Online path planning method and system for automatic grinding robot

Abstract

According to the online path planning method and system for the automatic polishing robot, the one-to-one mapping relation between the plane and the curved surface is established through orthographic projection, the complex curved surface is locally segmented into the simple curved surface, the algorithm time complexity of searching the coordinates of the path points on the surface of the model is reduced, the running speed is increased, and the real-time path planning requirement is met; according to the method, path planning can be directly carried out on point cloud data, the requirements for reconstruction algorithm precision and model quality are low, three-dimensional path planning can be converted into two-dimensional plane path planning, a two-dimensional path in any shape can be generated, and automatic and efficient online path planning of a robot is ensured.

Description

Technical Field

[0001] The present invention relates to the field of surface processing, and in particular to an online path planning method and system for an automatic polishing robot. Background Art

[0002] In the manufacturing industry, automation technology for workpiece surface processing is a pressing issue. In automotive manufacturing, only paint inspection and polishing processes still require manual work. The difficulty in automating these two processes lies in obtaining accurate three-dimensional coordinates of the workpiece surface, generating a coordinate sequence based on processing requirements, and guiding the robot to circulate along this coordinate sequence to achieve path planning for the workpiece surface. In automated polishing, the robot must traverse the polishing area along a non-repetitive path, effectively completing Coverage Path Planning (CPP).

[0003] In the past, offline planning was mainly used to construct a digital model of the workpiece using 3D modeling software. By importing surface analytical functions, the grinding path was calculated on the idealized model surface. However, there was a gap between the model and the actual workpiece, and the generated path needed to be manually adjusted, which could not meet the accuracy and efficiency requirements of the grinding process. The line planning method uses a visual camera or 3D scanning device to reconstruct the 3D model of the workpiece and automatically generate a coordinate sequence for the machining path on the digital model of the surface represented by a point cloud or mesh. Due to the disorder of the 3D point cloud and the independence of the point cloud index from the coordinates, the positioning of each path point coordinate requires traversing all vertices. In addition, the surface machining path must adhere closely to the surface. Not only must the distance between the path and the surface and the concavity and convexity of the surface be determined, but the relative posture changes of the model must also be considered. Therefore, automatic path planning on the surface of the point cloud model is quite difficult and challenging. Existing online path planning methods mainly rely on surface fitting, optimization search, and 2D coordinate UV unfolding. They are time-consuming and difficult to achieve the requirements of real-time accurate planning. Summary of the Invention

[0004] The purpose of the present invention is to provide an online path planning method and system for an automatic polishing robot, which obtains workpiece point cloud data and forms a workpiece point cloud model; segments and extracts a number of rectangular areas from the workpiece point cloud model; divides all rectangular areas into a grid map according to the robot polishing path parameters; creates rules to generate a two-dimensional path that traverses the grid map; collects and analyzes the workpiece surface image to obtain the boundary of the area to be polished on the workpiece surface; adjusts the path points of the two-dimensional path according to the boundary of the area to be polished; projects the two-dimensional path onto the point cloud surface to obtain a three-dimensional path; optimizes the path points of the three-dimensional path to obtain the actual polishing path of the robot on the workpiece surface, establishes a one-to-one mapping relationship between planes and surfaces through orthographic projection, locally segments the complex surface into simple surfaces, reduces the algorithm time complexity of searching the model surface path point coordinates, improves the running speed to meet the real-time path planning requirements, can directly perform path planning on the point cloud data, has low requirements on the reconstruction algorithm accuracy and model quality, can convert the three-dimensional path planning into two-dimensional plane path planning, generates a two-dimensional path of any shape, and ensures automatic and efficient online path planning of the robot.

[0005] The present invention is achieved through the following technical solutions: The online path planning method of the automatic polishing robot includes: Acquire workpiece point cloud data to form a workpiece point cloud model; segment the workpiece point cloud model to extract a plurality of rectangular areas; Divide all rectangular areas into a grid map based on the robot polishing path parameters; create rules to generate a two-dimensional path traversing the grid map; Collecting and analyzing the workpiece surface image to obtain the boundary of the area to be polished on the workpiece surface; adjusting the path points of the two-dimensional path according to the boundary of the area to be polished; The two-dimensional path is projected onto a point cloud surface to obtain a three-dimensional path; the path points of the three-dimensional path are optimized to obtain an actual grinding path of the robot on the workpiece surface.

[0006] Optionally, obtaining workpiece point cloud data to form a workpiece point cloud model; segmenting and extracting a plurality of rectangular areas from the workpiece point cloud model includes: Performing a deep scan and photographing of a workpiece to obtain point cloud data of the workpiece; performing pre-processing on the point cloud data by screening and removing abnormal data; and modeling the workpiece using the point cloud data to form a workpiece point cloud model; The workpiece point cloud model is subjected to normal clustering segmentation to obtain a plurality of curved surface regions; and each curved surface region is fitted into a rectangular region on a projection plane using principal component analysis.

[0007] Optionally, all rectangular areas are divided into a grid map according to the robot polishing path parameters; and rules are created to generate a two-dimensional path traversing the grid map, including: Divide all rectangular areas into a grid map according to the step size of the robot's grinding path, and each grid in the grid map is indexed by a two-dimensional coordinate (u, v); A zigzag or I-shaped path rule is created, and based on the path rule, a two-dimensional path that traverses the grid map is generated; the two-dimensional path is stored in a format of two mutually queryable arrays, one of which is a one-dimensional array of path point coordinates indexed by the path sequence, and the other is a two-dimensional array of sequence numbers in the path indexed by the coordinates.

