Surface non-contact measurement trajectory planning method based on Cartesian and cylindrical coordinate system

Abstract

The invention discloses a surface non-contact measurement trajectory planning method based on a Cartesian and cylindrical coordinate system, belongs to the technical field of measurement trajectory planning, and aims to solve the problems that in the prior art, a scanning mode is single, mixed features of a plane and a curved surface cannot be self-adapted, and the measurement precision is low. And therefore, the problems of non-optimal measurement path and insufficient identification sensitivity of the surface gradient defect can be solved. The method comprises the following steps: carrying out laser scanning on the surface of a mold by using pulse type scanning and triangular ranging type scanning to obtain scanning data; recognizing the surface of the mold according to the scanning data, constructing a mathematical model of the whole surface of the mold, and constructing a coordinate system; performing global path identification and planning on the mathematical model of the mold; scanning the surface of the mold according to the planned simulation path, judging defect points by adopting a threshold method, and confirming the positions of the defect points; and analyzing the corresponding coordinates and the relative coordinates corresponding to the defect points according to the positions of the defect points, and further calculating the optimal path trajectory between the defect points.

Description

Technical Field

[0001] This invention belongs to the field of measurement trajectory planning technology, specifically relating to a surface non-contact measurement trajectory planning method based on Cartesian and cylindrical coordinate systems. Background Technology

[0002] As a key forming tool in industrial production, the surface quality of molds directly affects the precision and performance of products. Traditional contact measurement (such as coordinate measuring machines) has inherent drawbacks such as slow measurement speed and easy scratching of workpiece surfaces. In recent years, although non-contact measurement technologies (such as laser scanning and structured light measurement) have been gradually applied, they still face the following technical bottlenecks in the inspection of complex curved surface molds: a single scanning mode is difficult to balance macroscopic efficiency and microscopic accuracy; pulse scanning is fast over long distances but lacks accuracy; triangulation has high accuracy but a limited measurement range; existing path planning methods are mostly based on a single coordinate system, either pure Cartesian or pure cylindrical coordinates, which cannot adapt to the mixed characteristics of planes and curved surfaces, resulting in non-optimal measurement paths; defect recognition algorithms generally use fixed threshold methods, which are not sensitive enough to the recognition of surface gradient defects, such as micro-dimples. Summary of the Invention

[0003] The purpose of this invention is to propose a surface non-contact measurement trajectory planning method based on Cartesian and cylindrical coordinate systems, which solves the problems of existing technologies such as single scanning mode, inability to adapt to mixed features of plane and curved surfaces, resulting in non-optimal measurement paths and insufficient sensitivity in identifying surface gradient defects.

[0004] To achieve the above objectives, the surface non-contact measurement trajectory planning method based on Cartesian and cylindrical coordinate systems of the present invention includes the following steps:

[0005] Step 1: Use pulse scanning and triangulation scanning to perform laser scanning on the mold surface to obtain scanning data;

[0006] Step 2: Based on the scanning data obtained in Step 1, identify the surface of the mold, construct a mathematical model of the overall mold surface, and establish a coordinate system;

[0007] Step 3: Perform global path identification and planning on the mathematical model of the mold obtained in Step 2;

[0008] Step 4: Scan the mold surface according to the simulation path planned in Step 3, use the threshold method to identify defect points, and confirm the location of defect points;

[0009] Step 5: Based on the defect point locations obtained in Step 4, analyze the corresponding coordinates and relative coordinates of the defect points, and further calculate the optimal path trajectory between the defect points.

[0010] The step one, which involves using pulse scanning and triangulation scanning to obtain scanning data on the mold surface, specifically involves: rapidly acquiring large-scale contour data of the mold surface through pulse scanning, and accurately acquiring local high-precision data through triangulation scanning.

[0011] Step two, which involves identifying the surface of the mold, constructing a mathematical model of the entire mold surface, and establishing a coordinate system, specifically includes:

[0012] First, a point cloud is constructed using the data obtained from triangulation scanning. The three-dimensional coordinates are calculated by measuring the distance d, using the following formula:

[0013]

[0014] Where: d is the measured distance;

[0015] c is the speed of light;

[0016] t is the round-trip time of the laser.

