A method and system for calibrating the coordinate system of multiple infrared laser sensors in situ

Abstract

The present invention provides a method and system for calibrating the coordinate system of multiple infrared laser sensors in situ, which can be realized by using a generally produced finely processed round object such as a train wheel. After the first calibration is completed, when the calibration is repeated to ensure the measurement accuracy, the system can use the existing calibration transformation matrix as a coarse matching rotation and translation matrix to save the step of manual coarse matching, thereby realizing a fast and efficient calibration system function. In addition, based on up to 2096 frames of cross-sectional data, the equation of the motion of the object center around the axis center is analyzed to more accurately determine the relative relationship between the object center and the axis center, thereby calibrating the relative relationship matrix between multiple infrared laser sensors and the axis center of the measuring device turntable with high precision.

Description

Technical Field

[0001] The present invention relates to the technical field of computer vision measurement, and in particular to a method and system for calibrating a coordinate system of a multi-infrared laser sensor in situ. Background Art

[0002] Laser sensor is a sensor that uses laser technology for measurement. It consists of a laser, a laser detector and a measuring circuit. Laser sensor is a new type of measuring instrument. Its advantages are that it can achieve non-contact long-distance measurement, fast speed, high accuracy, large range, and strong resistance to light and electrical interference.

[0003] Laser scanning measurement technology is one of the key research objects in the field of geometric measurement today. The purpose of this measurement technology is to obtain three-dimensional data information on the surface of the object to be measured. Through the process of laser projection and image acquisition, a laser digital image carrying the three-dimensional coordinate information of the object to be measured is obtained. The laser image is preprocessed and the center is extracted, and then a series of data processing and analysis are performed. According to the principle that points with different heights on the surface of the object have different imaging positions in the image in the laser triangulation method, the measurement results of the three-dimensional contour of the object surface can be obtained. This technology is widely used in reverse engineering, virtual reality, rapid prototyping technology and other fields. It has the advantages of low cost and high precision and has high application value. Compared with the early single-point laser scanning measurement technology, the line laser scanning measurement technology is more efficient. A one-time scan can obtain two-dimensional data on a line on the surface of the object to be measured. Then, the object can be moved relative to the laser plane to obtain three-dimensional data. Therefore, it has become the research and development direction of laser scanning measurement technology.

[0004] Before measuring a circular workpiece, the existing multi-infrared laser sensor measuring device needs to calibrate the coordinate system in situ, but the calibration method is mostly based on specially customized calibration objects, and the requirement for placing the calibration object in the center of the turntable is also very high. Summary of the invention

[0005] The problem to be solved by the present invention is that the existing calibration method cannot quickly and accurately calibrate the relative relationship between the multi-line laser and the coordinate system original position of the rotating axis center of the turntable of the measuring device.

[0006] In order to solve the above problems, on the one hand, the present invention provides a method for calibrating a coordinate system of multiple infrared laser sensors in situ, which comprises the following steps:

[0007] S1: Fix several infrared laser sensors at designated positions of the measuring device and adjust the laser coplanarity;

[0008] S2: Place a calibration object near the center of the rotating shaft of the measuring device and rotate it once, triggering the infrared laser sensor to collect data and calculate the rotation angle of each frame;

[0009] S3: Through the calibration system, a certain frame of data A is manually roughly matched to obtain a roughly matched rotation and translation matrix;

[0010] S4: performing point cloud registration on the roughly matched frame data A and the point cloud data of the high-precision CAD cross section made based on the calibration object to obtain an accurate rotation and translation matrix;

[0011] S5: Perform matrix transformation on all acquired frame data based on precise rotation and translation matrix to piece together each frame of cross-sectional data;

[0012] S6: Calculate the point cloud registration of all other frame data relative to the frame data A to obtain the horizontal displacement offset relative to the frame data A;

[0013] S7: Based on the horizontal displacement offset and rotation angle of all frame data, the motion equation of the calibration object moving in situ around the center of the rotation axis is fitted;

[0014] S8: Calculate the actual horizontal displacement offset of frame data A based on the rotation angle of frame data A, and subtract the actual horizontal displacement offset of frame data A from the x variable of the displacement matrix of the precise rotation and translation matrix to obtain the final calibration transformation matrix.

[0015] Preferably, the step S1 specifically includes the following steps:

[0016] Fix three infrared laser sensors at designated positions of the measuring device, and adjust the installation angle of each infrared laser sensor through the fixing device so that the infrared lasers of all infrared laser sensors are emitted on the same plane in space, wherein the first infrared laser sensor installed on the top emits laser vertically downward, the second infrared laser sensor installed in the middle emits laser horizontally, and the third infrared laser sensor installed at the bottom emits laser vertically upward.

[0017] Preferably, the step S2 specifically includes the following steps:

[0018] A circular calibration object is placed on the turntable shaft of the measuring device. The height of the circular calibration object is horizontal and there is no tilt. The circular calibration object is placed near the center of the shaft. Turn on the infrared laser sensor, and at the same time start the shaft of the measuring device to rotate 360 ​​degrees. The line laser is triggered at a fixed frame rate, and a total of 2096 frames of data are collected. The corresponding rotation angle of each frame of data is calculated based on the trigger timestamp.

