A robot measuring device and method for in-situ measurement of low-temperature surfaces

Abstract

The present invention relates to a robot measuring device and method for in-situ measurement of low-temperature profiles, in order to solve the technical problem that the data obtained by multiple measurements of the current robot measuring device has large errors when performing surface splicing, resulting in inaccurate measurement. The method includes: calibrating the robot measuring system; performing normal temperature measurement based on the calibrated system; coordinate conversion of point cloud data measured at normal temperature; normal temperature measurement data splicing including key point search, key point filtering, adjacent key point matrix conversion and data splicing; normal temperature dimension calculation; using the same method to perform low temperature measurement, coordinate system conversion, data splicing and low temperature dimension calculation; normal temperature and low temperature data comparison. The robot measuring system in the system includes a robotic arm and a three-dimensional profile measuring instrument arranged at the front end of the robotic arm, and the control processing system includes a memory and a processor, and a computer program is stored in the memory, and the program executes the above method when it runs in the processor.

Description

Technical Field

[0001] The invention relates to the field of robot measuring devices, and in particular to a robot measuring device and method for in-situ measurement of low-temperature surfaces. Background Art

[0002] The existing robot measurement system adopts the method of installing a machine vision three-dimensional surface dimension measuring instrument at the front end of the robot arm to integrate and build a flexible measurement system.

[0003] During the test, the robot arm drives the three-dimensional surface dimension measuring instrument to measure around the cryogenic equipment. After one round of measurement, it steps forward and steps around the cryogenic equipment to measure again. The above process is repeated until the cryogenic surface measurement is completed. The measurement range of the three-dimensional surface dimension measuring instrument is about 4.5cm when testing at the reference distance. For relatively large cryogenic equipment, the cryogenic surface dimensions cannot be obtained at one time. The data obtained from multiple tests need to be analyzed and spliced ​​to obtain the cryogenic surface dimensions. There is usually a certain error in the splicing process of the data obtained from multiple tests. Summary of the invention

[0004] The purpose of the present invention is to solve the technical problem that the data obtained by multiple measurements by the current robot measuring device has large errors when splicing the surface, resulting in inaccurate measurement, and propose a robot measuring device and method for in-situ measurement of low-temperature surfaces.

[0005] The technical solution provided by the present invention is:

[0006] A robot measurement method for in-situ measurement of cryogenic surfaces, which is special in that it is based on a robot measurement system, and the robot measurement system uses the robot arm zero coordinate system as a reference coordinate system;

[0007] The following steps are involved:

[0008] S1. Use the standard ball calibration method to calibrate multiple groups of posture DH parameters during the use of the robot measurement system;

[0009] S2. Normal temperature measurement: Based on the calibrated robot measurement system, use the robot arm to drive the three-dimensional surface measuring instrument within the measurement reference range of the three-dimensional surface measuring instrument, select a suitable reference distance, and circle the normal temperature test piece in multiple groups of postures for a total of K times of measurement to obtain K times of measurement point cloud data of the normal temperature test piece relative to the three-dimensional surface measuring instrument, where K ≥ 2 and K is an integer;

[0010] S3, coordinate system conversion: based on the DH parameters calibrated under multiple groups of postures of the robot measurement system, the K-times measurement point cloud data relative to the three-dimensional surface measuring instrument are converted into the K-times measurement point cloud data relative to the zero coordinate system of the robot arm;

[0011] S4. Normal temperature measurement data splicing:

[0012] Based on the K-times measured point cloud data of the robot zero coordinate system, the point cloud data of the L-th measurement is set as the current measurement point cloud set, and the point cloud data of the L+1-th measurement is set as the current target point cloud set, L<K, and L is an integer;

[0013] S4.1. Based on the adjacent point cloud A in the measured point cloud set or the target point cloud set n With A n+1 The relationship between the distance d and the set threshold y is used to determine the point cloud A. n+1 Is it a key point? If d>y, then point cloud A n+1 is the key point; if d<y, then point cloud A n+1 It’s not the key point;

[0014] S4.2, using a statistical filter to filter the point cloud key points screened out in step S4.1, remove the key points that do not meet the standards, and obtain the key points of the measurement point cloud set and the key points of the target point cloud set;

