Binocular vision system space error calibration method

Abstract

The application provides a binocular vision system space error calibration method, and inevitable errors still exist in the binocular vision system after traditional single and double target calibration, the method takes a laser tracker with higher measurement accuracy as a reference, provides a translational motion by means of a collaborative robot, and unifies the measurement data of the laser tracker and the measurement data of the binocular vision system by a designed coordinate system conversion method, so as to realize the estimation of the space measurement error of the binocular vision system, solve the calibration problem of the space error of the binocular vision system, and provide a further reference result for the error characteristics of the binocular vision system.

Description

Technical Field

[0001] This invention relates to the field of binocular stereo vision technology, and more specifically to a method for spatial error calibration of a binocular vision system. Background Technology

[0002] Binocular stereo vision technology is widely used in industrial 3D measurement and is a computer vision-based 3D measurement technique. Binocular stereo vision measurement technology achieves the calibration of system structural parameters through pre-calibration, thereby ensuring the accuracy of 3D measurements. However, due to the high complexity of the physical model of a binocular stereo vision system, existing models are insufficient to accurately characterize system characteristics, inevitably leading to certain calibration errors in existing binocular stereo vision system calibration methods. Furthermore, existing methods for evaluating calibration errors often rely on reprojection error, which fails to fully and accurately express the characteristics of calibration errors in 3D space. Summary of the Invention

[0003] Objective of the Invention: The technical problem to be solved by the present invention is to address the shortcomings of the prior art by providing a method for spatial error calibration of a binocular vision system, comprising the following steps:

[0004] Step 1: Fix two target holders on a metal plate that can be connected to the end flange of a collaborative robot (the UR5 model of UR Robotics is used in this method, but other models of collaborative robots can also be used). The target holders are used to place the reflective target ball that comes with the laser tracker and the vision target ball that corresponds to the binocular vision system, respectively. Fix the metal plate to the end of the collaborative robot.

[0005] Step 2: Place the laser tracker and the binocular vision system on the same side of the collaborative robot, and adjust the attitude of the metal plate fixed to the end of the collaborative robot in Step 1 to be approximately facing the binocular vision system, and the side of the metal plate on which the target is placed faces the binocular vision system.

[0006] Step 3: Arrange the reflective target ball and the visual target ball on the two target seats set up in Step 1, in random order, and guide the optical path of the laser tracker to the reflective target ball;

[0007] Step 4: Using the controller of the collaborative robot, control the robot to perform translational movements in the X, Y, and Z directions based on the end-effector TCP (tool centerposition) coordinate system. The human observes and records the coverage of the visual target ball at the robot's end in the measurement space of the binocular vision system within the robot's movement range.

[0008] Step 5: Sample the range of the binocular vision system measurement space that the visual target ball can cover under the recorded robot motion, sampling 60-90 points evenly.

[0009] Step 6: Based on the robot motion sampling points set in Step 5, control the robot to move to each point in sequence. Stop moving at each point. The laser tracker and binocular vision system measure the spatial coordinates of the reflective target ball and the visual target ball respectively, and save the measurement data at each sampling point.

[0010] Step 7: Process and analyze the measurement data of the laser tracker and the binocular vision system at each sampling point, and calculate the rotational transformation relationship between the coordinate systems of the laser tracker and the binocular vision system.

[0011] Step 8: Calculate the spatial error at each sampling point based on the rotation transformation relationship obtained in Step 7 and the measurement data at each sampling point saved in Step 6.

[0012] Preferably, in step 1, the target base, reflective target ball, and visual target ball are all standard parts with a standard size of 1.5 inches, wherein the diameter of the marking circle of the visual target ball is 12 mm.

[0013] Preferably, in step 1, the metal plate has a thickness greater than 5 mm and a plate area of ​​not less than 80 mm × 100 mm.

[0014] Preferably, in step 2, the laser tracker is positioned at a distance of no more than 2m from the collaborative robot.

[0015] Preferably, in step 6, the controlled robot moves sequentially to each point, and the motion is a translational motion relative to the TCP coordinate system.

[0016] Preferably, step 7 includes: taking the first sampling position as the initial reference position, and recording the coordinates of the reflective target ball and the visual target ball at the initial reference position, measured by the laser tracker and the binocular vision system, as P1 respectively. LT and P1 SV The coordinates of the reflective target sphere and the visual target sphere at other positions i, measured by the laser tracker and the binocular vision system, are denoted as P. i LT and P i SV Where i≥2, the measurement results of the reflective target ball and visual target ball at other sampling positions are subtracted from the corresponding measurement results at the initial reference position to obtain the translation vectors of the reflective target ball and visual target ball at each sampling position relative to the initial position, which are denoted as follows: and Based on the correspondence between each translation vector, the rotational transformation relationship between the laser tracker coordinate system and the binocular vision coordinate system is calculated.