[0019] Preferably, the step S3 specifically includes the following steps:

[0020] Through the computer calibration system, a certain frame of data A is manually selected for measurement, wherein each frame of data is composed of measurement segments of three infrared laser sensors; the measurement segments are manually dragged and combined with the high-precision CAD sections made based on the circular calibration object to obtain the roughly matched rotation and translation matrices [R0_0|T0_0], [R1_0|T1_0], [R2_0|T2_0] of the three measurement segments relative to the CAD section of the circular calibration object, wherein [R0_0|T0_0] is the roughly matched rotation and translation matrices of the measurement segment of the first infrared laser sensor. Rotation and translation matrix, R0_0 is the rotation matrix (3*3 matrix), T0_0 is the translation matrix (3*1 matrix), [R1_0|T1_0] is the rotation and translation matrix of the rough matching of the measurement segment of the second infrared laser sensor, R1_0 is the rotation matrix (3*3 matrix), T1_0 is the translation matrix (3*1 matrix), [R2_0|T2_0] is the rotation and translation matrix of the rough matching of the measurement segment of the third infrared laser sensor, R2_0 is the rotation matrix (3*3 matrix), T2_0 is the translation matrix (3*1 matrix).

[0021] Preferably, step S4 specifically includes the following steps:

[0022] Based on the coarse matching rotation and translation matrices [R0_0|T0_0], [R1_0|T1_0], [R2_0|T2_0], the ICP (iterative closest point) algorithm in the PCL (Point Cloud Library) is called to perform point cloud registration between 3D point cloud sets, that is, the measurement data of frame data A is registered with the point cloud data of the high-precision CAD section, and the matching relationship matrices [R0|T0], [R1|T1], [R2|T2] of the three measurement fragments are respectively obtained, and then the following formula is used.

[0023] R0_1=R0*R0_0

[0024] T0_1=R0*T0_0+T0

[0025] R1_1=R1*R1_0

[0026] T1_1=R1*T1_0+T1

[0027] R2_1=R2*R2_0

[0028] T2_1=R2*T2_0+T2

[0029] Obtain accurate rotation and translation matrices [R0_1|T0_1], [R1_1|T1_1], [R2_1|T2_1]; wherein, [R0|T0] is the matching relationship matrix of the measurement segment of the first infrared laser sensor, R0 is the rotation matrix (3*3 matrix), T0 is the translation matrix (3*1 matrix), [R1|T1] is the matching relationship matrix of the measurement segment of the second infrared laser sensor, R1 is the rotation matrix (3*3 matrix), T1 is the translation matrix (3*1 matrix), [R2|T2] is the matching relationship matrix of the measurement segment of the third infrared laser sensor, R2 is the rotation matrix (3*3 matrix), T2 is the translation matrix (3*1 matrix); [R0_1|T0_1] is the precise rotation and translation matrix of the measurement segment of the first infrared laser sensor, R0_1 is the rotation matrix (3*3 matrix), T0_1 is the translation matrix (3*1 matrix), [R1_1|T1_1] is the precise rotation and translation matrix of the measurement segment of the second infrared laser sensor, R1_1 is the rotation matrix (3*3 matrix), T1_1 is the translation matrix (3*1 matrix), [R2_1|T2_1] is the precise rotation and translation matrix of the measurement segment of the third infrared laser sensor, R2_1 is the rotation matrix (3*3 matrix), T2_1 is the translation matrix (3*1 matrix).

[0030] Preferably, the step S5 specifically includes the following steps:

[0031] Based on the precise rotation and translation matrices [R0_1|T0_1], [R1_1|T1_1], and [R2_1|T2_1], all 2096 frames of collected data are transformed into matrices to piece together the cross-sectional point cloud data of each frame. The cross-sectional point cloud data is converted from the three measurement fragment data.

[0032] Preferably, the step S6 specifically includes the following steps:

[0033] Perform point cloud registration on each frame of cross-sectional point cloud data other than frame data A relative to the cross-sectional point cloud data of frame data A to obtain the horizontal displacement offset X relative to frame data A. i , i=1, 2,…, 2095, the horizontal offset X0 of frame data A is set to 0.

[0034] Preferably, step S7 specifically includes the following steps:

[0035] The horizontal offset X based on all frame data j and rotation angle B j, j = 0, 1, ..., 2095, the motion equation of the circular calibration object moving in situ around the center of the rotation axis is fitted, X = a*sin(b+B)+c; X represents the relative horizontal offset of the current frame relative to the center of the rotation axis, in millimeters, B represents the angle of the current frame relative to the center of the rotation axis, in radians, a represents the amplitude of the sine curve during the 360-degree circular motion, b represents the curve transformation angle, in radians, and c represents the correction value of the offset, in millimeters.