[0017] The distance data d is converted into three-dimensional coordinate data (x, y, z) through trigonometric functions, thereby constructing a three-dimensional point cloud model;

[0018] The 3D point cloud model is further constructed and its extent is planned using data obtained from pulse scanning, specifically including:

[0019] Use pulsed scan data to supplement missing data in distant or low-precision areas of point cloud models;

[0020] The data from the two scanning methods are aligned and integrated using a data fusion algorithm to form a complete point cloud model of the mold surface.

[0021] Denoising and smoothing are performed on point cloud models to improve model quality;

[0022] Distinguishing between planar and curved regions:

[0023] The calculation is based on the RMSE (Root Mean Square Error) algorithm and the position data (i.e., 3D coordinate data) obtained from the scan. The specific steps are as follows:

[0024] Divide the point cloud model into several local regions;

[0025] For each local region, fit an ideal plane and calculate the root mean square error (RMSE) of the distances from all points within that region to the fitted plane. The formula is as follows:

[0026]

[0027]

[0028] Set threshold Trmes =0.5, if the difference between the RMSE of a certain region and the average RMSE of other regions is greater than or equal to T. rmse If the condition is met, the region is identified as a curved surface region; otherwise, it is a planar region.

[0029] For curved regions, a cylindrical coordinate system is constructed in the corresponding region of its 3D point cloud; for planar regions, a 2D Cartesian coordinate system is constructed; finally, a mathematical model of the entire mold surface is obtained, and the coordinate system type of each region is labeled.

[0030] Step three, which involves global path identification and planning of the mathematical model of the mold obtained in step two, specifically includes:

[0031] Select the scanning path mode according to the region type: use the grid scanning path in the Cartesian coordinate system for planar regions, and use the spiral scanning path in the cylindrical coordinate system for curved regions.

[0032] The global path planning uses a greedy algorithm, prioritizing the scanning of high-risk areas;

[0033] The goal of path planning is to minimize the total scan time and ensure full coverage, achieving a balance between efficiency and accuracy by dynamically adjusting the scan step size and speed.

[0034] The defect identification described in step four specifically involves:

[0035] Let the point cloud data be

[0036] Where: P i = (x i y i , z i ), x i y i Let z be the plane coordinates of the point. i Representative point P i Height;

[0037] Calculate the average height of all points: P j

[0038]

[0039] For any point P j Calculate the deviation of its height from the average height:

[0040]

[0041] like Then P j Points are identified as defect points, where T is a preset threshold.

[0042] Based on threshold discrimination, further threshold discrimination based on neighborhood analysis is performed;

[0043] For point Find its field Calculate the average height within the neighborhood. ;

[0044]

[0045] Where: m is the number of points in the neighborhood;

[0046] Calculation points The deviation of the height from the neighborhood average height:

[0047]

[0048] like ,but Determined as a defect point, T N The threshold for neighborhood analysis;

[0049] Finally, the coordinates of all defect points are output.

[0050] Step five involves analyzing the corresponding coordinates and relative coordinates of the defect points obtained in step four, and further calculating the optimal path trajectory between the defect points; specifically:

[0051] Extract the coordinates of all defect points and convert them to a unified coordinate system based on their location.

[0052] The Traveling Salesman Problem (TSP) optimization algorithm is used to calculate the shortest access path between defect points, ensuring that a detailed scan of all defect points is completed in the shortest time.

[0053] By combining the surface constraints of the mold, Dijkstra's algorithm is used to adjust the local path to avoid collisions and ensure scanning accuracy.

[0054] The final optimal path trajectory is generated, including the scanning order, movement path, and scanning parameters.

[0055] The beneficial effects of this invention are as follows: The surface non-contact measurement trajectory planning method based on Cartesian and cylindrical coordinate systems of this invention balances efficiency and accuracy through pulse and triangulation composite scanning, adapting to the measurement needs of complex curved surface molds; the path planning for mixed planar and curved surface features is optimized based on RMSE-based region discrimination and adaptive coordinate system construction; the defect identification method combining global threshold and neighborhood analysis improves the detection sensitivity of gradient defects; and the measurement time is minimized and the coverage is maximized through the collaborative planning of global path and local path of defect point. Attached Figure Description

[0056] Figure 1 This is an overall flowchart of the surface non-contact measurement trajectory planning method based on the Cartesian and cylindrical coordinate system of the present invention;