[0008] Optionally, collecting and analyzing a workpiece surface image to obtain a boundary of a to-be-polished area on the workpiece surface; and adjusting the path points of the two-dimensional path according to the boundary of the to-be-polished area include: Acquire a workpiece surface image with the normal direction of the rectangular area as the shooting optical axis, perform image segmentation, contour extraction and convex polygon fitting on the workpiece surface image, and obtain the boundary of the area to be polished on the workpiece surface; The coordinates of the contour line of the boundary of the area to be polished are normalized and transformed into the coordinate system of the grid map, the path points outside the contour line in the two-dimensional path are identified and filtered out, and then all the path points of the two-dimensional path are reordered.

[0009] Optionally, projecting the two-dimensional path onto a point cloud surface to obtain a three-dimensional path; and optimizing the path points of the three-dimensional path to obtain an actual grinding path of the robot on the workpiece surface, including: Projecting the two-dimensional path onto a point cloud surface, thereby transforming the path point coordinates of the two-dimensional path into three-dimensional coordinates, thereby obtaining a three-dimensional path; The coordinates of the path points of the three-dimensional path are interpolated and / or filled in to obtain the actual grinding path of the robot on the workpiece surface.

[0010] Automatic polishing robot online path planning system, including: Point cloud model building module, used to obtain workpiece point cloud data and form a workpiece point cloud model; A rectangular region segmentation module is used to segment the workpiece point cloud model and extract a plurality of rectangular regions; The grid map generation module is used to divide all rectangular areas into grid maps according to the robot polishing path parameters; a two-dimensional path generation module, configured to create rules for generating a two-dimensional path traversing the grid map; The boundary determination module is used to collect and analyze the workpiece surface image to obtain the boundary of the area to be polished on the workpiece surface; A two-dimensional path adjustment module, configured to adjust the path points of the two-dimensional path according to the boundary of the area to be polished; The three-dimensional path generation and optimization module is used to project the two-dimensional path onto the point cloud surface to obtain a three-dimensional path; optimize the path points of the three-dimensional path to obtain the actual grinding path of the robot on the workpiece surface.

[0011] Optionally, the point cloud model building module is used to acquire workpiece point cloud data and form a workpiece point cloud model, including: Performing a deep scan and photographing of a workpiece to obtain point cloud data of the workpiece; performing pre-processing on the point cloud data by screening and removing abnormal data; and modeling the workpiece using the point cloud data to form a workpiece point cloud model; The rectangular region segmentation module is used to segment the workpiece point cloud model and extract a plurality of rectangular regions, including: The workpiece point cloud model is subjected to normal clustering segmentation to obtain a plurality of curved surface regions; and each curved surface region is fitted into a rectangular region on a projection plane using principal component analysis.

[0012] Optionally, the grid map generation module is used to divide all rectangular areas into grid maps according to the robot polishing path parameters, including: Divide all rectangular areas into a grid map according to the step size of the robot's grinding path, and each grid in the grid map is indexed by a two-dimensional coordinate (u, v); The two-dimensional path generation module is used to create rules to generate a two-dimensional path traversing the grid map, including: A zigzag or I-shaped path rule is created, and based on the path rule, a two-dimensional path that traverses the grid map is generated; the two-dimensional path is stored in a format of two mutually queryable arrays, one of which is a one-dimensional array of path point coordinates indexed by the path sequence, and the other is a two-dimensional array of sequence numbers in the path indexed by the coordinates.

[0013] Optionally, the boundary determination module is used to collect and analyze the workpiece surface image to obtain the boundary of the area to be polished on the workpiece surface, including: Acquire a workpiece surface image with the normal direction of the rectangular area as the shooting optical axis, perform image segmentation, contour extraction and convex polygon fitting on the workpiece surface image, and obtain the boundary of the area to be polished on the workpiece surface; The two-dimensional path adjustment module is used to adjust the path points of the two-dimensional path according to the boundary of the area to be polished, including: The coordinates of the contour line of the boundary of the area to be polished are normalized and transformed into the coordinate system of the grid map, the path points outside the contour line in the two-dimensional path are identified and filtered out, and then all the path points of the two-dimensional path are reordered.

[0014] Optionally, the three-dimensional path generation and optimization module is used to project the two-dimensional path onto a point cloud surface to obtain a three-dimensional path; and optimize the path points of the three-dimensional path to obtain an actual grinding path of the robot on the workpiece surface, including: Projecting the two-dimensional path onto a point cloud surface, thereby transforming the path point coordinates of the two-dimensional path into three-dimensional coordinates, thereby obtaining a three-dimensional path; The coordinates of the path points of the three-dimensional path are interpolated and / or filled in to obtain the actual grinding path of the robot on the workpiece surface.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The online path planning method for an automatic polishing robot and the algorithm for searching the coordinates of path points on the model surface provided by the present application have low time complexity and fast running speed, and can meet the requirements of real-time path planning; online path planning is simple to implement, and path planning can be performed directly on point cloud data. Discrete point clouds facilitate the segmentation of complex models, avoiding the problem of re-triangulation after dividing the network model; it is suitable for projecting any non-repetitive two-dimensional path, and can be integrated with other two-dimensional plane map path planning algorithms; the path planning boundary can be set by visually identifying the polishing area, and path planning can be automatically implemented within the boundary; it is not only suitable for automatic polishing online path planning, but can also be used for other surface processing processes, and has wide scenario applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. Among them: Figure 1 It is a flow chart of the online path planning method for the automatic polishing robot provided by the present invention.