[0036] Preferably, the step S8 specifically includes the following steps:

[0037] Based on the rotation angle B(A) of frame data A, the actual offset of frame data A is obtained by subtracting X(A) from the x variable of the translation matrices T0_1, T1_1, and T2_1 of the precise rotation and translation matrices [R0_1|T0_1], [R1_1|T1_1], and [R2_1|T2_1], that is, by the following formula

[0038] R0_2=R0_1

[0039] T0_2=T0_1–(X(A), 0, 0)

[0040] R1_2=R1_1

[0041] T1_2=T1_1–(X(A), 0, 0)

[0042] R2_2=R2_1

[0043] T2_2=T2_1–(X(A), 0, 0)

[0044] Finally, the calibration transformation matrices [R0_2|T0_2], [R1_2|T1_2], and [R2_2|T2_2] of the three infrared laser sensors relative to the original position of the rotation axis center of the measuring device are obtained, wherein [R0_2|T0_2] is the calibration transformation matrix of the measurement segment of the first infrared laser sensor, R0_2 is a rotation matrix (a 3*3 matrix), and T0_2 is a translation matrix (a 3*1 matrix), [R1_2|T1_2] is the calibration transformation matrix of the measurement segment of the second infrared laser sensor, R1_2 is a rotation matrix (a 3*3 matrix), and T1_2 is a translation matrix (a 3*1 matrix), and [R2_2|T2_2] is the calibration transformation matrix of the measurement segment of the third infrared laser sensor, R2_2 is a rotation matrix (a 3*3 matrix), and T2_2 is a translation matrix (a 3*1 matrix).

[0045] On the other hand, the present invention further provides a system, which adopts the above-mentioned multi-infrared laser sensor coordinate system in-situ calibration method, wherein the system comprises:

[0046] An acquisition module, used for acquiring and saving the measurement segment point cloud data of multiple infrared laser sensors;

[0047] A 2D display module for visualizing the measurement segments of all frame data for manual review and evaluation of basic acquisition measurement frame quality;

[0048] The calibration module is used to manually drag the measurement fragment data of a certain frame for rough matching and perform the final calibration calculation;

[0049] The measurement and analysis module splices all frame data based on the final calibration transformation matrix, calculates the splicing accuracy of the frame data and the high-precision CAD cross-section frame, and gives an evaluation value of the calibration quality.

[0050] Compared with the prior art, the in-situ calibration method and system of the multi-infrared laser sensor coordinate system of the present invention have the following beneficial effects:

[0051] (1) The in-situ calibration method and system of the coordinate system of multiple infrared laser sensors described in the present invention can be implemented through generally produced finely processed circular objects such as train wheels. After the first calibration is completed, when the calibration is repeated to ensure the measurement accuracy, the system can use the existing calibration transformation matrix as a coarse matching rotation and translation matrix to eliminate the manual coarse matching steps, thereby achieving a fast and efficient calibration system function. In addition, based on up to 2096 frames of cross-sectional data, the equation of the motion of the object center around the center of the rotation axis is analyzed to more accurately determine the relative relationship between the object center and the center of the rotation axis, thereby calibrating the relative relationship matrix between multiple infrared laser sensors and the rotation axis center of the measuring device turntable with high precision;

[0052] (2) The in-situ calibration method and system of the multi-infrared laser sensor coordinate system of the present invention provides sufficient discrete data sets for fitting the equation of the motion of the center of the calibration object around the center of the rotation axis by synchronously and rapidly acquiring high-frame-rate data of the three infrared laser sensors on the measured object, so that the calibration is stable and accurate, and the calibration can be performed faster in the subsequent repeated calibration process;

[0053] (3) The method and system for calibrating the coordinate system of multiple infrared laser sensors in situ of the present invention can quickly and synchronously collect 2096 frames of cross-sectional data in a whole acquisition process of 12 seconds. The scanned cross-sectional data consists of measurement segments of multiple line lasers, and each segment is point cloud-aligned with the corresponding segment of the high-precision CAD cross section to achieve fast and high-precision point cloud alignment. In order to ensure the stability of the point cloud alignment, during the first calibration process, a method of manually rotating and translating the measurement segments is provided to perform rough matching and splicing. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a flow chart of the in-situ calibration method of the coordinate system of multiple infrared laser sensors of the present invention;

[0055] Figure 2 This is a module diagram of the in-situ calibration system of the multi-infrared laser sensor coordinate system of the present invention. DETAILED DESCRIPTION

[0056] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0057] Embodiment 1

[0058] A method for calibrating the coordinate system of multiple infrared laser sensors in situ is provided. Figure 1 As shown, the method includes the following steps:

[0059] S1: Fix several infrared laser sensors at designated positions of the measuring device and adjust the laser coplanarity;

[0060] S2: Place a calibration object near the center of the rotating shaft of the measuring device and rotate it once, triggering the infrared laser sensor to collect data and calculate the rotation angle of each frame;

[0061] S3: Through the calibration system, a certain frame of data A is manually roughly matched to obtain a roughly matched rotation and translation matrix;