[0057] Figure 2 This is an illustration of the optimal path analysis for defect points in the surface non-contact measurement trajectory planning method based on Cartesian and cylindrical coordinate systems of the present invention. Detailed Implementation

[0058] The embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0059] The surface non-contact measurement trajectory planning method based on Cartesian and cylindrical coordinate systems of the present invention includes the following steps:

[0060] Step 1: Use pulse scanning and triangulation scanning to perform laser scanning on the mold surface to obtain scanning data;

[0061] Step 2: Based on the scanning data obtained in Step 1, identify the surface of the mold, construct a mathematical model of the overall mold surface, and establish a coordinate system;

[0062] Step 3: Perform global path identification and planning on the mathematical model of the mold obtained in Step 2;

[0063] Step 4: Scan the mold surface according to the simulation path planned in Step 3, use the threshold method to identify defect points, and confirm the location of defect points;

[0064] Step 5: Based on the defect point locations obtained in Step 4, analyze the corresponding coordinates and relative coordinates of the defect points, and further calculate the optimal path trajectory between the defect points.

[0065] The step one, which involves using pulse scanning and triangulation scanning to obtain scanning data on the mold surface, specifically involves: rapidly acquiring large-scale contour data of the mold surface through pulse scanning, and accurately acquiring local high-precision data through triangulation scanning, thereby achieving complementarity between macroscopic and microscopic data.

[0066] To achieve efficient and accurate non-contact measurement trajectory planning for mold surfaces, this embodiment selects laser scanning as the core scanning method after comprehensively considering the characteristics of various scanning technologies. Laser scanning, with its significant advantages of high precision, high speed, and non-contact operation, can acquire accurate measurement data without damaging the mold surface.

[0067] To ensure the high authenticity and stability of the acquired scanning data, this step innovatively employs both pulse scanning and triangulation scanning techniques. Pulse scanning utilizes the principle of high-energy laser pulse emission and reception, with pulse widths typically in the nanosecond range, such as the common 5-10 nanoseconds. This enables rapid measurement over long distances, with an effective measurement distance of tens of meters, making it extremely efficient for acquiring macroscopic features of mold surfaces. Triangulation scanning, based on the principle of optical triangulation, calculates distance by precisely measuring the angle change formed after the laser beam reflects off the object's surface. Its measurement accuracy can reach sub-millimeter level, such as 0.1-0.5 millimeters, allowing for the delicate capture of microscopic details on the mold surface.

[0068] By employing two scanning methods in tandem, the mold surface can be scanned comprehensively and at multiple levels, laying a solid foundation for accurately acquiring data such as Cartesian and polar coordinates in subsequent processes. In the subsequent data processing stage, the high-quality scanned data allows for precise calculation of the coordinates of each point on the mold surface in both Cartesian and polar coordinate systems, providing reliable data support for mold design optimization and quality inspection.

[0069] Step two, which involves identifying the surface of the mold, constructing a mathematical model of the entire mold surface, and establishing a coordinate system, specifically includes:

[0070] Based on the composite scanning method in step one, the data obtained from scanning the mold surface is analyzed. First, a point cloud is constructed using data obtained from triangulation scanning (i.e., using a triangulation rangefinder). Triangulation scanning calculates the three-dimensional coordinates by measuring distance d, using the following formula:

[0071]

[0072] Where: d is the measured distance;

[0073] c is the speed of light;

[0074] t is the round-trip time of the laser.

[0075] The distance data d is converted into three-dimensional coordinate data (x, y, z) through trigonometric functions, thereby constructing a three-dimensional point cloud model;

[0076] The data obtained from pulse scanning (i.e., pulse scanner) is used to further construct and define the extent of the 3D point cloud model, specifically including:

[0077] Use pulsed scan data to supplement missing data in distant or low-precision areas of point cloud models;

[0078] Data from the two scanning methods are aligned and integrated using a data fusion algorithm (such as the ICP algorithm) to form a complete point cloud model of the mold surface.