[0017] Figure 2 is a 2D path on a grid map.

[0018] Figure 3 It is the true outline of the workpiece.

[0019] Figure 4 It is a two-dimensional path adjusted by the contour line of the boundary of the area to be polished on the surface of the workpiece.

[0020] Figure 5 It is a schematic diagram of the starting state of projecting a two-dimensional path onto the point cloud surface.

[0021] Figure 6 It is a schematic diagram of the final state of projecting the two-dimensional path onto the point cloud surface.

[0022] Figure 7 It is a structural diagram of the online path planning system for the automatic polishing robot provided by the present invention. DETAILED DESCRIPTION

[0023] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are described in detail below in conjunction with the accompanying drawings. It will be understood that the specific embodiments described herein are only used to explain the present application, rather than to limit the present application. It should also be noted that, for ease of description, only some, rather than all, structures related to the present application are shown in the accompanying drawings. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0024] As used herein, the terms "comprise," "comprising," and "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or elements is not limited to the listed steps or elements but may optionally include steps or elements not listed, or may optionally include other steps or elements inherent to the process, method, product, or apparatus.

[0025] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0026] See also Figure 1 As shown, an embodiment of the present application provides an online path planning method for an automatic polishing robot. The online path planning method for an automatic polishing robot includes: Acquire workpiece point cloud data to form a workpiece point cloud model; segment the workpiece point cloud model and extract several rectangular areas; Divide all rectangular areas into grid maps based on the robot's polishing path parameters; create rules to generate a two-dimensional path that traverses the grid map; Collect and analyze the workpiece surface image to obtain the boundary of the area to be polished on the workpiece surface; adjust the path points of the two-dimensional path according to the boundary of the area to be polished; The two-dimensional path is projected onto the point cloud surface to obtain a three-dimensional path; the path points of the three-dimensional path are optimized to obtain the actual grinding path of the robot on the workpiece surface.

[0027] The beneficial effects of the above embodiments are as follows: the online path planning method for the automatic polishing robot establishes a one-to-one mapping relationship between planes and curved surfaces through orthographic projection, locally divides complex curved surfaces into simple curved surfaces, reduces the algorithm time complexity of searching for the coordinates of path points on the model surface, improves the running speed to meet the real-time path planning requirements, can perform path planning directly on point cloud data, has low requirements on the accuracy of the reconstruction algorithm and the quality of the model, can convert three-dimensional path planning into two-dimensional plane path planning, and generate two-dimensional paths of arbitrary shapes, ensuring automatic and efficient online path planning of the robot.

[0028] In another embodiment, obtaining workpiece point cloud data to form a workpiece point cloud model; segmenting the workpiece point cloud model to extract a plurality of rectangular regions includes: Perform a deep scan and capture of the workpiece to obtain its point cloud data; perform pre-processing on the point cloud data by screening and removing abnormal data; and use the point cloud data to model the workpiece to form a point cloud model of the workpiece. The workpiece point cloud model is segmented by normal clustering to obtain several surface regions; each surface region is fitted into a rectangular region on the projection plane using principal component analysis.

[0029] In actual operation, a depth camera can be used to perform a deep scan and capture of the workpiece (especially the workpiece surface) to obtain point cloud data of the entire workpiece. Taking into account the influence of the depth camera's own internal factors and external environmental factors during the depth scan, the collected point cloud data will inevitably contain abnormal data such as erroneous data. In order to ensure the correctness of the point cloud data, the point cloud data is pre-processed for abnormal data screening and elimination, thereby eliminating erroneous data such as too large or too small in the point cloud data. The point cloud data is then used to perform three-dimensional modeling of the workpiece to obtain a workpiece point cloud model, thereby realizing the three-dimensional shape representation of the workpiece. The workpiece point cloud model is also clustered and divided into several curved surface areas according to direction. The surface curvature of each curved surface area changes slightly, and the maximum normal angle is less than 90 degrees. The principal component analysis (PCA) method is then used to fit each surface area into a rectangular area with the largest area on the projection plane S, and a two-dimensional plane coordinate system is established with the projection plane S as the plane where the XY axis is located. The point cloud coordinates corresponding to the point cloud data are transformed into the above coordinate system. During the specific transformation process, the coordinate value x and the coordinate value y of all point cloud coordinates (x, y, z) remain unchanged, and the coordinate value z of all point cloud coordinates (x, y, z) becomes zero, thereby realizing the transformation of point cloud coordinates from three-dimensional to two-dimensional.

[0030] In another embodiment, all rectangular areas are divided into a grid map according to the robot polishing path parameters; and rules are created to generate a two-dimensional path traversing the grid map, including: Divide all rectangular areas into grid maps according to the step size of the robot's grinding path, and each grid in the grid map is indexed by two-dimensional coordinates (u, v); Create a zigzag or I-shaped path rule and generate a two-dimensional path that traverses the grid map based on the path rule. The two-dimensional path is saved in two arrays that can query each other. One array is a one-dimensional array of the coordinates of the path points indexed by the path sequence, and the other array is a two-dimensional array of the sequence numbers in the path indexed by the coordinates.