[0062] S4: performing point cloud registration on the roughly matched frame data A and the point cloud data of the high-precision CAD cross section made based on the calibration object to obtain an accurate rotation and translation matrix;

[0063] S5: Perform matrix transformation on all acquired frame data based on precise rotation and translation matrix to piece together each frame of cross-sectional data;

[0064] S6: Calculate the point cloud registration of all other frame data relative to the frame data A to obtain the horizontal displacement offset relative to the frame data A;

[0065] S7: Based on the horizontal displacement offset and rotation angle of all frame data, the motion equation of the calibration object moving in situ around the center of the rotation axis is fitted;

[0066] S8: Calculate the actual horizontal displacement offset of frame data A based on the rotation angle of frame data A, and subtract the actual horizontal displacement offset of frame data A from the x variable of the displacement matrix of the precise rotation and translation matrix to obtain the final calibration transformation matrix.

[0067] Wherein, the step S1 specifically includes the following steps:

[0068] Fix three infrared laser sensors at designated positions of the measuring device, and adjust the installation angle of each infrared laser sensor through the fixing device so that the infrared lasers of all infrared laser sensors are emitted on the same plane in space, wherein the first infrared laser sensor installed on the top emits laser vertically downward, the second infrared laser sensor installed in the middle emits laser horizontally, and the third infrared laser sensor installed at the bottom emits laser vertically upward.

[0069] Three infrared laser sensors can simultaneously collect data of the measured object at a high frame rate and synchronously.

[0070] Wherein, the step S2 specifically includes the following steps:

[0071] A circular calibration object is placed on the turntable shaft of the measuring device. The height of the circular calibration object is horizontal and there is no tilt. The circular calibration object is placed near the center of the shaft. Turn on the infrared laser sensor, and at the same time start the shaft of the measuring device to rotate 360 ​​degrees. The line laser is triggered at a fixed frame rate, and a total of 2096 frames of data are collected. The corresponding rotation angle of each frame of data is calculated based on the trigger timestamp.

[0072] The circular calibration object only needs to be a finely processed non-reflective and non-light-absorbing circular object (such as a finely processed iron train wheel). Based on the characteristics of infrared laser, the calibration object rotates 360 degrees within 12 seconds, and a large amount of cross-sectional frame data can be collected to provide sufficient motion data for subsequent fitting to obtain accurate motion equations.

[0073] The calibration object used for calibration does not need to be placed at the center of the turntable with very high precision. The offset will be corrected later by fitting the motion equation.

[0074] Wherein, the step S3 specifically includes the following steps:

[0075] Through the computer calibration system, a certain frame of data A is manually selected for measurement, wherein each frame of data is composed of measurement segments of three infrared laser sensors; the measurement segments are manually dragged and combined with the high-precision CAD sections made based on the circular calibration object to obtain the roughly matched rotation and translation matrices [R0_0|T0_0], [R1_0|T1_0], [R2_0|T2_0] of the three measurement segments relative to the CAD section of the circular calibration object, wherein [R0_0|T0_0] is the roughly matched rotation and translation matrices of the measurement segment of the first infrared laser sensor. Rotation and translation matrix, R0_0 is the rotation matrix (3*3 matrix), T0_0 is the translation matrix (3*1 matrix), [R1_0|T1_0] is the rotation and translation matrix of the rough matching of the measurement segment of the second infrared laser sensor, R1_0 is the rotation matrix (3*3 matrix), T1_0 is the translation matrix (3*1 matrix), [R2_0|T2_0] is the rotation and translation matrix of the rough matching of the measurement segment of the third infrared laser sensor, R2_0 is the rotation matrix (3*3 matrix), T2_0 is the translation matrix (3*1 matrix).

[0076] The high-precision CAD section is the model CAD section used for forging the train wheel. It is the ideal cross-sectional shape of the forging die and is provided by the relevant personnel of the forging enterprise through AutoCAD and other software. The calibrated train wheel should be measured and analyzed by a high-precision 3D scanner, and the matching accuracy with the ideal CAD section should reach less than 0.1mm. However, the calibrated train wheel may also have defects in some parts, resulting in some frame data being unsatisfactory. Therefore, it is necessary to manually select a frame data with good scanning quality from the 2096 frame data. However, this situation rarely occurs. Generally, the quality of the selected frame data can meet the requirements. This is hereby explained.

[0077] After the computer calibration system uses the mouse and keyboard to rotate and translate to make the measurement segment coincide with the CAD section segment, the system will record the user's operation, save the final rotation angle and translation amount, as well as the original point data coordinates of the measurement segment and the point data coordinates after splicing, and the rotation and translation transformation matrix can be obtained through the point-to-point rotation angle and translation amount;

[0078] The measurement point data (x, 0, z) of the measurement segment of the first infrared laser sensor is obtained by

[0079] (x1, 0, z1)=R0_0*(x, 0, z)+T0_0

[0080] New coordinates can be calculated, and the point set fragments formed by these coordinate points will coincide with the corresponding fragments of the CAD section with low precision. Similarly, other measurement fragments are calculated and transformed in the same way.

