A high-precision pose measurement method based on butt joint surface features

Abstract

This invention relates to a high-precision pose measurement method based on docking surface features, to solve the technical problems of high equipment cost, complex calibration process, and susceptibility to industrial environment in existing large component assembly methods. The method includes: 1. building an assembly platform; 2. calibrating the intrinsic parameters of a monocular vision sensor; 3. calibrating the monocular vision sensor and motion execution components using a hand-eye calibration method; 4. calibrating the corresponding target and pin hole; 5. measuring the pose of the first component relative to the second component.

Description

Technical Field

[0001] This invention relates to a method for assembling pin holes on the mating surfaces of components, and more specifically to a high-precision pose measurement method based on the features of the mating surfaces. Background Technology

[0002] With the continuous improvement of the performance of aircraft, rockets and other products in the aerospace field, higher requirements for assembly precision and automation have been put forward for the docking and assembly of large components. The assembly work was initially carried out by manual operation, that is, the docking attitude of large components was estimated by human eyes and then manually adjusted. This assembly method is labor-intensive, has low work efficiency, and depends on the work experience of the staff. Therefore, the adoption of automated assembly has become an inevitable trend. In the process of assembling large components, the measurement system needs to measure the relative spatial attitude between docking components in real time and guide the motion execution components to adjust the attitude of the components.

[0003] Currently, most mature automated assembly processes utilize laser trackers for pose measurement. This method calculates spatial attitude using the features of a target sphere or mating surface, offering high accuracy and a large measurement range. However, laser trackers are expensive, require detailed calibration for each assembly, and are complex and demanding in terms of environmental requirements. Compared to laser trackers, 3D measurement methods offer advantages in terms of cost and ease of use. However, in practical applications, 3D measurement systems are susceptible to industrial environmental influences, necessitating regular calibration of system parameters to ensure measurement accuracy. If vibrations are present at the work site, the calibration frequency increases, potentially impacting application performance. Summary of the Invention

[0004] The purpose of this invention is to solve the technical problems of high equipment cost, complex calibration process and susceptibility to industrial environment in existing assembly methods for large components, and to propose a high-precision pose measurement method based on docking surface features.

[0005] The technical solution of this invention is as follows:

[0006] A high-precision pose measurement method based on docking surface features, characterized by the following steps:

[0007] S1. Assemble the assembly platform: This includes a first assembly unit and a second assembly unit arranged opposite to each other. The first assembly unit includes a first fixed docking vehicle, a movable motion execution component disposed on the upper surface of the first fixed docking vehicle, and a first component fixedly disposed on the motion execution component. The first component has at least four first pin holes on its docking surface. The second assembly unit includes a second fixed docking vehicle, a fixed execution component disposed on the second fixed docking vehicle, and a second component fixedly disposed on the fixed execution component. The second component has second pin holes on its docking surface that correspond one-to-one with the first pin holes. Establish a three-dimensional coordinate system O-xyz for the first assembly unit, with O at the center of the docking surface of the first component.

[0008] S2. Perform intrinsic parameter calibration on the vision sensor to obtain the calibration parameters of the vision sensor;

[0009] S3. Install the intrinsically calibrated vision sensor in the first assembly unit. Use the hand-eye calibration method to calibrate the vision sensor and the motion execution component, calculate the transformation relationship between the vision sensor coordinate system and the motion execution component coordinate system, and obtain the rotation matrix from the vision sensor coordinate system to the motion execution component coordinate system. Translation matrix

[0010] S4. The first target and the second target are set on the first component and the second component, respectively, and both the first target and the second target are within the field of view of the vision sensor described in step S3; a three-dimensional measurement system is used to calibrate the positions of the first target and the first pin hole, and the positions of the second target and the second pin hole, respectively; according to the sorting rules, the coordinates of the first target marker points are obtained and sorted as follows: The coordinates of the second target marker points are sorted as follows: The coordinates of the first pin hole are sorted as follows: The coordinates of the second pin hole are sorted as follows:

[0011] S5. Using the visual sensor from step S3, simultaneously acquire the first target image and the second target image. Through image processing algorithms and the coordinate system transformation matrix, calculate the pitch angle θ, yaw angle ψ, roll angle φ, displacement x, displacement y, and displacement z of the first component relative to the second component, thus completing the pose measurement of the first component relative to the second component.