[0031] In practice, see Figure 2 , divide all rectangular areas into grid_step according to the robot's step size during the grinding path (for example, corresponding to Figure 2 The grid map is divided into two grids with a side length of 0.2, and each grid in the grid map is indexed with a two-dimensional coordinate (u, v). Then a zigzag or I-shaped path rule is created. Figure 2 The path rule is an “I” shape. According to the above path planning, a two-dimensional path that traverses the entire grid map is generated (such as Figure 2 The blue two-dimensional path). Alternatively, the two-dimensional path can be saved in two mutually queryable arrays. One array is a one-dimensional array Point(index) = (u, v) that stores the coordinates (u, v) of the path points indexed by the path sequence. The other array is a two-dimensional array Index(u, v) = index that stores the sequence number of the coordinates in the path indexed by the coordinates (u, v). If a coordinate is not in the path, index = -1. This two-dimensional path saving format can be used to describe two-dimensional paths of arbitrary shapes, achieving applicability and compatibility with different workpiece surface shape scenarios.

[0032] In another embodiment, collecting and analyzing a workpiece surface image to obtain a boundary of a region to be polished on the workpiece surface; and adjusting path points of a two-dimensional path according to the boundary of the region to be polished include: The workpiece surface image is captured with the normal direction of the rectangular area as the shooting optical axis, and the workpiece surface image is subjected to image segmentation, contour extraction and convex polygon fitting to obtain the boundary of the area to be polished on the workpiece surface; The coordinates of the contour line of the boundary of the area to be polished are normalized and transformed into the coordinate system of the grid map. The path points outside the contour line in the two-dimensional path are identified and filtered out, and then all the path points of the two-dimensional path are reordered.

[0033] In actual operation, a high-definition camera can be used to capture the workpiece surface image in the normal direction of the rectangular area (i.e., the axis of the high-definition camera is controlled to be parallel to the normal direction of the rectangular area). The workpiece surface image is segmented, contour extracted, and convex polygon fitting is performed on the workpiece surface image to obtain the boundary of the area to be polished on the workpiece surface. Figure 3 and Figure 4 , respectively adjust the two-dimensional path according to the actual outline of the workpiece and the contour line of the boundary of the area to be polished on the workpiece surface, where Figure 4 The contour line adjustment of the two-dimensional path is Figure 3 Specifically, the coordinates of the contour line of the boundary of the area to be polished are normalized and transformed into the coordinate system of the grid map, so as to identify and filter out the path points outside the contour line in the two-dimensional path, and then re-sort all the path points of the two-dimensional path. If the area to be polished cannot be identified, the rectangular area is directly used as the boundary, so as to obtain the following Figure 4 The two-dimensional path shown provides a reliable basis for subsequent optimization of the three-dimensional path.

[0034] In another embodiment, the two-dimensional path is projected onto a point cloud surface to obtain a three-dimensional path; and the path points of the three-dimensional path are optimized to obtain an actual grinding path of the robot on the workpiece surface, including: Projecting the two-dimensional path onto the point cloud surface, thereby transforming the path point coordinates of the two-dimensional path into three-dimensional coordinates, thereby obtaining a three-dimensional path; The coordinates of the path points of the three-dimensional path are interpolated and / or filled in to obtain the actual grinding path of the robot on the workpiece surface.

[0035] In practice, in order to avoid retrieving coordinates in three-dimensional space, the point cloud is actually projected onto a two-dimensional plane to retrieve the nearest path point. As mentioned above, the point cloud data has been transformed into the projection plane coordinate system. You only need to set the Z coordinate of the point to zero to get the projected point, such as Figure 5 and Figure 6The following diagram shows the initial and final states of projecting a 2D path onto a point cloud surface. In the initial state, the 2D path is aligned with the point cloud surface; in the final state, the 2D path is completely projected onto the point cloud surface. Iterate through all vertices in the point cloud model and calculate the grid coordinates (u, v) of each vertex based on its (x, y) coordinates. The calculation method is (u, v) = (x / grid_step, y / grid_step). The path index corresponding to the coordinates (u, v) is queried, and the corresponding path point coordinates (Point(index)) are replaced with the vertex's 3D coordinates, i.e., Point(index(x / grid_step, y / grid_step)) = (x, y, z). The normal of the vertex is also output, thus obtaining the 3D path. Because multiple vertices may fall on the same grid, the vertex closest to the grid center is selected. Therefore, an array (Dis(index)) is maintained to store the minimum distance between each path point and the vertex. The present invention only needs to traverse all vertices once to find the mapping from two-dimensional coordinates to three-dimensional coordinates. Assuming that the number of point cloud vertices is n, the algorithm time complexity is O(n), and all vertices are calculated independently, and parallel computing acceleration is supported, thereby realizing real-time path planning.

[0036] In addition, the low local point density of the reconstructed point cloud model may result in the loss of some path points or the large distance between the path points and vertices. In this case, the path point coordinates are interpolated and / or supplemented. The coordinate optimization of the path point adopts the triangle barycentric coordinate interpolation, searches for the triangle mesh closest to the path projection point P, obtains the coordinates of the three vertices ABC, and uses the difference product method to calculate the area S of triangle ABC and its sub-triangles PBC, PCA, and PAB. ABC 、S PBC 、S PCA 、S PAB , path point coordinate z value The calculation formula is ,in 、 、 are the coordinate z values ​​of points A, B, and C respectively. If the distance between two adjacent points on the path is large and the normal direction changes greatly, the grinding path may deviate from the workpiece surface. In this case, the path point can be filled between the two points. The coordinates of the filled path point are determined by the intersection of the tangents of the two points along the path direction.