[0081] Wherein, the step S4 specifically includes the following steps:

[0082] Based on the coarse matching rotation and translation matrices [R0_0|T0_0], [R1_0|T1_0], [R2_0|T2_0], the ICP (iterative closest point) algorithm in the PCL (Point Cloud Library) is called to perform point cloud registration between 3D point cloud sets, that is, the measurement data of frame data A is registered with the point cloud data of the high-precision CAD section, and the matching relationship matrices [R0|T0], [R1|T1], [R2|T2] of the three measurement fragments are respectively obtained, and then the following formula is used.

[0083] R0_1=R0*R0_0

[0084] T0_1=R0*T0_0+T0

[0085] R1_1=R1*R1_0

[0086] T1_1=R1*T1_0+T1

[0087] R2_1=R2*R2_0

[0088] T2_1=R2*T2_0+T2

[0089] Obtain accurate rotation and translation matrices [R0_1|T0_1], [R1_1|T1_1], [R2_1|T2_1]; wherein, [R0|T0] is the matching relationship matrix of the measurement segment of the first infrared laser sensor, R0 is the rotation matrix (3*3 matrix), T0 is the translation matrix (3*1 matrix), [R1|T1] is the matching relationship matrix of the measurement segment of the second infrared laser sensor, R1 is the rotation matrix (3*3 matrix), T1 is the translation matrix (3*1 matrix), [R2|T2] is the matching relationship matrix of the measurement segment of the third infrared laser sensor, R2 is the rotation matrix (3*3 matrix), T2 is the translation matrix (3*1 matrix); [R0_1|T0_1] is the precise rotation and translation matrix of the measurement segment of the first infrared laser sensor, R0_1 is the rotation matrix (3*3 matrix), T0_1 is the translation matrix (3*1 matrix), [R1_1|T1_1] is the precise rotation and translation matrix of the measurement segment of the second infrared laser sensor, R1_1 is the rotation matrix (3*3 matrix), T1_1 is the translation matrix (3*1 matrix), [R2_1|T2_1] is the precise rotation and translation matrix of the measurement segment of the third infrared laser sensor, R2_1 is the rotation matrix (3*3 matrix), T2_1 is the translation matrix (3*1 matrix).

[0090] If one point cloud is just a rigid transformation of another point cloud, the rigid transformation matrix between the two point clouds can be obtained by continuously reducing the distance between the two point cloud points, so that the two point clouds can be completely spliced ​​together. The algorithm of this iterative closest point is ICP, and this rigid transformation matrix is ​​the required matching relationship matrix.

[0091] The measurement point data (x1, 0, z1) of the measurement segment of the first infrared laser sensor is obtained by

[0092] (x2, 0, z2) = R0_1*(x1, 0, z1) + T0_1

[0093] New coordinate points can be calculated, and the point set fragments formed by these coordinate points will coincide with the corresponding fragments of the CAD section with high precision. Similarly, other measurement fragments are calculated and transformed in the same way.

[0094] Wherein, the step S5 specifically includes the following steps:

[0095] Based on the precise rotation and translation matrices [R0_1|T0_1], [R1_1|T1_1], and [R2_1|T2_1], all 2096 frames of collected data are transformed into matrices to piece together the cross-sectional point cloud data of each frame. The cross-sectional point cloud data is converted from the three measurement fragment data.

[0096] The measurement fragment data is a three-dimensional point cloud data set consisting of P0_0 (x, 0, z), P0_1 (x, 0, z), and P0_2 (x, 0, z). The converted point set is:

[0097] P1_0(x,0,z)=R0_1*P0_0(x,0,z)+T0_1,

[0098] P1_1(x, 0, z)=R1_1*P0_1(x, 0, z)+T1_1,

[0099] P1_2(x, 0, z)=R2_1*P0_2(x, 0, z)+T2_1.

[0100] Wherein, the step S6 specifically includes the following steps:

[0101] Perform point cloud registration on each frame of cross-sectional point cloud data other than frame data A relative to the cross-sectional point cloud data of frame data A to obtain the horizontal displacement offset X relative to frame data A. i , i=1, 2,…, 2095, the horizontal offset X0 of frame data A is set to 0.

[0102] Wherein, the step S7 specifically includes the following steps:

[0103] The horizontal offset X based on all frame data j and rotation angle B j , j = 0, 1, ..., 2095, the motion equation of the circular calibration object moving in situ around the center of the rotation axis is fitted, X = a*sin(b+B)+c; X represents the relative horizontal offset of the current frame relative to the center of the rotation axis, in millimeters, B represents the angle of the current frame relative to the center of the rotation axis, in radians, a represents the amplitude of the sine curve during the 360-degree circular motion, b represents the curve transformation angle, in radians, and c represents the correction value of the offset, in millimeters.

