A linear laser-based method for positioning a PCD profile milling cutter

Abstract

A kind of PCD profile milling cutter positioning method based on line laser, first milling cutter is clamped on tool holder, control line laser sensor to the cutting edge part of the front end of milling cutter is scanned, and depth data is obtained;Then edge extraction and correction are carried out to depth map, and edge contour point cloud is obtained;Tangent feature-based PPF algorithm is used to search and match edge contour point cloud and standard spline curve sampling point cloud, then the blade plane where the cutting edge is located is found and fitted, and the normal vector of the plane is obtained;Finally, according to line laser sensor coordinate system conversion and the size of cutting edge contour shape design, the actual processing contour position is positioned;The non-contact positioning method is used in the application, and there is no damage to the cutting edge and blade plane, the positioning efficiency is high, the compatibility is high, and the precision meets the requirements of PCD profile milling cutter edge positioning machining.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of workpiece clamping positioning and linear laser online measurement, and particularly relates to a PCD profile milling cutter positioning method based on linear laser. BACKGROUND

[0002] The PCD profile milling cutter refers to a profile forming milling cutter with a non-standard shape of a polycrystalline diamond composite blade welded on a cemented carbide cutter base body, which is used for milling aluminum alloy, fiber and metal composite material, engineering plastic, engineering ceramic and hard wood, and is widely applied in the aviation, automobile, precision medical and 3C consumer product industries. The PCD profile milling cutter is generally processed by electric spark or laser. Firstly, the PCD composite blade is cut after a reserved allowance according to the designed profile shape, and then the blade is brazed on the cutter base body. Then, the milling cutter is clamped on a machine tool for blade processing. Since the rotary position of the cutter clamping is indefinite, the milling cutter has a rake angle and an axial angle, and a position error is caused by welding, so it is necessary to determine the milling cutter position, position the actual processing blade profile and generate an NC program.

[0003] The existing positioning measurement method is a probe contact type. Chinese patent CN201410648592.8 discloses a PCD cutter edge processing method. In the method, the positioning method of the PCD cutter is to preset the probe measurement point position parameters, control the probe movement track, use the probe to detect the blade position, rake angle and axial angle, and find the blade plane according to the reference axis and the multi-point coplanar principle. The method needs manual parameter setting, the probe point taking efficiency is low, and the probe and the cutter are easily damaged. SUMMARY

[0004] In order to overcome the above-mentioned defects of the prior art, the purpose of the present application is to provide a PCD profile milling cutter positioning method based on linear laser, which uses a linear laser sensor to scan and obtain depth data, uses image and point cloud processing technology to position the actual processing profile, does not need manual setting, does not damage the edge and the blade plane, has high positioning efficiency, and the precision meets the requirements of the PCD profile milling cutter edge positioning processing.

[0005] In order to achieve the above-mentioned purpose, the technical scheme adopted by the present application is as follows:

[0006] A PCD profile milling cutter positioning method based on linear laser, comprising the following steps:

[0007] 1) Clamping and scanning: clamp the milling cutter on the tool holder, and the tool holder is HSK-A vacuum tool holder; then use a line laser sensor to scan along the axial direction of the milling cutter, and obtain depth data S containing the front edge part of the milling cutter, wherein the line laser sensor is fixed beside the galvanometer in the five-axis laser equipment, the rotating shaft is controlled to make the milling cutter horizontal, the line laser scans the surface of the milling cutter by moving the X axis, and the coordinate system of the obtained depth data is the sensor coordinate system;

[0008] 2) Contour point cloud acquisition: edge extraction is performed on the depth data obtained by scanning in the form of a depth image, and then the background points in the edge points are converted to obtain a corrected edge contour point set C p {c pi (x,y,ushort), i=0~N-1}, wherein x and y are X and Y coordinate values of the ith point relative to the line laser, and ushort is the gray value of the point; according to the height proportion parameter η of the line laser sensor, the ushort value in the edge point is converted into a height value, that is, the edge contour point set C p is restored to point cloud data C;

[0009] 3) Edge contour matching: sampling points are obtained according to the standard spline curve of the edge contour to obtain a model point cloud M; a tangent line fitting is performed on the model point cloud M by principal component analysis, and then a PPF algorithm based on the tangent direction feature is used to search and match the model point cloud M and the scene point cloud C obtained in step 2) to obtain a spatial transformation matrix T m→c ;