[0012] Furthermore, step S5 specifically includes:

[0013] S5.1. Using the visual sensor in step S3, simultaneously acquire the first target image and the second target image, and all the marker points on the first target and the second target are within the corresponding images. Extract the center coordinates of the first target marker points and the second target marker points respectively through the image processing algorithm, and sort them according to the sorting rules in step S4 to obtain the sorted p1 of the first target marker point center coordinates and the sorted p2 of the second target marker point center coordinates.

[0014] S5.2. Based on the calibration parameters of the visual sensor in step S2, the sorting of the center coordinates of the first target marker point p1, the sorting of the center coordinates of the second target marker point p2, and the sorting of the coordinates of the first target marker point in the target world coordinate system... Sorting the coordinates of the second target marker point in the target world coordinate system Calculate the coordinates of the first and second targets in the visual sensor coordinate system respectively. According to coordinates The rotation transformation matrix (R) between the computer vision sensor coordinate system and the target world coordinate system. c ,T c );

[0015] S5.3, Based on the rotation transformation matrix (R) between the visual sensor coordinate system and the target world coordinate system. c ,T c The first pin hole coordinates in the target world coordinate system are sorted. Second pin hole coordinate sorting Calculate the coordinates of the first pin hole in the visual sensor coordinate system. Coordinates of the second pin hole in the visual sensor coordinate system

[0016] S5.4. Based on step S2, the rotation matrix from the visual sensor coordinate system to the motion execution component coordinate system. Translation matrix from the vision sensor coordinate system to the motion actuator coordinate system and the coordinates of the first pin hole in the visual sensor coordinate system Coordinates of the second pin hole in the visual sensor coordinate system Calculate the coordinates of the first pin hole in the coordinate system of the motion actuator. The coordinates of the second pin hole in the coordinate system of the motion actuator component

[0017] S5.5 Construct the SVD function and solve... Calculate the rigid body transformation rotation matrix R of the first and second pin holes in the coordinate system of the motion actuator under the ideal docking condition. b Translation matrix t b ;

[0018] S5.6, According to the rotation matrix R b Translation matrix t b By utilizing the relationship between Euler angles, rotation matrix, and translation matrix, the pitch angle θ, yaw angle ψ, roll angle φ, displacement x, displacement y, and displacement z of the first component relative to the second component are calculated, thus completing the pose measurement of the first component relative to the second component.

[0019] Furthermore, it also includes step S6,

[0020] S6. Based on the pitch angle θ, yaw angle ψ, roll angle φ, displacement x, displacement y, and displacement z obtained in step S5.6, calculate the alignment error value between the first pin hole and the second pin hole. If the alignment error value between the first pin hole and the second pin hole is less than a preset threshold, the pose adjustment of the first component relative to the second component is completed. If the alignment error value between the first pin hole and the second pin hole is greater than or equal to the preset threshold, adjust the first component to align the first pin hole and the second pin hole, and repeat steps S5.1 to S5.6 until the alignment error value between the first pin hole and the second pin hole is less than the preset threshold.

[0021] Further, in step S4, a three-dimensional measurement system is used to calibrate the positions of the first target and the first pin hole, as well as the positions of the second target and the second pin hole; specifically:

[0022] S4.1. Use a three-dimensional measurement system to obtain the three-dimensional coordinates of the first target marker point, the first pin hole, the second target marker point, and the second pin hole in the target world coordinate system.

[0023] S4.2. Sort the three-dimensional coordinates obtained in step S4.1 according to the sorting rules, and obtain the first target marker coordinates sorted as follows: The coordinates of the second target marker points are sorted as follows: The coordinates of the first pin hole are sorted as follows: The coordinates of the second pin hole are sorted as follows:

[0024] Further, in step S5.3, the coordinates of the first pin hole in the visual sensor coordinate system are calculated. Coordinates of the second pin hole in the visual sensor coordinate system The formula is

[0025]

[0026]

[0027] Further, in step S5.4, the coordinates of the first pin hole in the coordinate system of the motion actuator are calculated. Coordinates of the second pin hole in the motion actuator coordinate system Specifically

[0028]

[0029]

[0030] Further, in step S5.6, according to the rotation matrix R... b Using the relationship between Euler angles, rotation matrix, and translation matrix, the pitch angle θ, yaw angle ψ, and roll angle φ of the first component relative to the second component are calculated.

[0031] Extracting the rotation matrix R b Translation matrix t b The rotation and translation matrix M in the equation is

[0032]

[0033] Where, r 11 r 12 r 13 …r 32 r 33 It is the rotation matrix R b The elements t1, t2, and t3 are translation matrices t b The elements in the formula are used to calculate the pitch angle θ, yaw angle ψ, and roll angle φ of the first component relative to the second component.