[0104] Construct (X0, B0), (X1, B1), ..., (X 2095 , B 2095 ) discrete point set; construct the shortest distance from the discrete point set to the motion equation and the minimum target equation,

[0105] E0=X0–(a*sin(b+B0)+c)

[0106] E1=X1–(a*sin(b+B1)+c)

[0107] …

[0108] E 2095 =X 2095 –(a*sin(b+B 2095 )+c)

[0109] Find a, b, c such that E0+E1+…+E 2095 The sum is the smallest,

[0110] The gradient descent method is used to fit and approximate the motion equation X=a*sin(b+B)+c of the calibration object's in-situ motion around the center of the rotation axis.

[0111] Wherein, the step S8 specifically includes the following steps:

[0112] Based on the rotation angle B(A) of frame data A, the actual offset of frame data A is obtained by subtracting X(A) from the x variable of the translation matrices T0_1, T1_1, and T2_1 of the precise rotation and translation matrices [R0_1|T0_1], [R1_1|T1_1], and [R2_1|T2_1], that is, by the following formula

[0113] R0_2=R0_1

[0114] T0_2=T0_1–(X(A), 0, 0)

[0115] R1_2=R1_1

[0116] T1_2=T1_1–(X(A), 0, 0)

[0117] R2_2=R2_1

[0118] T2_2=T2_1–(X(A), 0, 0)

[0119] Finally, the calibration transformation matrices [R0_2|T0_2], [R1_2|T1_2], and [R2_2|T2_2] of the three infrared laser sensors relative to the original position of the rotation axis center of the measuring device are obtained, wherein [R0_2|T0_2] is the calibration transformation matrix of the measurement segment of the first infrared laser sensor, R0_2 is a rotation matrix (a 3*3 matrix), and T0_2 is a translation matrix (a 3*1 matrix), [R1_2|T1_2] is the calibration transformation matrix of the measurement segment of the second infrared laser sensor, R1_2 is a rotation matrix (a 3*3 matrix), and T1_2 is a translation matrix (a 3*1 matrix), and [R2_2|T2_2] is the calibration transformation matrix of the measurement segment of the third infrared laser sensor, R2_2 is a rotation matrix (a 3*3 matrix), and T2_2 is a translation matrix (a 3*1 matrix).

[0120] Since the position of the wheel placed on the axle by the robot arm is based on the placement accuracy of about 10mm floating horizontally to the left and right of the center of the axle, the wheel data measured subsequently are first calibrated through the three transformation matrices obtained. After the point cloud data conversion, it can be ensured that the stitched data of all measurement frames floats to the left and right of the CAD cross-section coordinate position, which is convenient for the subsequent point cloud alignment and motion equation fitting, so as to quickly stitch the data with high precision.

[0121] In this way, the method in this embodiment realizes the calibration of the relative relationship matrix between multiple line lasers and the center of the measuring machine turntable rotation axis after fixing multiple infrared lasers on the measuring machine, which is used for subsequent fast and high-precision calculation and stitching of cross-sectional frame data of the rotating object placed on the turntable.

[0122] Embodiment 2

[0123] A system is provided, which adopts the in-situ calibration method of the multi-infrared laser sensor coordinate system as described in Example 1, such as Figure 2 As shown, wherein the system comprises:

[0124] An acquisition module, used for acquiring and saving the measurement segment point cloud data of multiple infrared laser sensors;

[0125] A 2D display module for visualizing the measurement segments of all frame data for manual review and evaluation of basic acquisition measurement frame quality;

[0126] The calibration module is used to manually drag the measurement fragment data of a certain frame for rough matching and perform the final calibration calculation;

[0127] The measurement and analysis module splices all frame data based on the final calibration transformation matrix, calculates the splicing accuracy of the frame data and the high-precision CAD cross-section frame, and gives an evaluation value of the calibration quality.

[0128] Since the calibration of train wheels is a precision process, the quality of the frame data collected can generally meet the requirements of subsequent calibration; some key point data information is provided on the CAD section of the wheel to represent the corresponding size data; data search is performed on each frame of all the spliced ​​frame data to find the data coordinates of the two points closest to the key point data, and the distance from the key point to the closest point of the line segment between the two points is the measurement accuracy; finally, the measurement accuracy of all frame data is averaged to obtain the splicing accuracy;

[0129] The acquisition module is specifically used for:

[0130] The interface data of rotating objects is measured by synchronously exposing multiple line lasers triggered by the turntable. The acquisition process is stable and efficient, and the 2096 frames of measurement data collected can be quickly saved and visualized;

[0131] The calibration module is specifically used for:

[0132] Provides manual selection of a certain frame of measurement fragment data, and controls the selected measurement fragment to translate and rotate through the keyboard to make the measurement fragment and CAD section splice together, and records the translation and rotation values ​​to convert them into a roughly matched rotation and translation matrix. Finally, the system will convert the measurement fragment through the roughly matched rotation and translation matrix and then perform point cloud registration with the high-precision CAD section point cloud data, making the stability and matching accuracy of point cloud registration higher;

[0133] The measurement and analysis module is specifically used for:

[0134] After calibration or measurement is completed, the measurement accuracy of each key dimension of the wheel data and the wheel data constructed by the high-precision CAD section is provided. The high-precision CAD section produced is equivalent to the actual size of the wheel used for actual calibration, which can be used to judge the accuracy of the calibration measurement.