[0037] In one embodiment, before executing “obtaining workpiece point cloud data and forming a workpiece point cloud model”, the method may further include the following steps: Step A1: Scan the workpiece using a laser scanning shooting method, collect the scanned information, and obtain the original collection signal of the workpiece.

[0038] Step A2: Based on the improved Lambertian reflection model, the diffuse reflection and specular reflection components of the workpiece material are separated and quantified, and the original collected signal is corrected for the reflection signal intensity through normalization processing to obtain the corrected reflection signal intensity, thereby eliminating the interference of the scanning conditions.

[0039] In step A2, the following formula (1) can be used to obtain the corrected reflected signal intensity of each scanned point on the workpiece: (1-1) in, represents the corrected reflected signal intensity of the scanned point i on the workpiece (dimension is intensity); Represents the reference benchmark signal strength (dimension is intensity). This is a constant with a clear physical meaning and intensity dimension. Its value can be an empirical value, such as the signal strength value measured under standard conditions (specific distance, specific material standard plate, specific light source power). represents the original reflection signal strength of the i-th point (dimension is intensity). Represents the maximum intensity value (dimension is intensity) among the original emission signal intensity values ​​corresponding to all points on the workpiece; Represents the diffuse reflectance of the material at the i-th point, a coefficient between 0 and 1, dimensionless; Represents the specular reflectivity of the material at the i-th point, a dimensionless coefficient between 0 and 1; represents the incident angle at the i-th point; represents the observation angle at the i-th point; Represents the specular index or glossiness coefficient of the material at the i-th point. The larger the value, the smoother the material. Represents the diffuse reflectivity of the material at the jth point; represents the specular reflectivity of the material at the jth point, represents the incident angle at the jth point; represents the observation angle at the jth point; Represents the specular index or glossiness coefficient of the material at the jth point; Represents the set of all points on the workpiece surface.

[0040] The principle of the above formula (1-1) is as follows: This application is an improved Lambertian and non-Lambertian mixed reflection model. Its core principle is to eliminate the interference of measurement conditions and achieve material-adaptive signal correction through multi-level normalization and reference scaling. The formula includes the following processing: The first level normalization is achieved by The purpose is to eliminate the influence of absolute intensity. The principle is: the original signal intensity of the i-th point With the maximum intensity of the whole field By comparing, we get a dimensionless number between [0, 1], which eliminates the overall intensity scaling effect caused by light source power, global distance, etc.

[0041] The second level normalization is achieved by The purpose of this is to eliminate the reflection characteristics of the material and the influence of local geometry. The principle is: The numerator in this item is the theoretical reflection model for calculating the i-th point, which comprehensively considers diffuse reflection and specular reflection, and comprehensively represents the theoretical relative reflection ability of the material and geometric posture of the point.

[0042] The denominator in this term calculates the maximum value of this theoretical reflectance model at all points in the field. Dividing the numerator by this maximum value normalizes the local reflectivity differences caused by different materials (such as glossy metal and matte plastic) and different geometric postures (such as front and side), ultimately resulting in a material and geometric correction factor between (0, 1].

[0043] Finally, through , and the product of the two calculation results described in items 1 and 2 above to achieve benchmark scaling, so that the final output value of the formula is the correct corrected reflected signal strength of the i-th point.

[0044] The functions of the above formula (1) are: 1. Realize adaptive material correction: adapt to complex materials such as reflective, transparent, and light-absorbing materials through the diffuse reflection coefficient, specular reflection coefficient, and specular index; 2. Realize signal normalization: eliminate the influence of scanning distance and light source intensity; 3. Realize noise suppression precursor: provide weights for subsequent formulas (2-1), (2-2), and (2-3) to improve the accuracy of subsequent filtering.

[0045] Alternatively, the above formula (1) can also be expressed as follows: (1-2) Among them, Ni represents the unit normal vector at the i-th point (the unit vector located at the i-th point on the surface of the object and perpendicular to the surface of the object at that point); Li represents the unit incident light vector pointing to the light source at the i-th point; Vi represents the unit vector pointing from the i-th point to the camera or sensor (defining the direction in which the sensor observes the point); Ri represents the unit mirror reflection vector at the i-th point (according to the law of reflection of light, the unit vector of the ideal reflection direction calculated by the incident light vector Li and the normal vector Ni, for example, its calculation formula can be Nj represents the unit normal vector at the jth point (a unit vector located at the jth point on the surface of the object and perpendicular to the surface of the object at that point); Lj represents the unit incident light vector pointing from the jth point to the light source; Vj represents the unit vector pointing from the jth point to the camera or sensor (defining the direction in which the sensor observes the point); Rj represents the unit specular reflection vector at the jth point (a unit vector of the ideal reflection direction calculated from the incident light vector Lj and the normal vector Nj according to the law of reflection of light).

[0046] Step A3: Perform weighted neighborhood adaptive filtering based on the corrected reflected signal intensity to suppress noise, thereby obtaining filtered point cloud data of the workpiece.

