuring head to constrain joint angle adjustments.

[0077] The above scheme, after identifying incomplete coverage defects, retracts a third distance away from the workpiece from the current measuring head position and locks the measurement orientation. Its operating mechanism utilizes the characteristic that the field of view of an optical or 3D sensor expands divergently with increasing object distance. As the measuring head retracts away from the workpiece, the projected field of view formed on the workpiece surface is proportionally magnified, thus covering a wider area of ​​the workpiece surface in a single re-image, including features that were originally outside the edge of the field of view or missed. Simultaneously, maintaining the current measurement orientation ensures that the central reference of the re-image field of view remains stably aligned with the original target area, avoiding new blind spots or field of view deviations caused by posture changes. Through this targeted posture adjustment, re-image points matching the field of view expansion requirements can be quickly generated, directly targeting coverage blind spots for data acquisition. The above technical solution enables rapid and accurate reconstruction of supplementary poses with a wider field of view specifically for defect types with incomplete coverage. This helps eliminate scanning blind spots and edge loss phenomena in 3D measurement. Without the need for complex posture transformation, the measurement field of view is expanded with simple displacement adjustment, which significantly improves the integrity of the point cloud model and measurement efficiency, ensuring the high efficiency and accuracy of collaborative measurement operations between AGV and robotic arm.

[0078] In one embodiment, when there is more than one quality defect type, the reshoot location is determined by prioritizing the quality defect type ranked first, in the order of excessive point cloud noise, incomplete coverage, and insufficient point cloud density.

[0079] Specifically, having more than one quality defect type means that when scoring and identifying the quality of data captured from the same measurement viewpoint, at least two of the following defects are present: high point cloud noise, low coverage integrity, and low point cloud density. The highest-ranked quality defect type is used as the sole decision-making basis for determining the current reshoot position. By establishing a strict defect handling priority, the logic for generating the reshoot posture when multiple defects coexist is constrained. During point cloud measurement, excessive point cloud noise is often caused by factors such as the measurement head being too close to the workpiece surface, high light reflection, or poor incident angle. Strong noise data directly destroys the spatial geometric reference of the point cloud. Incomplete coverage represents the lack of effective geometric structure information within the measurement field of view. Insufficient point cloud density is only manifested as sparse local sampling points. Therefore, excessive point cloud noise is given the highest priority. The measurement head is first adjusted away from the workpiece, and the angle between the optical axis and the normal vector is optimized to suppress noise. Only then is the problem of incomplete coverage addressed by increasing the field of view. Finally, insufficient point cloud density is addressed, thereby ensuring the monotonicity and convergence of the reshoot action decision for each measurement dimension and avoiding conflicts in the direction of the robotic arm's posture adjustment.

[0080] In one embodiment, the AGV and robotic arm collaborative three-dimensional measurement method further includes: after each re-shoot, the quality score of the re-shoot data is scored again; if the quality score of the re-shoot data is lower than a preset quality threshold, the re-shoot position is re-determined and re-shoot is performed until the quality score of the re-shoot data is not lower than the preset quality threshold or the maximum number of re-shoots is reached.

[0081] Specifically, re-evaluating the quality of the re-shot data after each re-shoot refers to performing a quality review and quantitative assessment of the 3D measurement data generated in real time after each re-shoot. This process can be carried out using the method described in the above embodiment, i.e., using point cloud density. Point cloud noise and coverage integrity The quality score is calculated by weighted summation, thus obtaining an instant score value that reflects the current quality of the reshoot.

[0082] If the quality score of the re-enhanced data is lower than the preset quality threshold, re-determining the re-enhancing position and re-enhancing refers to replanning the spatial pose or measurement direction of the measuring head and performing a second measurement when the re-enhanced data fails to meet the standard, based on the type of quality defect. This can be achieved by further fine-tuning the pose offset along the optical axis of the measuring head or the workpiece normal, based on the previous re-enhancing position; or by recalculating the optimal observation angle and distance of the measuring head based on the quality defects of the previous failed acquisition, and driving the robotic arm to move to the new position to re-enhance the image.