[0017] Preferably, in step 7, the rotational transformation relationship between the laser tracker coordinate system and the binocular vision coordinate system is calculated using the least squares algorithm. The calculation expression is as follows: Where R represents the obtained rotational transformation relationship.

[0018] Preferably, step 8 includes: based on the translation vector of the reflective target ball at sampling position i relative to the initial position. The rotational transformation relationship R between the laser tracker coordinate system and the binocular vision coordinate system is obtained, with reference to the measured value v of the laser tracker. i LT Calculate the standard reference value of the translation vector of the visual target ball at position i relative to the initial position visual target ball in the binocular vision coordinate system. in The standard reference value for the coordinates of the visual target ball at position i is Therefore, the measurement error of the binocular vision system at position i Where P i SV The measurement value of the visual target point at this location by the binocular vision system is given; the spatial error of the binocular vision system at each sampling point location is calculated.

[0019] The present invention also provides a storage medium storing a computer program or instructions, which, when the computer program or instructions are run, implement the aforementioned method for spatial error calibration of a binocular vision system.

[0020] Beneficial effects: This invention utilizes a calibration system based on a collaborative robot and a laser tracker. The collaborative robot provides translational motion in space, and the measurement based on the laser tracker serves as a reference benchmark, providing an evaluation standard for the spatial measurement error of the binocular vision system. On this basis, the spatial error of the binocular vision system is calibrated, providing an effective means of evaluating the error characteristics of the binocular vision system, and achieving a more accurate and comprehensive error assessment of the binocular vision system. Attached Figure Description

[0021] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments, and the advantages of the present invention in the above and / or other aspects will become clearer.

[0022] Figure 1 This is a flowchart illustrating the implementation of the binocular vision spatial error calibration method according to an embodiment of the present invention. Detailed Implementation

[0023] like Figure 1 As shown, the present invention provides a spatial error calibration method for a binocular vision system, comprising the following steps:

[0024] Step S1: Fix two target holders on a metal plate that can be connected to the end flange of the collaborative robot. These two holders are used to place the reflective target ball for the laser tracker and the vision target ball for the binocular vision system, respectively. Then, fix the metal plate to the end of the collaborative robot.

[0025] First, for the target mount, a 3-point target mount can be selected. The sizes of the target mount, reflective target ball, and visual target ball need to be corresponding, and 1.5 inches can be selected.

[0026] Secondly, the distance between the two target mounts depends on the size of the selected metal plate, and is generally set between 10 and 20 centimeters.

[0027] Step S2: Place the laser tracker and the binocular vision system on the same side of the collaborative robot, and adjust the attitude of the metal plate fixed to the end of the collaborative robot in step S1 to be approximately facing the binocular vision system, with the side of the metal plate on which the target is placed facing the binocular vision system.

[0028] In this embodiment, a collaborative robot of model UR5 is used. The robot's state can be set to the teaching mode in its teach pendant, and then the robot's joint arm can be dragged by hand to move.

[0029] In this embodiment, the thickness of the metal plate should be greater than 5mm, and the area of ​​the plate should be not less than 80mm×100mm; the laser tracker should be placed at a distance of no more than 2m from the collaborative robot.

[0030] Step S3: Arrange the laser tracker reflective target ball and visual target ball on the two target seats set up in step S1, in random order, and guide the optical path of the laser tracker to the reflective target ball.

[0031] In this embodiment, the GTS series laser tracker from Shenzhen Zhongtu Instrument Co., Ltd. is used. The operation of guiding the optical path of the laser tracker to the reflective target ball is based on the operation procedure of the laser tracker. The reflective target ball is initially located at the bird's nest point of the laser tracker. The reflective target ball is manually held and the prism of the target ball is kept aligned with the optical path of the laser tracker. It is moved stably to the set target base, and the light is kept continuous during the process. In this way, the optical path of the laser tracker is guided to the reflective target ball arranged on a target base.

[0032] Step S4: Using the controller of the collaborative robot, control the robot to perform translational movements in the X, Y, and Z directions based on its end-effector TCP coordinate system, and analyze the range of the measurement space of the binocular vision system that the visual target ball can cover within the robot's range of motion.