[0015] S4.3, calculating the points with the closest Euclidean distances between the key points of the measurement point cloud and the key points of the target point cloud, and taking a series of points with the closest Euclidean distances as registration points;

[0016] According to the formula

[0017]

[0018] Solve the rotation matrix R and translation vector T. When fm is the smallest, the rotation matrix R and translation vector T are the required rotation matrix and translation vector; where pi is a key point in the measured point cloud set, pji is the registration point in the target point cloud set corresponding to the key point pi in the measured point cloud set, and np is the number of key points in the measured point cloud set;

[0019] Then according to the formula p′ i =R·p i +T, so that the coordinates of the measured point cloud set are transformed into the coordinates of the target point cloud set to form the current measured point cloud set, which is the spliced ​​point cloud data of the Lth measured point cloud data and the L+1th measured point cloud data, where p′ i The point cloud after the key point pi in the measurement point cloud set is converted to the coordinates of the target point cloud set;

[0020] S4.4, according to the current measurement point cloud set in step S4.3 and the current target point cloud set composed of the measurement data adjacent to the current measurement point cloud set, repeat steps S4.1 to S4.3, complete the splicing of K measurement data in adjacent order, and finally obtain the normal temperature test piece point cloud set containing K measurement point cloud data in the same coordinates;

[0021] S5. Calculation of characteristic dimensions at room temperature: According to the spatial coordinate values ​​of the point cloud of the test piece at room temperature, the geometric dimensions of the test piece at room temperature are obtained;

[0022] S6, low temperature measurement: using the same reference distance as step S2, the same object to be tested is measured K times in the same posture under low temperature state, and K measurement point cloud data of the low temperature object to be tested relative to the three-dimensional surface measuring instrument are obtained;

[0023] S7, coordinate system conversion: based on the DH parameters calibrated under multiple postures of the robot measurement system, the K measurement point cloud data of the low-temperature test piece relative to the three-dimensional surface measuring instrument in step S6 are converted into K measurement point cloud data relative to the zero coordinate system of the robot arm;

[0024] S8, low temperature measurement data stitching: the point cloud data in the zero coordinate system of the robot arm is stitched in the same manner as step S4 to obtain a low temperature test piece point cloud set containing K times of measurement point cloud data;

[0025] S9, low temperature characteristic dimension calculation: according to the spatial coordinate value of the point cloud of the low temperature test piece, the geometric dimension of the low temperature test piece is obtained;

[0026] S10, comparing the size data of the corresponding parameters of the test piece at normal temperature and low temperature, and obtaining the deformation data of the test piece at low temperature.

[0027] Furthermore, the specific process of the standard ball calibration method in step S1 is as follows:

[0028] S1.1. Install the three-dimensional surface measuring instrument at the front end of the robotic arm, fix the bottom end of the robotic arm, use the standard ball as the target, the center of the standard ball as the fixed point, and the standard ball array with known ball center distance as the reference. Define the spatial connection between the robotic arm and the standard ball array as the equivalent robotic arm to form a virtual constraint.

[0029] S1.2, setting a group of robot arm postures, measuring and recording the coordinates of the robot arm and the standard ball array respectively, solving the robot arm posture according to the virtual constraint, and completing the DH parameter correction of the robot arm under the group of robot arm postures;

[0030] S1.3. Set other robot arm postures and repeat step S1.2 until the robot arm DH parameter calibration of multiple groups of postures is completed.

[0031] Furthermore, in step S4.2, the key points of the point cloud are filtered using a statistical filter as follows: the key point data is input into the statistical filter, the distance from each key point to the adjacent point is calculated and the average distance is solved, and the relationship between the average distance and the standard range is determined. If the average distance is within the standard range, the key point is retained; if the average distance is outside the standard range, the key point is removed.

[0032] Furthermore, the accuracy of the three-dimensional surface measuring instrument is less than 0.05 mm.

[0033] Furthermore, the three-dimensional surface measuring instrument is a line laser measuring instrument, which obtains a line laser point cloud during measurement.