[0047] Wherein, step A3 can be specifically implemented as follows: (2-1) (2-1) (2-3) in, (2-4) in, 、 、 Represents the filtered coordinates corresponding to the i-th point The x, y, and z coordinate values ​​of the i-th point; N is the number of neighboring points of the i-th point, and N is a preset positive integer, for example, the value of N can be 20; is the weight of the jth neighboring point of the ith point, 、 、 The original coordinates of the jth neighboring point of the i-th point The x, y, and z coordinate values ​​of is the mean of the corrected reflected signal strengths of all points in the neighborhood of the i-th point, is the standard deviation of the corrected reflected signal strength of all points in the neighborhood of the i-th point.

[0048] The principle of the above formulas (2-1), (2-2), and (2-3) is to calculate the weights based on the output of formula (1-1) or (1-2) and perform dynamic weighted averaging on the neighborhood points; The functions of the above formula are: 1. Achieve precise noise suppression and improve the signal-to-noise ratio of the point cloud; 2. Achieve feature preservation: By differentiating weights, it avoids edge blurring caused by traditional filtering, such as gear tooth profiles and thin-walled structures; 3. Achieve data purification: Output the filtered point cloud to provide high-quality geometric data for subsequent formulas.

[0049] Step A4: For missing areas of the workpiece point cloud data, such as when there is no valid coordinate data in the point cloud In the blank area, the missing points are interpolated according to the filtered point cloud data to fill the point cloud data in the missing area.

[0050] Wherein, step A4 can be implemented as follows: Step A41: determining missing areas of the point cloud data of the workpiece; Specifically: 1. Determine a distance threshold, which is R1 (the value of R1 can be 3-5) times the global point cloud average distance, which refers to the average distance between all points in the point cloud and their nearest neighbor points; 2. Calculate the average distance of A (the value of A can be 20-50) neighboring points around each point in the point cloud, determine the relationship between the average distance corresponding to each point and the distance threshold, and regard points with corresponding average distances greater than the distance threshold as potential missing boundary points; cluster the potential missing boundary points to form a closed contour, which is the contour of the missing area; Step A42: Based on the boundary of the missing area, expand outward by a width of R2 times the average spacing of the global point cloud. The annular area (the value of R2 can be 3~20) is used as the candidate neighborhood of the missing area; Step A43: Determine whether the point cloud density within the candidate neighborhood is greater than or equal to the global point cloud density (the point cloud density of areas other than the missing area on the workpiece). If so, use the candidate neighborhood as the neighborhood of the missing area and proceed to step A44. If not, output a rescan prompt. Step A44: uniformly select Q boundary points on the boundary of the missing region, where Q is equal to or greater than 3; Step A45: uniformly select M points in the neighborhood of the missing region as reference points; fit a minimum three-dimensional bounding box of the missing region using the boundary points, generate three-dimensional grid points at preset intervals within the minimum three-dimensional bounding box, each grid point corresponding to a sampling point, and determine the coordinates of the interpolation points in the missing region based on the distances from each sampling point to the Q boundary points, the distances from each reference point to the Q boundary points, the coordinates of each reference point, and the rate of change of the normal vector corresponding to each reference point; The coordinates of the interpolation points in the missing area can be calculated using the following formula: (3-1) (3-2) (3-3) in, (3-4) in, 、 、 Indicates the interpolation point coordinates obtained by traversing to the tth sampling point according to the above formula The x, y, and z coordinate values ​​(coordinate values ​​in the three-dimensional coordinate system); 、 、 Indicates the coordinates of the kth reference point The x, y, and z coordinate values ​​of the reference points; M represents the total number of reference points, which can be 20 to 50; Indicates the weight of the kth reference point, which represents the kth reference point to the current interpolation point The degree of impact, The larger the value, the greater the influence of the reference point on the final interpolation result. p represents the distance attenuation coefficient, and its value is greater than or equal to 1.

[0051] The functions of formulas (3-1), (3-2), and (3-3) are: High-fidelity completion of missing areas in the point cloud: Based on the known point cloud around the missing area, intelligently generate three-dimensional coordinate points in the missing area, effectively solving the problem of missing data caused by material or occlusion, and providing a complete and accurate three-dimensional model for path planning, ensuring that the subsequently generated robot grinding path can closely fit the workpiece surface and improve grinding accuracy.

[0052] The above steps A1-A4 collaboratively solve the core problems of signal distortion, noise interference, and data missing in point cloud data acquisition of complex material workpieces, providing an accurate and complete data foundation for the online path planning of this application and improving planning accuracy.

[0053] The value of p can be determined in advance according to the material of the workpiece, as shown in Table 1 below: Table 1

[0054] See also Figure 7 As shown, an embodiment of the present application provides an online path planning system for an automatic polishing robot. The online path planning system for an automatic polishing robot includes: Point cloud model building module, used to obtain workpiece point cloud data and form a workpiece point cloud model; Rectangular region segmentation module, used to segment the workpiece point cloud model and extract several rectangular regions; The grid map generation module is used to divide all rectangular areas into grid maps according to the robot polishing path parameters; A 2D path generation module, used to create rules to generate 2D paths that traverse the grid map; The boundary determination module is used to collect and analyze the workpiece surface image to obtain the boundary of the area to be polished on the workpiece surface; A two-dimensional path adjustment module is used to adjust the path points of the two-dimensional path according to the boundary of the area to be polished; The three-dimensional path generation and optimization module is used to project the two-dimensional path onto the point cloud surface to obtain a three-dimensional path; optimize the path points of the three-dimensional path to obtain the actual grinding path of the robot on the workpiece surface.