[0079] Denoising and smoothing are performed on point cloud models to improve model quality;

[0080] Distinguishing between planar and curved regions:

[0081] The calculation is based on the RMSE (Root Mean Square Error) algorithm and the position data (i.e., 3D coordinate data) obtained from the scan. The specific steps are as follows:

[0082] Divide the point cloud model into several local regions;

[0083] For each local region, fit an ideal plane and calculate the root mean square error (RMSE) of the distances from all points within that region to the fitted plane. The formula is as follows:

[0084]

[0085]

[0086] Set threshold T rmes =0.5, if the difference between the RMSE of a certain region and the average RMSE of other regions is greater than or equal to T. rmse If the condition is met, the region is identified as a curved surface region; otherwise, it is a planar region.

[0087] For curved regions, a cylindrical coordinate system (based on the principal axes of the surface) is constructed in the corresponding region of its 3D point cloud; for planar regions, a two-dimensional Cartesian coordinate system (based on the plane normal vector) is constructed; finally, a mathematical model of the entire mold surface is obtained, and the coordinate system type of each region is labeled.

[0088] Step three, which involves global path identification and planning of the mathematical model of the mold obtained in step two, specifically includes:

[0089] Select the scan path mode according to the region type (planar or curved): planar regions use a raster scan path in Cartesian coordinates, and curved regions use a spiral scan path in cylindrical coordinates;

[0090] The global path planning uses a greedy algorithm, prioritizing the scanning of high-risk areas (such as areas with large curvature changes and areas with a high incidence of historical defects).

[0091] The goal of path planning is to minimize the total scan time and ensure full coverage, achieving a balance between efficiency and accuracy by dynamically adjusting the scan step size and speed.

[0092] The defect identification described in step four specifically involves:

[0093] Let the point cloud data be

[0094] Where: P i = (xi y i , z i ), x i y i Let z be the plane coordinates of the point. i Representative point P i Height;

[0095] Calculate the average height of all points: P j

[0096]

[0097] For any point P j Calculate the deviation of its height from the average height:

[0098]

[0099] like P j Points are identified as defect points, where T is a preset threshold.

[0100] Based on threshold discrimination, further threshold discrimination based on neighborhood analysis is performed;

[0101] For point Calculate the average height within the neighborhood.

[0102]

[0103] Where: m is the number of points in the neighborhood;

[0104] Calculation points The deviation of the height from the average height of the neighborhood;

[0105]

[0106] like ,but Determined as a defect point, T N The threshold for neighborhood analysis;

[0107] Finally, the coordinates of all defect points are output.

[0108] Step five involves analyzing the corresponding coordinates and relative coordinates of the defect points obtained in step four, and further calculating the optimal path trajectory between the defect points; specifically:

[0109] Extract the coordinates (Cartesian or cylindrical coordinates) of all defect points and convert them to a unified coordinate system (global Cartesian coordinate system or global cylindrical coordinate system) based on their location.

[0110] The Traveling Salesman Problem (TSP) optimization algorithm is used to calculate the shortest access path between defect points, ensuring that a detailed scan of all defect points is completed in the shortest time.

[0111] By combining mold surface constraints (such as obstacles and curvature changes), local path adjustment is performed using the Dijkstra algorithm to avoid collisions and ensure scanning accuracy.

[0112] The optimal path trajectory is finally generated, including scanning order, movement path and scanning parameters (such as speed and laser power).

[0113] Based on the defect points identified by scanning, the corresponding coordinates and relative coordinates of the defect points are analyzed, and the optimal path trajectory between defect points is further calculated to facilitate the location of defect points and the rapid processing of consecutive defect points in subsequent processing.

Claims

1. A surface non-contact measurement trajectory planning method based on Cartesian and cylindrical coordinate systems, characterized in that, Includes the following steps: Step 1: Use pulse scanning and triangulation scanning to perform laser scanning on the mold surface to obtain scanning data; Step 2: Based on the scanning data obtained in Step 1, identify the surface of the mold, construct a mathematical model of the overall mold surface, and establish a coordinate system; Step 3: Perform global path identification and planning on the mathematical model of the mold obtained in Step 2; Step 4: Scan the mold surface according to the simulation path planned in Step 3, use the threshold method to identify defect points, and confirm the location of defect points; Step 5: Based on the defect point locations obtained in Step 4, analyze the corresponding coordinates and relative coordinates of the defect points, and further calculate the optimal path trajectory between the defect points.