[0034] The present invention also provides a robot measuring device for in-situ measurement of low-temperature profiles, which is special in that it includes a robot measuring system and a control processing system;

[0035] The robot measurement system includes a mechanical arm and a three-dimensional surface measuring instrument arranged at the front end of the mechanical arm;

[0036] The mechanical arm is used to drive the three-dimensional surface measuring instrument to move;

[0037] The control processing system includes a memory and a processor. The memory stores a computer program. When the program runs in the processor, the above method is executed.

[0038] Furthermore, the accuracy of the three-dimensional surface measuring instrument is less than 0.05 mm.

[0039] Furthermore, the three-dimensional surface measuring instrument is a line laser measuring instrument, and the measurement result is line laser point cloud data.

[0040] Beneficial effects of the present invention:

[0041] 1. In the data stitching process, the robot measurement method for in-situ measurement of low-temperature surfaces provided by the present invention searches and filters the key points in the point cloud data, aligns the key points of the finally selected measurement point cloud set and the points with the closest Euclidean distance in the key points of the target point cloud set, and transforms the coordinates of the measurement point cloud set into the coordinates of the target point cloud set to complete the stitching of two adjacent data. This stitching method of unifying the coordinates each time makes the key point alignment of the point cloud data more accurate, further improving the accuracy of the measurement data.

[0042] 2. The robot measurement method for in-situ measurement of low-temperature surfaces provided by the present invention converts multiple measurement point cloud data measured by the robot measurement system relative to the three-dimensional surface measuring instrument into point cloud data relative to the zero coordinate system of the robot arm, so that the splicing process of multiple data is in the same coordinate system, and the measured data has high accuracy and small error.

[0043] 3. The present invention adopts the standard ball calibration method to pre-calibrate the DH parameters of each group of postures of the robot system during the measurement process, rather than using the DH parameters given by the robot arm before leaving the factory. This can avoid the influence of deviations in the actual processing and installation process on the measurement, improve the accuracy of the DH parameters of the robot system, and increase the accuracy of data measurement. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 The present invention is a flow chart of the robot measurement method for in-situ measurement of low-temperature surfaces. DETAILED DESCRIPTION

[0045] This embodiment provides a robot measurement method for in-situ measurement of a low-temperature profile, comprising the following steps:

[0046] S1. Build and calibrate the measurement system: Install the three-dimensional surface measuring instrument on the robotic arm to form a robot measurement system. The robot measurement system is connected to the control processing system. The robot measurement system uses the robotic arm zero coordinate system as the reference coordinate system.

[0047] The 3D surface measuring instrument must meet the requirements of an accuracy of less than 0.05mm, be light in weight and small in size, be able to be installed on a robotic arm, be able to measure the reflective surface of stainless steel, and provide point cloud data; the robotic arm must be able to install the 3D surface measuring instrument, generally by making a suitable fixture to fix the 3D surface measuring instrument on the robotic arm.

[0048] For the selected robot arm, the DH parameters are generally given at the factory, but due to the heating of the robot arm, the dynamic influence of the unmodeled model, the actual processing and installation deviation, etc., the DH parameters of the robot arm will drift, resulting in a decrease in accuracy. Therefore, this embodiment uses the standard ball calibration method to calibrate the DH parameters of multiple groups of postures during the use of the robot measurement system. The specific process is:

[0049] S1.1. Install the three-dimensional surface measuring instrument at the front end of the robotic arm, fix the bottom end of the robotic arm, use the standard ball as the target, the center of the standard ball as the fixed point, and the standard ball array with known ball center distance as the reference. Define the spatial connection between the robotic arm and the standard ball array as the equivalent robotic arm to form a virtual constraint.

[0050] S1.2, setting a group of robot arm postures, measuring and recording the coordinates of the robot arm and the standard ball array respectively, solving the robot arm posture according to the virtual constraint, and completing the DH parameter correction of the robot arm under the group of robot arm postures;

[0051] S1.3. Set other robot arm postures and repeat step S1.2 until the robot arm DH parameter calibration of multiple groups of postures is completed.