[0135] In this way, the system in this embodiment realizes the calibration of the relative relationship matrix between multiple line lasers and the center of the measuring machine turntable rotation axis after fixing multiple infrared lasers on the measuring machine, which is used for subsequent fast and high-precision calculation and stitching of cross-sectional frame data of the rotating object placed on the turntable.

[0136] It should be understood that parts not described in detail in this specification are all prior art.

[0137] Although the present invention is disclosed as above, the protection scope of the present invention is not limited thereto. Those skilled in the art may make various changes and modifications without departing from the spirit and scope of the present invention, and these changes and modifications will fall within the protection scope of the present invention.

Claims

1. A method for calibrating the coordinate system of multiple infrared laser sensors in situ, characterized in that: The steps include: S1: Fix a plurality of infrared laser sensors at designated positions of the measuring device and adjust the laser coplanarity; the step S1 specifically includes the following steps: Fix three infrared laser sensors at designated positions of the measuring device, and adjust the installation angle of each infrared laser sensor by means of the fixing device so that the infrared lasers of all infrared laser sensors are emitted on the same plane in space, wherein the first infrared laser sensor installed at the top emits laser vertically downward, the second infrared laser sensor installed in the middle emits laser horizontally, and the third infrared laser sensor installed at the bottom emits laser vertically upward; S2: Place a calibration object near the center of the rotating shaft of the measuring device and rotate it once, triggering the infrared laser sensor to collect data and calculate the rotation angle of each frame; S3: manually perform a rough match on a frame of data A through a calibration system to obtain a rough matched rotation and translation matrix; the step S3 specifically includes the following steps: Through the computer calibration system, a certain frame of measurement data A is manually selected, wherein each frame of data is composed of measurement segments of three infrared laser sensors; the measurement segments are manually dragged and combined with the high-precision CAD sections made based on the circular calibration object to obtain the roughly matched rotation and translation matrices [R0_0|T0_0], [R1_0|T1_0], [R2_0|T2_0] of the three measurement segments relative to the CAD section of the circular calibration object, wherein [R0_0|T0_0] is the rotation and translation matrix of the first infrared laser sensor. The coarse matching rotation and translation matrix of the measurement segment of the first infrared laser sensor, R0_0 is a 3*3 rotation matrix, T0_0 is a 3*1 translation matrix, [R1_0|T1_0] is a coarse matching rotation and translation matrix of the measurement segment of the second infrared laser sensor, R1_0 is a 3*3 rotation matrix, T1_0 is a 3*1 translation matrix, [R2_0|T2_0] is a coarse matching rotation and translation matrix of the measurement segment of the third infrared laser sensor, R2_0 is a 3*3 rotation matrix, T2_0 is a 3*1 translation matrix S4: performing point cloud registration on the roughly matched frame data A and the point cloud data of the high-precision CAD cross section made based on the calibration object to obtain an accurate rotation and translation matrix; S5: Perform matrix transformation on all acquired frame data based on precise rotation and translation matrix to piece together each frame of cross-sectional data; S6: Calculate the point cloud registration of all other frame data relative to the frame data A to obtain the horizontal displacement offset relative to the frame data A; S7: Based on the horizontal displacement offset and rotation angle of all frame data, the motion equation of the calibration object moving in situ around the center of the rotation axis is fitted; S8: Calculate the actual horizontal displacement offset of frame data A based on the rotation angle of frame data A, and subtract the actual horizontal displacement offset of frame data A from the x variable of the displacement matrix of the precise rotation and translation matrix to obtain the final calibration transformation matrix.

2. The method for calibrating the coordinate system of multiple infrared laser sensors in situ according to claim 1, characterized in that: The step S2 specifically includes the following steps: A circular calibration object is placed on the turntable shaft of the measuring device. The height of the circular calibration object is horizontal and there is no tilt. The circular calibration object is placed near the center of the shaft. Turn on the infrared laser sensor, and at the same time start the shaft of the measuring device to rotate 360 ​​degrees. The line laser is triggered at a fixed frame rate, and a total of 2096 frames of data are collected. The corresponding rotation angle of each frame of data is calculated based on the trigger timestamp.