[0083] The maximum number of retakes refers to a pre-configured upper limit threshold for closed-loop retrying for the same measurement task, used to constrain repeated retake behavior. The maximum number of retakes can be a pre-configured fixed iteration count. If the maximum number of retakes is reached, it indicates that the required 3D data cannot be obtained, and the retake request is recorded in the retake queue, to be tried again at the next measurement point. If the maximum number of retakes has not been reached, the retake position is determined and it is checked whether the robotic arm can reach it. If so, the robotic arm moves the measuring head to the retake position to perform the retake.

[0084] The above solution establishes an immediate data quality closed-loop verification mechanism after each re-shoot. If the re-shoot data fails to meet the quality threshold, it automatically triggers secondary pose calculation and cyclic re-shooting, while introducing a maximum number of re-shoots as a boundary condition for loop exit. When the re-shoot data quality reaches a preset standard or the retry count reaches its limit, the system automatically exits the re-shooting loop and enters subsequent logic. This mechanism deeply integrates data acquisition, quality evaluation, and secondary pose optimization, enabling the robotic arm's re-shooting actions to have adaptive correction capabilities and avoiding the propagation of measurement errors caused by a single failed re-shoot.

[0085] In one embodiment, the AGV and robotic arm collaborative three-dimensional measurement method further includes: After the robotic arm has completed all the measurement viewpoints of the current measurement site, if the reshoot queue is not empty, the reshoot requirements in the reshoot queue are merged with the measurement viewpoints of the next measurement site into a set. The AGV is controlled to move to the next measurement point, and the vibration signal is collected by the accelerometer and the amplitude is calculated. If the amplitude is not greater than the preset amplitude threshold, the robotic arm's measuring head is first controlled to traverse and process the reshooting requirements in the set; if the reshooting requirements in the set are completed, the corresponding reshooting requirements are deleted from the reshooting queue; otherwise, the corresponding reshooting requirements are retained in the reshooting queue. The robotic arm's measuring head is then moved to various measuring viewpoints within the assembly to take pictures and measurements.

[0086] Specifically, after the robotic arm has traversed all measurement viewpoints at the current measurement site, it determines whether this is the last measurement site. If not, it checks if the replacement queue is not empty. If the replacement queue is not empty, the replacement requests in the queue are merged with the measurement viewpoints for the next measurement site into a single set, and the replacement requests are marked as priority processing. This serves to reconstruct the execution target sequence of the robotic arm at the next measurement site, thereby unifying the management of tasks to be executed. In specific implementations, this set can be constructed as a dynamic linked list or task array structure containing task attribute tags, with replacement requests and regular measurement viewpoints assigned different task identifiers; alternatively, it can be constructed as a double-ended queue or a tree-like task structure supporting priority scheduling, placing replacement requests at the priority execution level for efficient task retrieval and status tracking.

[0087] Before taking pictures and measuring at the next measurement point, the vibration signal is collected by the accelerometer and the amplitude is calculated. The vibration state of the AGV is determined by the amplitude. When the conditions are met, the robotic arm and measuring head are controlled to perform three-dimensional measurement to avoid excessive vibration that would lead to unsatisfactory measurement data.

[0088] The process iterates through the set of retake requests and updates the retake queue. Its purpose is to prioritize retake tasks left over from previous measurement sites and achieve closed-loop task management. Specifically, the robotic arm controller sequentially reads the retake pose data from the set and performs inverse kinematics calculations. If the inverse kinematics solution is within the current joint limits and the motion trajectory is free from collision interference, the retake request is deemed executable, and the robotic arm is moved into position for data acquisition. After acquisition, a confirmation command is sent to dequeue the corresponding item from the retake queue. If a retake is not completed due to unreachability or a low retake quality score, the item is skipped and marked in the retake queue, allowing subsequent measurement sites to attempt it again.

[0089] The above solution encapsulates the backlog of reshooting requests and the inherent viewpoint of the new measurement site into a composite execution set before the AGV moves to the next measurement site. After the AGV arrives and the vehicle vibration convergence meets the standard, a strict hierarchical scheduling logic of "reshooting first, regular requests follow" is adopted. The robotic arm prioritizes calling computing resources and mechanical actions to attempt reshooting tasks within the set. Each successful reshoot immediately updates and simplifies the reshooting queue. After the reshooting status is cleared or confirmed, it smoothly switches to continuous traversal shooting of the new site's own measurement viewpoint. This dynamic merging and priority clearing mechanism ensures state consistency during cross-site task flow, tightly connecting AGV relocation actions with robotic arm reshooting actions, and eliminating redundant path planning caused by isolated reshooting.