2. The surface non-contact measurement trajectory planning method based on Cartesian and cylindrical coordinate systems according to claim 1, characterized in that, The step one, which involves using pulse scanning and triangulation scanning to obtain scanning data on the mold surface, specifically involves: rapidly acquiring large-scale contour data of the mold surface through pulse scanning, and accurately acquiring local high-precision data through triangulation scanning.

3. The surface non-contact measurement trajectory planning method based on Cartesian and cylindrical coordinate systems according to claim 1, characterized in that, Step two, which involves identifying the surface of the mold, constructing a mathematical model of the entire mold surface, and establishing a coordinate system, specifically includes: First, a point cloud is constructed using the data obtained from triangulation scanning. The three-dimensional coordinates are calculated by measuring the distance d, using the following formula: ； Where: d is the measured distance; c is the speed of light; t is the round-trip time of the laser. The distance data d is converted into three-dimensional coordinate data (x, y, z) through trigonometric functions, thereby constructing a three-dimensional point cloud model; The 3D point cloud model is further constructed and its extent is planned using data obtained from pulse scanning, specifically including: Use pulsed scan data to supplement missing data in distant or low-precision areas of point cloud models; The data from the two scanning methods are aligned and integrated using a data fusion algorithm to form a complete point cloud model of the mold surface. Denoising and smoothing are performed on point cloud models to improve model quality; Distinguishing between planar and curved regions: The calculation is based on the RMSE algorithm and the location data obtained from the scan. The specific steps are as follows: Divide the point cloud model into several local regions; For each local region, fit an ideal plane and calculate the root mean square error (RMSE) of the distances from all points within that region to the fitted plane. The formula is as follows: ； ； Set threshold T rmes =0.5, if the difference between the RMSE of a certain region and the average RMSE of other regions is greater than or equal to T. rmse If the condition is met, the region is identified as a curved surface region; otherwise, it is a planar region. For curved regions, a cylindrical coordinate system is constructed in the corresponding region of its 3D point cloud; for planar regions, a 2D Cartesian coordinate system is constructed; finally, a mathematical model of the entire mold surface is obtained, and the coordinate system type of each region is labeled.

4. The surface non-contact measurement trajectory planning method based on Cartesian and cylindrical coordinate systems according to claim 1, characterized in that, Step three, which involves global path identification and planning of the mathematical model of the mold obtained in step two, specifically includes: Select the scanning path mode according to the region type: use the grid scanning path in the Cartesian coordinate system for planar regions, and use the spiral scanning path in the cylindrical coordinate system for curved regions. The global path planning uses a greedy algorithm, prioritizing the scanning of high-risk areas; The goal of path planning is to minimize the total scan time and ensure full coverage, achieving a balance between efficiency and accuracy by dynamically adjusting the scan step size and speed.

5. The surface non-contact measurement trajectory planning method based on Cartesian and cylindrical coordinate systems according to claim 1, characterized in that, The defect identification described in step four specifically involves: Let the point cloud data be ; Where: P i = (x i y i , z i ), x i y i Let z be the plane coordinates of the point. i Representative point P i Height; Calculate the average height of all points: P j ； For any point P j Calculate the deviation of its height from the average height: ； like Then P j Points are identified as defect points, where T is a preset threshold. Based on threshold discrimination, further threshold discrimination based on neighborhood analysis is performed; For point Find its field Calculate the average height within the neighborhood. ; ； Where: m is the number of points in the neighborhood; Calculation points The deviation of the height from the neighborhood average height: ； like ,but Determined as a defect point, T N The threshold for neighborhood analysis; Finally, the coordinates of all defect points are output.

6. The surface non-contact measurement trajectory planning method based on Cartesian and cylindrical coordinate systems according to claim 1, characterized in that, Step five involves analyzing the corresponding coordinates and relative coordinates of the defect points obtained in step four, and further calculating the optimal path trajectory between the defect points; specifically: Extract the coordinates of all defect points and convert them to a unified coordinate system based on their location. The Traveling Salesman Problem optimization algorithm is used to calculate the shortest access path between defect points, ensuring that a detailed scan of all defect points is completed in the shortest time. By combining the surface constraints of the mold, Dijkstra's algorithm is used to adjust the local path to avoid collisions and ensure scanning accuracy. The final optimal path trajectory is generated, including the scanning order, movement path, and scanning parameters.

Images