[0010] 4) Blade plane fitting: according to the spatial transformation matrix T m→c obtained in step 3), the model point cloud M is transformed into the point cloud S, a subset point cloud S p of the point cloud S within the bounding box range of the model point cloud M is selected, and then a random sample consensus algorithm is used to fit a blade plane in the point cloud S p , and a face normal vector n l (x,y,z) of the blade plane in the sensor coordinate system is obtained;

[0011] 5) Contour positioning: by pre-calibrating the spatial transformation matrix T l→g between the known line laser sensor coordinate system and the galvanometer coordinate system, the blade plane normal vector is converted into the galvanometer coordinate system T l→g n l (x,y,z) to obtain n g (x,y,z); then taking the center of the tool body rotation shaft as a reference, the actual machined edge contour curve is positioned according to the design size parameters of the edge contour shape.

[0012] The milling cutter is clamped on the HSK tool holder in the step 1), the tool holder is fixed on the rotary table, the relative position of the line laser sensor and the galvanometer is fixed, and the line laser is as parallel as possible to the Y axis of the machine tool during installation; the rotation axis of the rotary table is controlled to make the milling cutter axis direction as parallel as possible to the X axis of the machine tool, the Z axis is controlled to make the distance between the milling cutter and the line laser sensor, the sensor parameters are set, and the scanning of the milling cutter is completed by moving the X axis.

[0013] The specific method of edge extraction in the step 2) is as follows: the depth data obtained by scanning in the step 1) is read in the form of a depth image, wherein the gray value of a certain pixel represents the height value of the point, the edge gradient and direction of the image are calculated using a sobel operator, the edge points are judged, and then non-maximum suppression is performed to screen out the edge points; whether there is a non-edge point with a higher height value in the eight-neighborhood of the edge point is searched, and the background points are converted and corrected; according to the height proportion parameter η of the line laser sensor, the edge points are restored to point cloud data C.

[0014] The specific method of the PPF algorithm based on the tangent feature in the step 3) is as follows: the model point cloud M is obtained by sampling a standard tool edge profile spline curve, the tangent direction of each point is calculated by principal component analysis, the points with a large direction change rate are screened out as feature points, and the four-dimensional feature vector of each two points in the feature points is calculated Wherein d represents the vector between two points, n1 represents the tangent vector of a point, and n2 represents the tangent vector of another point, and an offline hash table is made; by performing feature matching, hash table query, coordinate alignment, pose voting and ICP accurate registration on the scene edge point cloud C and the model point cloud M, the space transformation matrix T of the model point cloud to the tool edge profile part of the scene point cloud is obtained m→c .

[0015] The PCD profile milling cutter positioning method based on the line laser is used for scanning the cutter surface by using a line laser sensor to obtain the normal vector of the cutter blade plane, after coordinate system conversion, the actual tool edge profile trajectory curve to be machined is positioned according to the design size parameters of the tool edge profile shape; and a machining program is generated by using NC automatic programming software.

[0016] The PCD cutter positioning method based on the line laser has the advantages that the PCD cutter, especially the PCD profile milling cutter, can be measured and positioned, the single-edge positioning time is less than 3 seconds, the efficiency is greatly higher than that of the method using a probe for positioning, the line laser sensor is used for non-contact measurement, the probe and the cutter blade are not damaged, the positioning process does not need manual pre-setting of parameters, the degree of automation is high, the cutter model and size are compatible, and the machining center is easy to integrate; the positioning requirements of the PCD cutter edge machining are met, and the positioning efficiency of the PCD cutter edge machining is improved. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1Method flowchart for the embodiment of the present application.

[0018] Figure 2 Edge profile point cloud extraction flowchart for the embodiment of the present application.

[0019] Figure 3 Edge profile point cloud extracted for the embodiment of the present application.

[0020] Figure 4 Model point cloud of the standard blade profile for the embodiment of the present application.

[0021] Figure 5 Blade profile matching flowchart for the embodiment of the present application. DETAILED DESCRIPTION

[0022] The present application will be further described below in conjunction with the drawings and embodiments.