[0090] This application provides an electronic device, such as... Figure 3 As shown, it includes: a processor 21 and a memory 22; wherein, the memory 22 stores a computer program 23, which is adapted to be loaded by the processor 21 and executed by the AGV and robotic arm collaborative three-dimensional measurement method of one or more embodiments above.

[0091] Here, electronic device 20 refers to a hardware system capable of processing information and executing instructions. Electronic device 20 can be a main controller located inside the AGV, used to control the movement of the AGV, the movement of the robotic arm, and the taking of images by the measuring head.

[0092] Processor 21 is the core computing unit of electronic device 20, responsible for interpreting and executing instructions in computer programs, and performing data processing and control operations. Processor 21 can be a central processing unit (CPU), such as a processor based on x86 architecture or ARM architecture, or it can be a microcontroller (MCU).

[0093] Memory 22 is a hardware component used to store data and computer programs. Memory 22 may include random access memory (RAM) for temporary storage of runtime data and program instructions, and read-only memory (ROM) or flash memory for long-term storage of the operating system, firmware, and applications. As one implementation, memory 22 may use DDR SDRAM as the main memory, supplemented by NAND Flash as non-volatile memory. In another implementation, memory 22 may be integrated inside the processor, such as through on-chip SRAM and on-chip Flash.

[0094] Computer program 23 is a set of instructions used to direct the processor to perform specific tasks. Computer program 23 can be pre-programmed into memory 22 as firmware, or it can run as application software on an operating system. As one implementation, computer program 23 can be an embedded application written in C / C++ that interacts directly with the hardware. In another implementation, computer program 23 can be test management software developed using high-level languages ​​such as Python or Java, running on top of an operating system.

[0095] Loading and execution refers to the process by which processor 21 reads the instructions of computer program 23 from memory 22 and processes and operates them according to the logical order of the instructions. This typically involves the coordinated work of internal processor components such as the program counter, instruction register, and arithmetic logic unit. In one implementation, processor 21 loads a bootloader from non-volatile memory 22 at startup, and then the bootloader loads the measurement method control program into RAM and begins execution. In another implementation, the operating system is responsible for loading the measurement method control program from the file system into memory and scheduling the processor to execute the instructions of that application.

[0096] This application provides a computer storage medium that stores multiple instructions, which are adapted to be loaded by a processor and executed by the AGV and robotic arm collaborative three-dimensional measurement method of one or more of the above embodiments.

[0097] Computer storage media refers to a physical device capable of storing digital data or computer instructions. Examples of computer storage media include, but are not limited to, hard disk drives (HDDs), solid-state drives (SSDs), flash memory (such as USB flash drives and SD cards), phase-change memory (PCRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transfer medium.

[0098] The purpose of computer storage media is to provide persistent or temporary storage space for computer programs, allowing program code and data to be accessed and used by the processor when needed. Computer storage media stores multiple instructions, which are collections of computer program code stored in binary form. These instructions represent a logical sequence of actions to implement a specific function or perform a specific task. These instructions can be a complete executable program or part of a program, such as a library file or script. Instructions are designed to be loaded and executed by the processor, meaning that the instructions stored on the medium can be read into the computer's internal memory (such as cache or registers), and then decoded and executed by the processor according to the logical order of the instructions. The processor executes these instructions to control the computer hardware to perform various operations, thereby realizing the various steps and functions defined in the 3D measurement method.

[0099] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus, and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0100] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0101] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0102] Those skilled in the art will understand that embodiments of this application can be provided as methods, apparatus, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0103] It should be noted that, due to space limitations, this application specification does not exhaustively list all possible implementation methods. Those skilled in the art should be able to conceive after reading this application specification that, as long as the technical features do not contradict each other, any combination of technical features can constitute an optional implementation method.