3. The method for calibrating the coordinate system of multiple infrared laser sensors in situ according to claim 2, characterized in that: The step S4 specifically includes the following steps: Based on the coarse matching rotation and translation matrices [R0_0|T0_0], [R1_0|T1_0], [R2_0|T2_0], the ICP algorithm in the PCL point cloud library is called to perform point cloud registration between 3D point cloud sets, that is, the measurement data of frame data A is registered with the point cloud data of the high-precision CAD section, and the matching relationship matrices [R0|T0], [R1|T1], [R2|T2] of the three measurement fragments are respectively obtained. Then, the following formula R0_1=R0*R0_0 T0_1=R0*T0_0+T0 R1_1=R1*R1_0 T1_1=R1*T1_0+T1 R2_1=R2*R2_0 T2_1=R2*T2_0+T2 Obtain accurate rotation and translation matrices [R0_1|T0_1], [R1_1|T1_1], [R2_1|T2_1]; wherein, [R0|T0] is the matching relationship matrix of the measurement segment of the first infrared laser sensor, R0 is a 3*3 rotation matrix, T0 is a 3*1 translation matrix, [R1|T1] is the matching relationship matrix of the measurement segment of the second infrared laser sensor, R1 is a 3*3 rotation matrix, T1 is a 3*1 translation matrix, [R2|T2] is the matching relationship matrix of the measurement segment of the third infrared laser sensor, R2 is a 3*3 rotation matrix, T2 is 3*1 translation matrix; [R0_1|T0_1] is the precise rotation and translation matrix of the measurement segment of the first infrared laser sensor, R0_1 is a 3*3 rotation matrix, T0_1 is a 3*1 translation matrix, [R1_1|T1_1] is the precise rotation and translation matrix of the measurement segment of the second infrared laser sensor, R1_1 is a 3*3 rotation matrix, T1_1 is a 3*1 translation matrix, [R2_1|T2_1] is the precise rotation and translation matrix of the measurement segment of the third infrared laser sensor, R2_1 is a 3*3 rotation matrix, T2_1 is a 3*1 translation matrix.

4. The method for calibrating the coordinate system of multiple infrared laser sensors in situ according to claim 3, characterized in that: The step S5 specifically includes the following steps: Based on the precise rotation and translation matrices [R0_1|T0_1], [R1_1|T1_1], and [R2_1|T2_1], all 2096 frames of collected data are transformed into matrices to piece together the cross-sectional point cloud data of each frame. The cross-sectional point cloud data is converted from the three measurement fragment data.

5. The method for calibrating the coordinate system of multiple infrared laser sensors in situ according to claim 4, characterized in that: The step S6 specifically includes the following steps: Perform point cloud registration on each frame of cross-sectional point cloud data other than frame data A relative to the cross-sectional point cloud data of frame data A to obtain the horizontal displacement offset X relative to frame data A. i , i=1, 2,…, 2095, the horizontal offset X0 of frame data A is set to 0.

6. The method for calibrating the coordinate system of multiple infrared laser sensors in situ according to claim 5, characterized in that: The step S7 specifically includes the following steps: The horizontal offset X based on all frame data j and rotation angle B j , j = 0, 1, ..., 2095, the motion equation of the circular calibration object moving in situ around the center of the rotation axis is fitted, X = a*sin(b+B)+c; X represents the relative horizontal offset of the current frame relative to the center of the rotation axis, in millimeters, B represents the angle of the current frame relative to the center of the rotation axis, in radians, a represents the amplitude of the sine curve during the 360-degree circular motion, b represents the curve transformation angle, in radians, and c represents the correction value of the offset, in millimeters.

7. The method for calibrating the coordinate system of multiple infrared laser sensors in situ according to claim 6, characterized in that: The step S8 specifically includes the following steps: Based on the rotation angle B(A) of frame data A, the actual offset of frame data A is obtained by subtracting X(A) from the x variable of the translation matrices T0_1, T1_1, and T2_1 of the precise rotation and translation matrices [R0_1|T0_1], [R1_1|T1_1], and [R2_1|T2_1], that is, by the following formula R0_2=R0_1 T0_2=T0_1–(X(A), 0, 0) R1_2=R1_1 T1_2=T1_1–(X(A), 0, 0) R2_2=R2_1 T2_2=T2_1–(X(A), 0, 0) Finally, the calibration transformation matrices [R0_2|T0_2], [R1_2|T1_2], and [R2_2|T2_2] of the three infrared laser sensors relative to the center of the rotation axis of the measuring device are obtained, wherein [R0_2|T0_2] is the calibration transformation matrix of the measurement segment of the first infrared laser sensor, R0_2 is a 3*3 rotation matrix, and T0_2 is a 3*1 translation matrix, [R1_2|T1_2] is the calibration transformation matrix of the measurement segment of the second infrared laser sensor, R1_2 is a 3*3 rotation matrix, and T1_2 is a 3*1 translation matrix, and [R2_2|T2_2] is the calibration transformation matrix of the measurement segment of the third infrared laser sensor, R2_2 is a 3*3 rotation matrix, and T2_2 is a 3*1 translation matrix.

8. A system using the in-situ calibration method of a multi-infrared laser sensor coordinate system as claimed in any one of claims 1 to 7, characterized in that: The system comprises: An acquisition module, used for acquiring and saving the measurement segment point cloud data of multiple infrared laser sensors; A 2D display module for visualizing the measurement segments of all frame data for manual review and evaluation of basic acquisition measurement frame quality; The calibration module is used to manually drag the measurement fragment data of a certain frame for rough matching and perform the final calibration calculation; The measurement and analysis module splices all frame data based on the final calibration transformation matrix, calculates the splicing accuracy of the frame data and the high-precision CAD cross-section frame, and gives an evaluation value of the calibration quality.

Images