[0052] Each group of calibrated robot arm postures is used for multiple measurements when measuring the workpiece to be measured. Before use, each group of robot arm postures is calibrated using the standard ball calibration method, and the control processing system saves each group of robot arm postures. When measuring, the measurement is performed according to the calibrated groups of robot arm postures.

[0053] S2. Normal temperature measurement: Based on the calibrated robot measurement system, use the robot arm to drive the 3D surface measuring instrument within the measurement reference range of the 3D surface measuring instrument, select a suitable reference distance, and circle the workpiece to be tested at normal temperature in multiple postures for a total of K measurements to obtain K measurement point cloud data of the workpiece to be tested at normal temperature relative to the 3D surface measuring instrument; the K measurement point cloud data are line laser point cloud data, where K ≥ 2 and K is an integer.

[0054] The measurement reference range of a three-dimensional surface measuring instrument refers to the distance range between the measuring instrument and the workpiece to be measured when the measuring accuracy and precision of the measuring instrument are high.

[0055] S3, coordinate system conversion: based on the DH parameters calibrated under multiple groups of postures of the robot measurement system, the K-times measurement point cloud data relative to the three-dimensional surface measuring instrument are converted into the K-times measurement point cloud data relative to the zero coordinate system of the robot arm;

[0056] Converting K times of measured point cloud data into a unified coordinate system is helpful to improve the accuracy of data stitching.

[0057] S4. Normal temperature measurement data splicing:

[0058] Based on the K times measured point cloud data of the robot zero coordinate system, in this embodiment, the point cloud data of the first measurement is set as the current measurement point cloud set, and the point cloud data of the second measurement is set as the current target point cloud set;

[0059] S4.1 Based on the adjacent point cloud A in the measured point cloud set or the target point cloud set n With A n+1 The relationship between the distance d and the set threshold y is used to determine the point cloud A. n+1 Is it a key point? If d>y, then point cloud A n+1 is the key point; if d<y, then point cloud A n+1 It is not a critical point; the threshold is an empirical value obtained based on test results.

[0060] S4.2 uses a statistical filter to filter the point cloud key points screened out in step S4.1, removes key points that do not meet the standards, and obtains the key points of the measurement point cloud set and the key points of the target point cloud set.

[0061] The specific method is: input the key point data into the statistical filter, calculate the distance from each key point to the adjacent point and solve the average distance, judge the relationship between the average distance and the standard range, if the average distance is within the standard range, the key point is retained, if the average distance is outside the standard range, the key point is a noise point or an error point, and it is removed.

[0062] The standard range is determined based on the distribution characteristics of the key points of the point cloud. Each measured point cloud data expresses a certain amount of information. The denser the point cloud in an area, the greater the amount of information. Noise points or error points are useless information or have very little information. Considering the characteristics of outlier point clouds, the distance from a key point of a point cloud to its adjacent points is calculated, the average distance and variance are solved, and points with variances outside of 3σ are defined as outside the standard range and are removed.

[0063] S4.3 calculates the points with the closest Euclidean distances between the key points of the measurement point cloud and the key points of the target point cloud, and uses a series of points with the closest Euclidean distances as registration points;

[0064] According to the formula

[0065]

[0066] Solve the rotation matrix R and translation vector T. When fm is the smallest, the rotation matrix R and translation vector T are the required rotation matrix and translation vector; where pi is a key point in the measured point cloud set, pji is the registration point in the target point cloud set corresponding to the key point pi in the measured point cloud set, and np is the number of key points in the measured point cloud set;

[0067] Then according to the formula: p′ i =R·p i +T, so that the coordinates of the measured point cloud set are transformed into the coordinates of the target point cloud set to form the current measured point cloud set, which is the spliced ​​point cloud data of the first measured point cloud data and the second measured point cloud data, where p′ i The point cloud after the key point pi in the measurement point cloud set is converted to the coordinates of the target point cloud set;

[0068] S4.4 repeats steps S4.1 to S4.3 according to the current measured point cloud set in step S4.3 and the current target point cloud set formed by the point cloud data measured for the third time, and completes the splicing of the K times of measured data in sequence, and finally obtains the normal temperature test piece point cloud set containing the K times of measured point cloud data in the same coordinates;