[0104] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0105] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A collaborative three-dimensional measurement method using an AGV and a robotic arm, characterized in that, It includes: The AGV stops after reaching the measurement point, and the vibration signal is collected by the acceleration sensor to calculate the amplitude. If the amplitude exceeds the preset amplitude threshold and continues for more than the preset time, the AGV will be controlled to make a fine adjustment of its position or retract the robotic arm to lower the center of gravity, and then the vibration signal will be collected again to determine whether the amplitude exceeds the limit. If the amplitude is not greater than the preset amplitude threshold, the robotic arm is controlled to move the measuring head sequentially to each measuring viewpoint contained in the current measuring point according to the motion path to take pictures. After each measurement viewpoint is captured, the captured data is scored for quality; if the quality score is lower than the preset quality threshold, a reshoot location is determined. Determine whether the replacement shooting position is within the reach of the robotic arm at the current measurement point of the AGV: if yes, control the robotic arm to reach the replacement shooting position and perform replacement shooting when the amplitude is not greater than the preset amplitude threshold; if no, record the replacement shooting requirement in the replacement shooting queue and perform replacement shooting at the next measurement point of the AGV. Once the AGV has completed all measurement points and the robotic arm has completed all measurement viewpoints, if the reshoot queue is not empty, then the reshoot requirement in the reshoot queue will be marked as abnormal.

2. The AGV and robotic arm collaborative three-dimensional measurement method as described in claim 1, characterized in that, After the controlled AGV reaches the measurement point, it stops, and the vibration signal is collected by the acceleration sensor to calculate the amplitude, including: The AGV stops after reaching the measurement point, and vibration signals are collected in real time by a triaxial accelerometer installed on the AGV. The vibration signals include acceleration components in three directions. , where t is time; Calculate the vibration amplitude A(t): The vibration amplitude is smoothed by applying a moving average filter. : Where N is the length of the sliding window. The sampling period.

3. The AGV and robotic arm collaborative three-dimensional measurement method as described in claim 1, characterized in that, The quality scoring metrics include, but are not limited to: point cloud density. Point cloud noise and coverage integrity ; The quality score Q is a weighted average of all indicators: in, The weights of each indicator are given.

4. The AGV and robotic arm collaborative three-dimensional measurement method as described in claim 3, characterized in that, If the quality score is lower than a preset quality threshold, the method for determining the reshoot location includes: If the quality score is lower than the preset quality threshold, first identify the type of quality defect: if the point cloud density is lower than the density threshold, the quality defect is insufficient point cloud density; if the point cloud noise is higher than the noise threshold, the quality defect is excessive point cloud noise; if the coverage integrity is lower than the integrity threshold, the quality defect is incomplete coverage. The location for reshoots was then determined based on the quality defects.

5. The AGV and robotic arm collaborative three-dimensional measurement method as described in claim 4, characterized in that, If the quality defect is insufficient point cloud density, take the current measuring head position as the starting point, move a first distance towards the workpiece, keep the current measuring direction unchanged, and obtain the re-shooting position.

6. The AGV and robotic arm collaborative three-dimensional measurement method as described in claim 4, characterized in that, If the quality defect is excessive point cloud noise, take the current measuring head position as the starting point, move a second distance away from the workpiece to obtain the retake position, and then adjust the orientation of the measuring head to reduce the angle between the measuring head optical axis and the normal vector of the workpiece surface.

7. The AGV and robotic arm collaborative three-dimensional measurement method as described in claim 4, characterized in that, If the quality defect is incomplete coverage, take the current measuring head position as the starting point, move a third distance away from the workpiece, keep the current measuring direction unchanged, and obtain the position for re-taking.

8. The AGV and robotic arm collaborative three-dimensional measurement method as described in claim 4, characterized in that, When there is more than one quality defect type, the reshoot location is determined by prioritizing the quality defect type ranked first, in the order of excessive point cloud noise, incomplete coverage, and insufficient point cloud density.

9. The AGV and robotic arm collaborative three-dimensional measurement method as described in claim 1, characterized in that, Also includes: After each reshoot, the quality score of the reshoot data is calculated again. If the quality score of the reshoot data is lower than the preset quality threshold, the reshoot location is determined and reshoot is performed again until the quality score of the reshoot data is not lower than the preset quality threshold or the maximum number of reshoots is reached.

10. An electronic device, characterized in that, It includes a processor and a memory; wherein the memory stores a computer program adapted to be loaded by the processor and executed as described in any one of claims 1-9, the AGV and robotic arm collaborative three-dimensional measurement method.