[0023] As Figure 1 shown, a PCD profile milling cutter positioning method based on a line laser, comprising the following steps:

[0024] 1) Clamping and scanning: clamp the milling cutter on the tool holder, and the tool holder is HSK-A vacuum tool holder; then use the line laser sensor to scan along the axial direction of the milling cutter, and obtain the depth data S containing the front blade part of the milling cutter, wherein the line laser sensor is fixed beside the galvanometer in the five-axis laser equipment, the rotating shaft is controlled to make the milling cutter horizontal, the line laser scans the surface of the milling cutter by moving the X-axis, and the coordinate system of the obtained depth data is the sensor coordinate system;

[0025] In order to ensure that the line laser sensor (Keyence LJ-X8080, height range: ±20.5mm, laser line length: 35mm) has better scanning effect, the line laser is as parallel as possible to the machine tool Y-axis during installation; when scanning, the Z-axis is controlled to make the milling cutter and the line laser sensor at a suitable distance, the frequency is set to 1kHz, the exposure is 160us, the filtering processing is 17-point median and 8-time smoothing, there is no interval data, the milling cutter is scanned by moving the X-axis, and the single-blade scanning time is less than 2 seconds;

[0026] 2) Profile point cloud acquisition: as Figure 2 shown, the edge extraction is performed on the depth data obtained by scanning in the form of a depth image, the sobel operator is used to calculate the edge gradient and direction of the image, the edge points are determined, then the non-maximum suppression is performed, and the edge points are screened out; whether there is a non-edge point with a higher height value in the eight-neighborhood of the edge point is searched, the background points in the edge points are converted, and the corrected edge profile point set C is obtained p {c pi(x, y, ushort), i = 0 ~ N-1}, wherein x, y are the X, Y coordinate values of the i-th point relative to the line laser, and ushort is the grayscale value of the point; according to the height proportion parameter η = 1.6 of the line laser sensor, the ushort value in the edge point is converted into a height value:

[0027] c i .x = c pi .x

[0028] c i .y = c pi .y

[0029] c i .z = (c pi .ushort - 2 16 / 2) * η / 1000

[0030] wherein c i is the i-th point in the edge contour point cloud C, c pi is the i-th point in the edge contour point set, that is, the edge contour point set C p is restored into a point cloud data to obtain the edge contour point cloud C as shown in Figure 3 ;

[0031] 3) Blade profile matching: sampling points according to the standard spline curve of the blade profile, as shown in Figure 4 , to obtain the model point cloud M; performing tangent fitting on the model point cloud M by principal component analysis, and then searching and matching the model point cloud M and the obtained scene point cloud C by using the PPF algorithm based on the tangent direction features to obtain the spatial transformation matrix T m→c ;

[0032] The specific process is shown in Figure 5 , the model point cloud M is obtained by sampling the standard blade profile spline curve, the tangent direction of each point is calculated by principal component analysis, and the points with large direction change rates are selected as feature points, and the four-dimensional feature vector of each two points in the feature points is calculated , wherein d represents the vector between two points, n1 represents the tangent vector of a point, and n2 represents the tangent vector of another point, and an offline hash table is made; the scene edge point cloud C is down-sampled, the tangent feature vectors of the points are calculated, the points with large direction change rates are selected as feature points, and the spatial transformation matrix T m→c from the model point cloud M to the blade profile part of the scene point cloud C is obtained by feature matching, hash table query, coordinate alignment, pose voting and ICP precise registration on the model point cloud M.

[0033] 4) Blade plane fitting: according to the spatial transformation matrix T m→cTransform the model point cloud M into the point cloud S, select a subset point cloud S of the point cloud S within the bounding box range of the model point cloud M p Then fit the blade plane in the point cloud S p Through the random sampling consistency algorithm, obtain the face normal vector n l (x, y, z) of the blade plane in the sensor coordinate system;

[0034] 5) Contour positioning: through the pre-calibration of the space conversion matrix T l→g between the known line laser sensor coordinate system and the galvanometer coordinate system, convert the blade plane normal vector into the galvanometer coordinate system T l→g n l (x, y, z) to obtain n g (x, y, z); then, taking the center of the tool body rotation shaft as the reference, according to the design size parameters of the blade contour shape, the actual machining blade contour curve can be positioned, and the machining program can be generated through the NC automatic programming software.