[0034]

[0035] Furthermore, in step S3, the vision sensor after internal parameter calibration is installed in the first assembly unit. Specifically, the first assembly unit is equipped with multiple vision sensors.

[0036] In step S4.1, at least one first target and at least one second target are set so that the field of view of each visual sensor simultaneously includes the first target and the second target.

[0037] In step S5.6, multiple vision sensors calculate multiple sets of pitch angle θ, yaw angle ψ, roll angle φ, displacement x, displacement y, and displacement z of the first component relative to the second component;

[0038] Based on the preset threshold ranges for pitch angle, yaw angle, roll angle, displacement x, displacement y, and displacement z, if pitch angle θ, yaw angle ψ, roll angle φ, displacement x, displacement y, and displacement z are all within their respective threshold ranges, then the set (θ, ψ, φ, x, y, z) is a valid value; if pitch angle θ, yaw angle ψ, roll angle φ, displacement x, displacement y, or displacement z is outside their respective threshold ranges, then the set (θ, ψ, φ, x, y, z) is an invalid value; the average of each set of valid values ​​is calculated and used as the pitch angle, yaw angle, roll angle, displacement x, displacement y, and displacement z of the first component relative to the second component;

[0039] If multiple vision sensors calculate multiple sets of pitch angle θ, yaw angle ψ, roll angle φ, displacement x, displacement y, and displacement z of the first component relative to the second component, and all of these values ​​are invalid, then return to step S5.1 to remeasure.

[0040] Furthermore, the first target and the second target are standard targets that are respectively installed on the first component and the second component;

[0041] Alternatively, the first target may be a feature inherent in the first component that has target attributes, and the second target may be a feature inherent in the second component that has target attributes.

[0042] Further, in step S3, the vision sensor with calibrated intrinsic parameters is installed in the first assembly unit, and the calibration of the vision sensor and motion execution component is performed using the hand-eye calibration method, specifically as follows:

[0043] The vision sensor is installed on the first fixed docking vehicle, and the hand-eye calibration fixture is fixedly installed on the motion execution component.

[0044] Alternatively, the vision sensor can be mounted on the motion actuator, and the hand-eye calibration fixture can be fixedly mounted in an area outside the motion actuator and the first component and within the field of view of the vision sensor, with the hand-eye calibration fixture remaining in a fixed position.

[0045] The beneficial effects of this invention are:

[0046] 1. The high-precision pose measurement method based on docking surface features proposed in this invention achieves coordinate system transformation during measurement by calibrating the intrinsic parameters of the vision sensor, the calibration between the vision sensor and the motion execution component, and the positions of the first target and the first pin hole, as well as the positions of the second target and the second pin hole. The above calibration only needs to be performed once for continuous use, simplifying the tedious calibration operation.

[0047] 2. This invention achieves docking of the first and second components by measuring the pose relationship between the first and second components in the coordinate system of the motion execution component and adjusting the motion execution component. Compared with the laser tracker measurement-guided docking method, it is easier to operate and has lower cost. Compared with the three-dimensional measurement-guided docking method, it has the advantages of high stability and good flexibility, and simplifies the operation and maintenance process.

[0048] 3. The method of the present invention is simple to calculate and has a fast calculation speed.

[0049] 4. The method provided by this invention can perform pose measurement tasks during docking of large components of different models. It has many application scenarios, a simple system structure, is easy to deploy, and has a simple calibration process with good stability. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of the high-precision pose measurement method based on docking surface features according to the present invention.

[0051] Figure 2 This is a schematic diagram of an embodiment of the high-precision pose measurement device based on docking surface features of the present invention;

[0052] Figure 3 This is a schematic diagram of the first pin hole on the mating surface in an embodiment of the present invention;

[0053] Figure 4 This is a schematic diagram of the hand-eye calibration principle in an embodiment of the present invention;

[0054] Figure 5 This is a schematic diagram illustrating the pose coordinate system transformation of the first component relative to the second component in an embodiment of the present invention;

[0055] The attached figures are labeled as follows:

[0056] 11-First fixed docking vehicle, 12-Motion execution component, 13-First component, 14-First target, 15-First pin hole, 16-Vision sensor, 21-Second fixed docking vehicle, 22-Fixed execution component, 23-Second component, 24-Second target, 25-Second pin hole. Detailed Implementation

[0057] See Figure 1 This embodiment provides a high-precision pose measurement method based on docking surface features, which includes the following steps:

[0058] S1. Assemble the platform: See [link / reference] Figure 2The assembly includes a first assembly unit and a second assembly unit arranged opposite to each other. The first assembly unit includes a first fixed docking vehicle 11, a movable motion execution component 12 disposed on the upper surface of the first fixed docking vehicle 11, and a first component 13 fixedly disposed on the motion execution component 12. The first component 13 has at least four first pin holes 15. See [link to documentation]. Figure 3 The second assembly unit includes a second fixed docking vehicle 21, a fixed execution component 22 disposed on the second fixed docking vehicle 21, and a second component 23 fixedly disposed on the fixed execution component 22. The second component 23 has second pin holes 25 that correspond one-to-one with the first pin holes 15. A three-dimensional coordinate system O-xyz is established for the first assembly unit, with O at the center of the docking surface of the first component 23.

[0059] S2. Perform intrinsic parameter calibration on the vision sensor 16. In this embodiment, the vision sensor 16 is a monocular vision sensor. In other embodiments, it can also be a binocular vision sensor or other measurement sensors. Obtain the calibration parameters of the monocular vision sensor and complete the intrinsic parameter calibration of the monocular vision sensor. In this embodiment, the intrinsic parameter calibration of the monocular vision sensor is achieved through a calibration board. The calibration method refers to Zhang Zhengyou's paper "A flexible new technique for camera calibration. IEEE Transactions on Pattern Analysis and Machine Intelligence [JJ. 2000, 22(11): 1330-1334".

[0060] S2.1. Use a monocular vision sensor to repeatedly acquire images of the internal parameter calibration board at different positions. Specifically, within the field of view of the monocular vision sensor, move the internal parameter calibration board freely 8-10 times without parallel movement, and try to cover the entire field of view with the movement range. Take a picture of the internal parameter calibration board once with the monocular vision sensor after each movement to obtain an image of the internal parameter calibration board.

[0061] S2.2, Let the coordinates of the center P of any feature point on the internal parameter calibration plate in the world coordinate system of the calibration plate be (X... w ,Y w Z w ), (X w ,Y w Z w The coordinates of the projection point on the image plane are (u, v); based on the perspective projection model of a monocular vision sensor, the effective focal length (f) of the monocular vision sensor in the X1 and Y1 directions is obtained. x ,f y The principal point coordinates (u0, v0) of the intersection of the optical axis of the monocular vision sensor and the image plane in the image pixel coordinate system are given by the following formula:

[0062]

[0063] Where: A is the intrinsic parameter matrix of the monocular vision sensor; (f x ,f y ) represents the effective focal length of the monocular vision sensor in the X1 and Y1 directions, which is the ratio of the physical size of a unit pixel in the X1 and Y1 axes of the image physical coordinate system to the focal length; (R,T) represents the external parameters of the monocular vision sensor, indicating the transformation relationship between the calibration board world coordinate system and the camera coordinate system; λ is a non-zero arbitrary scale factor.

[0064] Based on the lens distortion model of a monocular vision sensor, the radial distortion coefficients (k1, k2) and tangential distortion coefficients (p1, p2) are obtained, as shown in the following formulas:

[0065]

[0066] Where (x1, y1) are the ideal image coordinates, (x d ,y d ) represents the actual image coordinates, and r represents the distance between the actual image coordinates and the principal point coordinates of the monocular vision sensor.

[0067] S3. In order to guide the motion execution component 12 to achieve precise docking of the first component 13 and the second component 23, it is necessary to calculate the transformation relationship between the monocular vision sensor coordinate system and the motion execution component coordinate system by using the hand-eye calibration method.

[0068] The specific process of hand-eye calibration includes: S3.1, installing the monocular vision sensor after the internal parameter calibration in step S2 on the first fixed docking vehicle 11 of the first assembly unit, and fixing the hand-eye calibration fixture on the motion execution component 12. The hand-eye calibration fixture includes a mechanical fixture and a third target. Using the mechanical fixture, the third target is fixedly connected to the motion execution component 12. The motion execution component 12 is moved along its own coordinate system, moving three positions in each direction. During the movement, the third target is kept within the field of view of the monocular vision sensor.

[0069] S3.2. Each time the position is moved, the monocular vision sensor is activated to acquire the image of the third target at the current position. All markers of the third target should be included in the image of the third target.