[0069] S5. Calculation of characteristic dimensions at room temperature: According to the spatial coordinate values ​​of the point cloud of the test piece at room temperature, the geometric dimensions of the test piece at room temperature are obtained;

[0070] S6, low temperature measurement: using the same reference distance as step S2, the same object to be tested is measured K times in the same posture under low temperature state, and K measurement point cloud data of the low temperature object to be tested relative to the three-dimensional surface measuring instrument are obtained;

[0071] S7, coordinate system conversion: based on the DH parameters calibrated under multiple postures of the robot measurement system, the K measurement point cloud data of the low-temperature test piece relative to the three-dimensional surface measuring instrument in step S6 are converted into K measurement point cloud data relative to the zero coordinate system of the robot arm;

[0072] S8, low temperature measurement data stitching: the point cloud data in the zero coordinate system of the robot arm is stitched in the same manner as step S4 to obtain a low temperature test piece point cloud set containing K times of measurement point cloud data;

[0073] S9, low temperature characteristic dimension calculation: according to the spatial coordinate value of the point cloud of the low temperature test piece, the geometric dimension of the low temperature test piece is obtained;

[0074] S10, comparing the size data of the corresponding parameters of the test piece at normal temperature and low temperature, and obtaining the deformation data of the test piece at low temperature.

Claims

1. A robot measurement method for in-situ measurement of low-temperature surfaces, characterized in that: Based on the robot measurement system, the robot measurement system uses the robot arm zero coordinate system as the reference coordinate system; The following steps are involved: S1. Use the standard ball calibration method to calibrate multiple groups of posture DH parameters during the use of the robot measurement system; S2. Normal temperature measurement: Based on the calibrated robot measurement system, use the robot arm to drive the three-dimensional surface measuring instrument within the measurement reference range of the three-dimensional surface measuring instrument, select a suitable reference distance, and circle the normal temperature test piece in multiple groups of postures for a total of K times of measurement to obtain K times of measurement point cloud data of the normal temperature test piece relative to the three-dimensional surface measuring instrument, where K ≥ 2 and K is an integer; S3, coordinate system conversion: based on the DH parameters calibrated under multiple groups of postures of the robot measurement system, the K-times measurement point cloud data relative to the three-dimensional surface measuring instrument are converted into the K-times measurement point cloud data relative to the zero coordinate system of the robot arm; S4. Normal temperature measurement data splicing: Based on the K-times measured point cloud data in the zero coordinate system of the robot arm, the point cloud data of the L-th measurement is set as the current measurement point cloud set, and the point cloud data of the L+1-th measurement is set as the current target point cloud set, L<K, and L is an integer; S4.1 Based on the adjacent point cloud A in the measured point cloud set or the target point cloud set n With A n+1 The relationship between the distance d and the set threshold y is used to determine the point cloud A. n+1 Is it a key point? If d>y, then point cloud A n+1 is the key point; if d<y, then point cloud A n+1 It’s not the key point; S4.2 uses a statistical filter to filter the point cloud key points screened out in step S4.1, removes the key points that do not meet the standards, and obtains the key points of the measurement point cloud set and the key points of the target point cloud set; S4.3 calculates the points with the closest Euclidean distances between the key points of the measurement point cloud and the key points of the target point cloud, and uses a series of points with the closest Euclidean distances as registration points; According to the formula Solve the rotation matrix R and translation vector T. When fm is the smallest, the rotation matrix R and translation vector T are the required rotation matrix and translation vector; where pi is a key point in the measured point cloud set, pji is the registration point in the target point cloud set corresponding to the key point pi in the measured point cloud set, and np is the number of key points in the measured point cloud set; Then according to the formula p′ i =R·p i +T, so that the coordinates of the measured point cloud set are transformed into the coordinates of the target point cloud set to form the current measured point cloud set, which is the spliced ​​point cloud data of the Lth measured point cloud data and the L+1th measured point cloud data, where p′ i The point cloud after the key point pi in the measurement point cloud set is converted to the coordinates of the target point cloud set; S4.4 repeats steps S4.1 to S4.3 according to the current measurement point cloud set in step S4.3 and the current target point cloud set formed by the measurement data adjacent to the current measurement point cloud set, and completes the splicing of the K measurement data in sequence in adjacent order, and finally obtains the normal temperature test piece point cloud set containing the K measurement point cloud data in the same coordinates; S5. Calculation of characteristic dimensions at room temperature: Calculate the geometric dimensions of the test piece at room temperature according to the spatial coordinate values ​​of the point cloud of the test piece at room temperature; S6, low temperature measurement: using the same reference distance as step S2, the same object to be tested is measured K times in the same posture under low temperature state, and K measurement point cloud data of the low temperature object to be tested relative to the three-dimensional surface measuring instrument are obtained; S7, coordinate system conversion: based on the DH parameters calibrated under multiple postures of the robot measurement system, the K-times measurement point cloud data of the low-temperature test piece relative to the three-dimensional surface measuring instrument in step S6 are converted into point cloud data relative to the zero coordinate system of the robot arm; S8, low temperature measurement data stitching: the point cloud data in the zero coordinate system of the robot arm is stitched in the same manner as step S4 to obtain a low temperature test piece point cloud set containing K times of measurement point cloud data; S9, low temperature characteristic dimension calculation: according to the spatial coordinate value of the point cloud of the low temperature test piece, the geometric dimension of the low temperature test piece is obtained; S10, comparing the size data of the corresponding parameters of the test piece at normal temperature and low temperature, and obtaining the deformation data of the test piece at low temperature.