[0055] In another embodiment, the point cloud model building module is used to acquire workpiece point cloud data and form a workpiece point cloud model, including: Perform a deep scan and capture of the workpiece to obtain its point cloud data; perform pre-processing on the point cloud data by screening and removing abnormal data; and use the point cloud data to model the workpiece to form a point cloud model of the workpiece. The rectangular region segmentation module is used to segment the workpiece point cloud model and extract several rectangular regions, including: The workpiece point cloud model is segmented by normal clustering to obtain several surface regions; each surface region is fitted into a rectangular region on the projection plane using principal component analysis.

[0056] In another embodiment, the grid map generation module is used to divide all rectangular areas into grid maps according to the robot polishing path parameters, including: Divide all rectangular areas into grid maps according to the step size of the robot's grinding path, and each grid in the grid map is indexed by two-dimensional coordinates (u, v); The 2D Path Generation module is used to create rules to generate 2D paths that traverse a grid map, including: Create a zigzag or I-shaped path rule and generate a two-dimensional path that traverses the grid map based on the path rule. The two-dimensional path is saved in two arrays that can query each other. One array is a one-dimensional array of the coordinates of the path points indexed by the path sequence, and the other array is a two-dimensional array of the sequence numbers in the path indexed by the coordinates.

[0057] In another embodiment, the boundary determination module is used to collect and analyze the workpiece surface image to obtain the boundary of the area to be polished on the workpiece surface, including: The workpiece surface image is captured with the normal direction of the rectangular area as the shooting optical axis, and the workpiece surface image is subjected to image segmentation, contour extraction and convex polygon fitting to obtain the boundary of the area to be polished on the workpiece surface; The 2D path adjustment module is used to adjust the path points of the 2D path according to the boundary of the area to be polished, including: The coordinates of the contour line of the boundary of the area to be polished are normalized and transformed into the coordinate system of the grid map. The path points outside the contour line in the two-dimensional path are identified and filtered out, and then all the path points of the two-dimensional path are reordered.

[0058] In another embodiment, the 3D path generation and optimization module is used to project the 2D path onto the point cloud surface to obtain a 3D path; and optimize the path points of the 3D path to obtain the actual grinding path of the robot on the workpiece surface, including: Projecting the two-dimensional path onto the point cloud surface, thereby transforming the path point coordinates of the two-dimensional path into three-dimensional coordinates, thereby obtaining a three-dimensional path; The coordinates of the path points of the three-dimensional path are interpolated and / or filled in to obtain the actual grinding path of the robot on the workpiece surface.

[0059] The operation and effects of the automatic polishing robot online path planning system of the present invention are corresponding to and consistent with the above-mentioned automatic polishing robot online path planning method, and the automatic polishing robot online path planning system will not be repeated here.

[0060] In general, the online path planning method and system for the automatic polishing robot establishes a one-to-one mapping relationship between planes and curved surfaces through orthographic projection, locally dividing complex curved surfaces into simple curved surfaces, reducing the algorithm time complexity of searching for the coordinates of path points on the model surface, and improving the running speed to meet the requirements of real-time path planning. It can perform path planning directly on point cloud data, has low requirements on the accuracy of the reconstruction algorithm and model quality, can convert three-dimensional path planning into two-dimensional plane path planning, and generate two-dimensional paths of arbitrary shapes, ensuring automatic and efficient online path planning of the robot.

[0061] The above is only a specific embodiment of the present invention, and any other improvements made based on the concept of the present invention are considered to be within the scope of protection of the present invention.

Claims

1. An online path planning method for an automatic polishing robot, characterized in that: include: Acquire workpiece point cloud data to form a workpiece point cloud model; segment the workpiece point cloud model to extract a plurality of rectangular areas; Divide all rectangular areas into a grid map based on the robot polishing path parameters; create rules to generate a two-dimensional path traversing the grid map; Collecting and analyzing the workpiece surface image to obtain the boundary of the area to be polished on the workpiece surface; adjusting the path points of the two-dimensional path according to the boundary of the area to be polished; Projecting the two-dimensional path onto the point cloud surface to obtain a three-dimensional path; The path points of the three-dimensional path are optimized to obtain the actual grinding path of the robot on the workpiece surface.

2. The online path planning method for an automatic polishing robot according to claim 1, wherein: Acquiring workpiece point cloud data to form a workpiece point cloud model; segmenting the workpiece point cloud model to extract a plurality of rectangular areas includes: Performing a deep scan and photographing of a workpiece to obtain point cloud data of the workpiece; performing pre-processing on the point cloud data by screening and removing abnormal data; and modeling the workpiece using the point cloud data to form a workpiece point cloud model; The workpiece point cloud model is subjected to normal clustering segmentation to obtain a plurality of curved surface regions; and each curved surface region is fitted into a rectangular region on a projection plane using principal component analysis.