[0070] S3.3 Extract the image coordinates of the marker points of the third target and match them one-to-one with the marker point coordinates of the third target coordinate system;

[0071] S3.4, see also Figure 4Let B be the rotational transformation in the coordinate system of the motion actuator, and A be the rotational transformation in the coordinate system of the monocular vision sensor. Based on the hand-eye calibration equation: AX = XB, calculate X. Obtain the rotation matrix from the monocular vision sensor coordinate system to the motion actuator coordinate system. and the translation matrix from the monocular vision sensor coordinate system to the motion actuator coordinate system

[0072] Understandably, in step S3.1, the monocular vision sensor can be mounted on the motion execution component 12, and the corresponding third target is set in an area outside the motion execution component 12 and the first component 13 by mechanical tooling and is located within the field of view of the monocular vision sensor, and the third target remains in a fixed position.

[0073] It should be noted that the installation position of the monocular vision sensor during the calibration of the monocular vision sensor and motion execution component 12 using the hand-eye calibration method in step S3 cannot be changed after calibration. Subsequent operations should continue to maintain the same position; otherwise, recalibration is required.

[0074] S4. The first target 14 is fixedly connected to the first component 13 by pasting or spraying, and the second target is fixed to the second component 23 by pasting or spraying, with both targets 14 and 24 within the field of view of the monocular vision sensor described in step S3. The number of marker points on the first target 14 and the second target 24 can be flexibly adjusted, wherein the marker points at the four corners of the target are coded marker points, used for identifying the target and sorting the non-coded marker points on the target. It can be understood that the first target 14 is an inherent feature of the first component 1 with target attributes, and the second target 24 is an inherent feature of the second component 23 with target attributes.

[0075] In addition, to improve the accuracy of the measurement, multiple monocular vision sensors and multiple first targets 14 and multiple second targets 24 can be set. Preferably, a monocular vision sensor is set on the left side of the first component 13, a monocular vision sensor is set on the right side, and a monocular vision sensor is set on the top side, while ensuring that the field of view of each monocular vision sensor includes both the first target 14 and the second target 24.

[0076] The positions of the first target 14 and the first pin hole 15, and the positions of the second target 24 and the second pin hole 25 were calibrated using a three-dimensional measurement system.

[0077] S4.1. Use a three-dimensional measurement system to obtain the three-dimensional coordinates of the first target marker point, the first pin hole 15, the second target marker point, and the second pin hole 25 in the target world coordinate system.

[0078] S4.2 Sort the three-dimensional coordinates obtained in step S4.1 according to the sorting rules, and obtain the coordinates of the first target marker point as P1. w The coordinates of the second target marker are sorted as follows: The coordinates of the first pin hole are sorted as follows: The coordinates of the second pin hole are sorted as follows:

[0079] S5, see also Figure 5 Measure the pose of the first component 13 relative to the second component 23:

[0080] S5.1. Using the monocular vision sensor in step S3, simultaneously acquire the first target image and the second target image, and all the marker points on the first target 14 and the second target 24 are within the corresponding images. Extract the center coordinates of the marker points of the first target 14 and the second target 24 respectively through the image processing algorithm, and sort them according to the sorting rules in step S4.3 to obtain the sorted p1 of the center coordinates of the first target marker points and the sorted p2 of the center coordinates of the second target marker points.

[0081] S5.2. Based on the calibration parameters of the monocular vision sensor in step S2, the coordinates of the center of the first target marker point are sorted as p1, the coordinates of the center of the second target marker point are sorted as p2, and the coordinates of the first target marker point in the target world coordinate system are sorted as P1. w The coordinates of the second target marker are sorted as follows The PNP algorithm was used to calculate the coordinates of the first target 14 and the second target 24 in the monocular vision sensor coordinate system. According to coordinates Calculate the rotation transformation matrix (R) between the monocular vision sensor coordinate system and the target world coordinate system. c ,T c ).

[0082] S5.3, Based on the rotation transformation matrix (R) between the monocular vision sensor coordinate system and the target world coordinate system. c ,T c The coordinates of the first pin hole in the target world coordinate system are ordered as follows: The coordinates of the second pin hole are sorted as follows: According to the formula

[0083]

[0084]

[0085] Calculate the coordinates of the first pin hole 15 in the monocular vision sensor coordinate system as follows: The coordinates of the second pin hole 25 in the monocular vision sensor coordinate system are:

[0086] S5.4. Based on step S2, the rotation matrix from the monocular vision sensor coordinate system to the motion execution component coordinate system. Translation matrix from the monocular vision sensor coordinate system to the motion actuator coordinate system and the coordinates of the first pin hole 15 in the monocular vision sensor coordinate system Coordinates of the second pin hole 25 in the monocular vision sensor coordinate system According to the formula

[0087]

[0088]

[0089] Calculate the coordinates of the first pin hole 15 in the motion actuator coordinate system. Coordinates of the second pin hole 25 in the coordinate system of the motion actuator

[0090] S5.5 Construct the SVD function and solve... Solve the rigid body transformation relationship rotation matrix R of the first pin hole 15 and the second pin hole 25 in the coordinate system of the motion actuator under the ideal docking condition. b Translation matrix t b .