Claims

1. A PCD contour milling cutter positioning method based on line laser, characterized in that, Includes the following steps: 1) Clamping and scanning: The end mill is clamped on the tool holder, which is an HSK-A vacuum tool holder; then, a line laser sensor is used to scan along the axial direction of the end mill to obtain depth data S including the cutting edge of the end mill. The line laser sensor is fixed next to the galvanometer in the five-axis laser equipment. The rotation axis is controlled to keep the end mill horizontal. The line laser scans the surface of the end mill by moving the X-axis. The coordinate system of the obtained depth data is the sensor coordinate system. 2) Contour point cloud acquisition: Edges are extracted from the depth data obtained by scanning in the form of a depth image. Then, background points in the edge points are converted to obtain the corrected edge contour point set C. p {c pi Let C be a set of edge points, i = 0 to N-1, where x and y are the X and Y coordinates of the i-th point relative to the line laser, and ushort is the gray value of that point. Based on the height scaling parameter η of the line laser sensor, the ushort values ​​in the edge points are converted into height values, i.e., the edge contour point set C is obtained. p Restore to point cloud data C; 3) Blade contour matching: Sampling points are obtained based on the standard spline curve of the blade contour to obtain the model point cloud M; tangent fitting is performed on the model point cloud M using principal component analysis, and then the PPF algorithm based on tangent direction features is used to search and match the model point cloud M and the scene point cloud C obtained in step 2) to obtain the spatial transformation matrix T. m→c ; 4) Blade plane fitting: Based on the spatial transformation matrix T obtained in step 3), m→c Transform the model point cloud M into point cloud S, and select a subset of point cloud S within the bounding box of model point cloud M. p Then, the random sampling consensus algorithm is used in the point cloud S p By fitting the blade plane, its surface normal vector n in the sensor coordinate system is obtained. l (x,y,z); 5) Contour localization: By pre-calibrating the known spatial transformation matrix T between the linear laser sensor coordinate system and the galvanometer coordinate system. l→g Transform the blade plane normal vector to the galvanometer coordinate system T. l→g n l (x,y,z) yields n g (x,y,z); Then, taking the center of the tool body rotation axis as the reference, the actual machining tool profile curve is located according to the design dimension parameters of the tool profile shape.

2. The method according to claim 1, characterized in that: In step 1), the milling cutter is clamped on the HSK tool holder, the tool holder is fixed on the rotary table, and the relative positions of the line laser sensor and the galvanometer are fixed. During installation, the line laser is as parallel as possible to the Y-axis of the machine tool. The rotation axis of the rotary table is controlled to make the direction of the milling cutter axis as parallel as possible to the X-axis of the machine tool. The Z-axis is controlled to make the distance between the milling cutter and the line laser sensor. The sensor parameters are set, and the scanning of the milling cutter is completed by moving the X-axis.

3. The method according to claim 1, characterized in that, The specific method for edge extraction in step 2) is as follows: the depth data obtained in step 1) is read in the form of a depth image, where the gray value of a certain pixel represents the height value of that point. The Sobel operator is used to calculate the edge gradient and direction of the image, and edge-like points are identified. Then, non-maximum suppression is performed to filter out edge points. In the eight-neighborhood of the edge point, it is searched to see if there are any non-edge points with a height value greater than it. The background points in the search are converted and corrected. According to the height ratio parameter η of the line laser sensor, the edge points are restored to point cloud data C.

4. The method according to claim 1, characterized in that, The specific steps of the PPF algorithm based on tangent features in step 3) are as follows: A model point cloud M is obtained by sampling the standard blade profile spline curve; the tangent direction of each point is calculated using principal component analysis; points with a large rate of change of direction are selected as feature points; and the four-dimensional feature vector of every two feature points is calculated. Where d represents the vector between two points, n1 represents the tangent vector of one point, and n2 represents the tangent vector of the other point, and an offline hash table is created; by performing feature matching, hash table lookup, coordinate alignment, pose voting, and ICP precise registration on the scene edge point cloud C and the model point cloud M, the spatial transformation matrix T from the model point cloud to the blade outline of the scene point cloud is obtained. m→c .

5. The method according to claim 1, characterized in that: The PCD contour milling cutter positioning method based on line laser is to use a line laser sensor to scan the milling cutter surface to obtain the normal vector of the cutting edge plane. After coordinate system transformation, the actual cutting edge contour trajectory curve to be machined is located according to the design size parameters of the cutting edge contour shape. The machining program is generated by NC automatic programming software.

Images