3. The online path planning method for an automatic polishing robot according to claim 2, wherein: Divide all rectangular areas into a grid map based on the robot polishing path parameters; create rules to generate a two-dimensional path traversing the grid map, including: Divide all rectangular areas into a grid map according to the step size of the robot's grinding path, and each grid in the grid map is indexed by a two-dimensional coordinate (u, v); A zigzag or I-shaped path rule is created, and a two-dimensional path traversing the grid map is generated based on the path rule. The two-dimensional path is stored in two mutually queryable arrays, one of which is a one-dimensional array of path point coordinates indexed by the path sequence, and the other is a two-dimensional array of sequence numbers in the path indexed by the coordinates.

4. The online path planning method for an automatic polishing robot according to claim 1, wherein: Collect and analyze the workpiece surface image to obtain the boundary of the area to be polished on the workpiece surface; Adjusting the path points of the two-dimensional path according to the boundary of the area to be polished includes: Acquire a workpiece surface image with the normal direction of the rectangular area as the shooting optical axis, perform image segmentation, contour extraction and convex polygon fitting on the workpiece surface image, and obtain the boundary of the area to be polished on the workpiece surface; The coordinates of the contour line of the boundary of the area to be polished are normalized and transformed into the coordinate system of the grid map, the path points outside the contour line in the two-dimensional path are identified and filtered out, and then all the path points of the two-dimensional path are reordered.

5. The online path planning method for an automatic polishing robot according to claim 1, wherein: Projecting the two-dimensional path onto the point cloud surface to obtain a three-dimensional path; Optimizing the path points of the three-dimensional path to obtain the actual grinding path of the robot on the workpiece surface includes: Projecting the two-dimensional path onto a point cloud surface, thereby transforming the path point coordinates of the two-dimensional path into three-dimensional coordinates, thereby obtaining a three-dimensional path; The coordinates of the path points of the three-dimensional path are interpolated and / or filled in to obtain the actual grinding path of the robot on the workpiece surface.

6. Automatic polishing robot online path planning system, characterized by: include: Point cloud model building module, used to obtain workpiece point cloud data and form a workpiece point cloud model; A rectangular region segmentation module is used to segment the workpiece point cloud model and extract a plurality of rectangular regions; The grid map generation module is used to divide all rectangular areas into grid maps according to the robot polishing path parameters; a two-dimensional path generation module, configured to create rules for generating a two-dimensional path traversing the grid map; The boundary determination module is used to collect and analyze the workpiece surface image to obtain the boundary of the area to be polished on the workpiece surface; A two-dimensional path adjustment module, configured to adjust the path points of the two-dimensional path according to the boundary of the area to be polished; A three-dimensional path generation and optimization module, used for projecting the two-dimensional path onto a point cloud surface to obtain a three-dimensional path; The path points of the three-dimensional path are optimized to obtain the actual grinding path of the robot on the workpiece surface.

7. The automatic polishing robot online path planning system according to claim 6, characterized in that: The point cloud model building module is used to obtain workpiece point cloud data and form a workpiece point cloud model, including: Performing a deep scan and photographing of a workpiece to obtain point cloud data of the workpiece; performing pre-processing on the point cloud data by screening and removing abnormal data; and modeling the workpiece using the point cloud data to form a workpiece point cloud model; The rectangular region segmentation module is used to segment the workpiece point cloud model and extract a plurality of rectangular regions, including: The workpiece point cloud model is subjected to normal clustering segmentation to obtain a plurality of curved surface regions; and each curved surface region is fitted into a rectangular region on a projection plane using principal component analysis.

8. The automatic polishing robot online path planning system according to claim 6, characterized in that: The grid map generation module is used to divide all rectangular areas into grid maps according to the robot polishing path parameters, including: Divide all rectangular areas into a grid map according to the step size of the robot's grinding path, and each grid in the grid map is indexed by a two-dimensional coordinate (u, v); The two-dimensional path generation module is used to create rules to generate a two-dimensional path traversing the grid map, including: A zigzag or I-shaped path rule is created, and a two-dimensional path traversing the grid map is generated based on the path rule. The two-dimensional path is stored in two mutually queryable arrays, one of which is a one-dimensional array of path point coordinates indexed by the path sequence, and the other is a two-dimensional array of sequence numbers in the path indexed by the coordinates.

9. The automatic polishing robot online path planning system according to claim 6, characterized in that: The boundary determination module is used to collect and analyze the workpiece surface image to obtain the boundary of the area to be polished on the workpiece surface, including: Acquire a workpiece surface image with the normal direction of the rectangular area as the shooting optical axis, perform image segmentation, contour extraction and convex polygon fitting on the workpiece surface image, and obtain the boundary of the area to be polished on the workpiece surface; The two-dimensional path adjustment module is used to adjust the path points of the two-dimensional path according to the boundary of the area to be polished, including: The coordinates of the contour line of the boundary of the area to be polished are normalized and transformed into the coordinate system of the grid map, the path points outside the contour line in the two-dimensional path are identified and filtered out, and then all the path points of the two-dimensional path are reordered.

10. The automatic polishing robot online path planning system according to claim 6, characterized in that: The three-dimensional path generation and optimization module is used to project the two-dimensional path onto the point cloud surface to obtain a three-dimensional path; Optimizing the path points of the three-dimensional path to obtain the actual grinding path of the robot on the workpiece surface includes: Projecting the two-dimensional path onto a point cloud surface, thereby transforming the path point coordinates of the two-dimensional path into three-dimensional coordinates, thereby obtaining a three-dimensional path; The coordinates of the path points of the three-dimensional path are interpolated and / or filled in to obtain the actual grinding path of the robot on the workpiece surface.

Images