[0091] S5.6, According to the rotation matrix R b Translation matrix t b By utilizing the relationship between Euler angles and the rotation matrix, the pitch angle θ, yaw angle ψ, roll angle φ, displacement x, displacement y, and displacement z of the first component 13 relative to the second component 23 are calculated, thus completing the pose measurement of the first component 13 relative to the second component 23; specifically, the rotation matrix R is extracted. b Translation matrix t b The rotation and translation matrix M in the equation is

[0092]

[0093] Where, r 11 r 12 r 13 …r 32 r 33 It is the rotation matrix R b The elements t1, t2, and t3 are translation matrices t b The elements in the equation; the pitch angle θ, yaw angle ψ, and roll angle φ of the first component 13 relative to the second component 23 are calculated using the following formulas.

[0094]

[0095] Complete the pose measurement of the first component 13 relative to the second component 23.

[0096] When multiple vision sensors 16 are set, the multiple vision sensors 16 calculate multiple sets of pitch angle θ, yaw angle ψ, roll angle φ, displacement x, displacement y, and displacement z of the first component 13 relative to the second component 23.

[0097] Based on the preset threshold ranges for pitch angle, yaw angle, roll angle, displacement x, displacement y, and displacement z, if pitch angle θ, yaw angle ψ, roll angle φ, displacement x, displacement y, and displacement z are all within their respective threshold ranges, then the group (θ, ψ, φ, x, y, z) is a valid value; if pitch angle θ, yaw angle ψ, roll angle φ, displacement x, displacement y, or displacement z is outside their respective threshold ranges, then the group (θ, ψ, φ, x, y, z) is an invalid value; the average of each group of valid values ​​is calculated and used as the pitch angle, yaw angle, roll angle, displacement x, displacement y, and displacement z of the first component 13 relative to the second component 23;

[0098] The average of the effective values ​​of each group is used as the pitch angle, yaw angle, and roll angle of the first component 13 relative to the second component 23. If multiple vision sensors 16 calculate multiple sets of invalid values ​​for the pitch angle θ, yaw angle ψ, roll angle φ, displacement x, displacement y, and displacement z of the first component 13 relative to the second component 23, the process returns to step S5.1 to remeasure.

[0099] S6. Based on the pitch angle θ, yaw angle ψ, and roll angle φ obtained in step S5.6, calculate the alignment error value between the first pin hole 15 and the second pin hole 25. If the alignment error value between the first pin hole 15 and the second pin hole 25 is less than a preset threshold, the pose adjustment of the first component 13 relative to the second component 23 is completed. If the alignment error value between the first pin hole 15 and the second pin hole 25 is greater than or equal to the preset threshold, adjust the first component 13 to align the first pin hole 15 and the second pin hole 25, and repeat steps S5.1 to S5.6 until the alignment error value between the first pin hole 15 and the second pin hole 25 is less than the preset threshold.

[0100] A high-precision pose measurement method based on docking surface features involves: 1. Calibrating the intrinsic parameters of a monocular vision sensor; 2. Determining the transformation relationship between the monocular vision sensor coordinate system and the motion execution component coordinate system; 3. Calibrating the positions of the first target 14 and the first pin hole 15, and the positions of the second target 24 and the second pin hole 25, respectively. This calibration only needs to be performed once for continuous measurement. If the corresponding monocular vision sensor, first target 14, or second target 24 becomes detached or needs to be replaced, recalibration is required. Recalibration is also necessary if the device experiences a severe impact or shaking. During the subsequent docking pose measurement, a monocular vision sensor captures a set of images containing all the marker points of the first target 14 and the second target 24. Target recognition technology is used to extract the coordinates of all the marker points of the two targets and sort them according to specific rules. Based on the monocular vision sensor intrinsic parameters, distortion system coefficients, and the coordinate system transformation relationship with the motion execution component, the three-dimensional coordinates of the two target marker points are transformed into the coordinate system of the motion execution component. Then, the relative pose of the first target 14 with respect to the second target 24 in the coordinate system of the motion execution component is calculated using the target and corresponding pin hole position calibration parameters.