2. The robot measurement method for in-situ measurement of low-temperature profiles according to claim 1, characterized in that: The specific process of the standard ball calibration method in step S1 is as follows: S1.

1. Install the three-dimensional surface measuring instrument at the front end of the robotic arm, fix the bottom end of the robotic arm, use the standard ball as the target, the center of the standard ball as the fixed point, and the standard ball array with known ball center distance as the reference. Define the spatial connection between the robotic arm and the standard ball array as the equivalent robotic arm to form a virtual constraint. S1.2, setting a group of robot arm postures, measuring and recording the coordinates of the robot arm and the standard ball array respectively, solving the robot arm posture according to the virtual constraint, and completing the DH parameter correction of the robot arm under the group of robot arm postures; S1.

3. Set other robot arm postures and repeat step S1.2 until the robot arm DH parameter calibration of multiple groups of postures is completed.

3. The robot measurement method for in-situ measurement of low-temperature profiles according to claim 1 or 2, characterized in that: In step S4.2, the key points of the point cloud are filtered using a statistical filter. Specifically, the key point data is input into the statistical filter, the distance from each key point to the adjacent point is calculated and the average distance is solved, and the relationship between the average distance and the standard range is determined. If the average distance is within the standard range, the key point is retained; if the average distance is outside the standard range, the key point is removed.

4. The robot measurement method for in-situ measurement of low-temperature profiles according to claim 3, characterized in that: The accuracy of the three-dimensional surface measuring instrument is less than 0.05 mm.

5. The robot measurement method for in-situ measurement of low-temperature profiles according to claim 4, characterized in that: The three-dimensional surface measuring instrument is a line laser measuring instrument, and the measurement result is line laser point cloud data.

6. A robot measuring device for in-situ measurement of low-temperature surfaces, characterized in that: Including robot measurement system and control processing system; The robot measurement system includes a mechanical arm and a three-dimensional surface measuring instrument arranged at the front end of the mechanical arm; The mechanical arm is used to drive the three-dimensional surface measuring instrument to move; The control processing system includes a memory and a processor. The memory stores a computer program. When the program runs in the processor, the method described in any one of claims 1 to 5 is executed.

7. The robot measuring device for in-situ measurement of low-temperature profiles according to claim 6, characterized in that: The accuracy of the three-dimensional surface measuring instrument is less than 0.05 mm.

8. The robot measuring device for in-situ measurement of low-temperature profiles according to claim 7, characterized in that: The three-dimensional surface measuring instrument is a line laser measuring instrument, and the measurement result is line laser point cloud data.

Images