Claims

1. A high-precision pose measurement method based on docking surface features, characterized in that, Includes the following steps: S1. Constructing an assembly platform: including a first assembly unit and a second assembly unit arranged opposite to each other. The first assembly unit includes a first fixed docking vehicle (11), a movable motion execution component (12) disposed on the upper surface of the first fixed docking vehicle (11), and a first component (13) fixedly disposed on the motion execution component (12). The first component (13) has at least four first pin holes (15) on its docking surface. The second assembly unit includes a second fixed docking vehicle (21), a fixed execution component (22) disposed on the second fixed docking vehicle (21), and a second component (23) fixedly disposed on the fixed execution component (22). The second component (23) has second pin holes (25) on its docking surface that correspond one-to-one with the first pin holes (15). Establish a three-dimensional coordinate system O-xyz for the first assembly unit, with O at the center of the docking surface of the first component (13). S2. Perform intrinsic parameter calibration on the vision sensor (16) and obtain the calibration parameters of the vision sensor (16); S3. Install the vision sensor (16) after internal parameter calibration in the first assembly unit, and use the hand-eye calibration method to calibrate the vision sensor (16) and the motion execution component (12), calculate the transformation relationship between the vision sensor coordinate system and the motion execution component coordinate system, and obtain the rotation matrix from the vision sensor coordinate system to the motion execution component coordinate system. Translation matrix ; S4. The first target (14) and the second target (24) are set on the first component (13) and the second component (23), and both the first target (14) and the second target (24) are within the field of view of the vision sensor (16) described in step S3; the positions of the first target (14) and the first pin hole (15) and the positions of the second target (24) and the second pin hole (25) are calibrated using a three-dimensional measurement system; according to the sorting rules, the coordinates of the first target marker points are obtained and sorted as follows: The coordinates of the second target marker are sorted as follows: The coordinates of the first pin hole are sorted as follows: The coordinates of the second pin hole are sorted as follows: ; Step S4 is as follows: S4.

1. Use a three-dimensional measurement system to obtain the three-dimensional coordinates of the first target marker point, the first pin hole (15), the second target marker point, and the second pin hole (25) in the target world coordinate system. S4.

2. Sort the three-dimensional coordinates obtained in step S4.1 according to the sorting rules, and obtain the first target marker coordinates sorted as follows: The coordinates of the second target marker are sorted as follows: The coordinates of the first pin hole are sorted as follows: The coordinates of the second pin hole are sorted as follows: ; S5. Using the vision sensor (16) from step S3, simultaneously acquire the first target image and the second target image. Through image processing algorithms and coordinate system transformation matrices, calculate the pitch angle of the first component (13) relative to the second component (23). Yaw angle Roll angle Displacement x, displacement y, and displacement z are used to complete the pose measurement of the first component (13) relative to the second component (23); Step S5 is as follows: S5.

1. Using the visual sensor (16) in step S3, simultaneously acquire the first target image and the second target image, and ensure that all marker points on the first target (14) and the second target (24) are within the corresponding images. Extract the center coordinates of the first target marker points and the second target marker points respectively using the image processing algorithm, and sort them according to the sorting rules in step S4 to obtain the sorted center coordinates of the first target marker points. Sort the center coordinates of the second target marker point ; S5.

2. Sort according to the calibration parameters of the vision sensor (16) in step S2 and the center coordinates of the first target marker. The second target marker point center coordinates are sorted. And the coordinate order of the first target marker point in the target world coordinate system Sorting the coordinates of the second target marker point in the target world coordinate system Calculate the coordinates of the first target (14) and the second target (24) in the visual sensor coordinate system respectively. According to coordinates The rotation transformation matrix between the computer vision sensor coordinate system and the target world coordinate system ( , ); S5.3, Based on the rotation transformation matrix between the visual sensor coordinate system and the target world coordinate system ( , (and the first pin hole coordinates in the target world coordinate system) Second pin hole coordinate sorting Calculate the coordinates of the first pin hole (15) in the visual sensor coordinate system. The coordinates of the second pin hole (25) in the visual sensor coordinate system The calculation formula is: ； ； S5.

4. Based on step S2, the rotation matrix from the visual sensor coordinate system to the motion execution component coordinate system. Translation matrix from the vision sensor coordinate system to the motion actuator coordinate system and the coordinates of the first pin hole (15) in the visual sensor coordinate system The coordinates of the second pin hole (25) in the visual sensor coordinate system Calculate the coordinates of the first pin hole (15) in the motion actuator coordinate system. The coordinates of the second pin hole (25) in the motion actuator coordinate system The calculation formula is: ； ； S5.5 Construct the SVD function and solve... Calculate the rigid body transformation rotation matrix of the first pin hole (15) and the second pin hole (25) in the coordinate system of the motion actuator under the ideal docking condition. Translation matrix ; S5.6, Based on the rotation matrix Translation matrix By using the relationship between Euler angles, rotation matrix, and translation matrix, the pitch angle of the first component (13) relative to the second component (23) can be calculated. Yaw angle Roll angle The displacements x, y, and z are used to complete the pose measurement of the first component (13) relative to the second component (23).

2. The high-precision pose measurement method based on docking surface features according to claim 1, characterized in that: It also includes step S6, S6. The pitch angle obtained in step S5.6 Yaw angle Roll angle Calculate the alignment error between the first pin hole (15) and the second pin hole (25) using displacements x, y, and z. If the alignment error between the first pin hole (15) and the second pin hole (25) is less than a preset threshold, the pose adjustment of the first component (13) relative to the second component (23) is completed. If the alignment error between the first pin hole (15) and the second pin hole (25) is greater than or equal to the preset threshold, the first component (13) is adjusted to align the first pin hole (15) and the second pin hole (25). Repeat steps S5.1 to S5.6 until the alignment error between the first pin hole (15) and the second pin hole (25) is less than the preset threshold.

3. The high-precision pose measurement method based on docking surface features according to claim 1, characterized in that: In step S5.6, according to the rotation matrix Using the relationship between Euler angles, rotation matrix, and translation matrix, the pitch angle of the first component (13) relative to the second component (23) is calculated. Yaw angle Roll angle Specifically, Extracting the rotation matrix Translation matrix The rotation and translation matrix M in the equation is ； Where, r 11 r 12 r 13 …r 32 r 33 It is a rotation matrix The elements t1, t2, and t3 are translation matrices. Elements in; calculate the pitch angle of the first component (13) relative to the second component (23). Yaw angle Roll angle The formula is as follows 。 4. The high-precision pose measurement method based on docking surface features according to claim 3, characterized in that: In step S3, the vision sensor (16) after internal parameter calibration is installed in the first assembly unit. Specifically, the first assembly unit is equipped with multiple vision sensors (16). In step S4.1, at least one first target (14) and at least one second target (24) are set so that the field of view of each vision sensor (16) simultaneously includes the first target (14) and the second target (24). In step S5.6, multiple vision sensors (16) calculate multiple sets of pitch angles of the first component (13) relative to the second component (23). Yaw angle Roll angle Displacement x, displacement y, displacement z; Based on the preset pitch angle threshold range, yaw angle threshold range, roll angle threshold range, displacement x threshold range, displacement y threshold range, and displacement z threshold range, if the pitch angle... Yaw angle Roll angle If displacements x, y, and z are all within their respective threshold ranges, then this group ( , , (x, y, z) are valid values; if the pitch angle Yaw angle Roll angle If displacement x, displacement y, or displacement z is outside the corresponding threshold range, then this group ( , , x, y, z) are invalid values; the average of each group of valid values ​​is calculated and used as the pitch angle, yaw angle, roll angle, displacement x, displacement y, and displacement z of the first component (13) relative to the second component (23); If multiple vision sensors (16) calculate multiple sets of pitch angles of the first component (13) relative to the second component (23), Yaw angle Roll angle If displacement x, displacement y, and displacement z are all invalid values, return to step S5.1 to remeasure.

5. The high-precision pose measurement method based on docking surface features according to claim 4, characterized in that: The first target (14) and the second target (24) are standard targets that are respectively installed on the first component (13) and the second component (23); Alternatively, the first target (14) is a feature inherent to the first component (13) that has target attributes, and the second target (24) is a feature inherent to the second component (23) that has target attributes.

6. The high-precision pose measurement method based on docking surface features according to claim 5, characterized in that: In step S3, the vision sensor (16) after internal parameter calibration is installed in the first assembly unit, and the calibration of the vision sensor (16) and the motion execution component (12) is carried out using the hand-eye calibration method as follows: The vision sensor (16) is installed on the first fixed docking vehicle (11), and the hand-eye calibration fixture is fixedly installed on the motion execution component (12); Alternatively, the vision sensor (16) can be mounted on the motion execution component (12), and the hand-eye calibration fixture can be fixedly mounted in an area outside the motion execution component (12) and the first component (13) and within the field of view of the vision sensor (16), with the hand-eye calibration fixture remaining in the same position.

Images