[0033] Step S5: Sample the range of the binocular vision system measurement space that the visual target ball can cover under the recorded robot motion, and uniformly sample 60-90 points;

[0034] In this embodiment, the range of the binocular vision system measurement space that the visual target ball can cover under robot motion is recorded as 500×500×500mm. 3 Sixty points were sampled evenly in this space.

[0035] Step S6: According to the robot motion sampling points set in step S5, control the robot to move to each point in sequence. Stop the movement at each point. The laser tracker and the binocular vision system measure the spatial coordinates of the reflective target ball and the visual target ball respectively, and save the measurement data at each sampling point.

[0036] For the operation of controlling the robot to move sequentially to each point, the motion used is translational motion relative to the TCP coordinate system;

[0037] Step S7: Process and analyze the measurement data of the laser tracker and the binocular vision system at each sampling point, and calculate the rotational transformation relationship between the coordinate systems of the laser tracker and the binocular vision system.

[0038] For the calculation of the rotation transformation relationship, the first sampling position is used as the initial reference position. The measurement results of the reflective target ball and the visual target ball at other sampling positions are subtracted from the corresponding measurement results at the initial reference position to obtain the translation vector of the reflective target ball and the visual target ball at each sampling position relative to the initial position. Based on the correspondence between the translation vectors, the rotation transformation relationship between the laser tracker coordinate system and the binocular vision coordinate system is calculated.

[0039] To solve the rotation transformation relationship, the rotation transformation relationship between the laser tracker coordinate system and the binocular vision coordinate system is calculated based on the correspondence between each translation vector. The algorithm used is a solution method based on singular value decomposition.

[0040] Step S8: Calculate the spatial error at each sampling point based on the rotational transformation relationship between the laser tracker and the binocular vision system coordinate system calculated in step S7 and the measurement data at each sampling point saved in step S6.

[0041] For the calculation of spatial error, based on the translation vector of each sampling point relative to the initial position obtained by the laser tracker, and according to the obtained rotation transformation relationship between the laser tracker and the binocular vision system, the translation vector of each sampling point relative to the initial position sampling point in the binocular vision coordinate system is calculated. The theoretical coordinate result at each sampling point is obtained by adding the measurement coordinate of the visual target ball at the initial position by the binocular vision system to the translation vector measured by the laser tracker and transformed by coordinate transformation. The theoretical coordinate result at each sampling point is obtained by subtracting the measurement result of the visual target ball at each sampling point by the binocular vision system.

[0042] Based on the steps described above in this embodiment, the measurement error at each sampling point in the measurement space of the binocular vision system can be evaluated, thereby providing an effective means for calibrating the spatial error distribution of the binocular vision measurement system.

[0043] This invention provides a method for spatial error calibration of a binocular vision system. Many methods and approaches exist for implementing this technical solution; the above description is merely a preferred embodiment of the invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this invention, and these improvements and modifications should also be considered within the scope of protection of this invention. All components not explicitly stated in this embodiment can be implemented using existing technologies.

Claims

1. A method for spatial error calibration of a binocular vision system, characterized in that, Includes the following steps: Step 1: Fix two target holders on a metal plate that can be connected to the end flange of the collaborative robot. These are used to place the reflective target ball that comes with the laser tracker and the vision target ball that corresponds to the binocular vision system, respectively. Fix the metal plate to the end of the collaborative robot. Step 2: Place the laser tracker and the binocular vision system on the same side of the collaborative robot, and adjust the attitude of the metal plate fixed to the end of the collaborative robot in Step 1 to be approximately facing the binocular vision system, and the side of the metal plate on which the target is placed faces the binocular vision system. Step 3: Arrange the reflective target ball and the visual target ball on the two target seats set up in Step 1, in random order, and guide the optical path of the laser tracker to the reflective target ball; Step 4: Using the controller of the collaborative robot, control the robot to perform translational movements in the X, Y, and Z directions with the end-effector TCP coordinate system as the reference. Observe and record the coverage of the visual target ball at the end of the robot in the measurement space of the binocular vision system within the robot's movement range. Step 5: Sample the range of the binocular vision system measurement space that the visual target ball can cover under the recorded robot motion. Step 6: Based on the robot motion sampling points set in Step 5, control the robot to move to each point in sequence. Stop moving at each point. The laser tracker and binocular vision system measure the spatial coordinates of the reflective target ball and the visual target ball respectively, and save the measurement data at each sampling point. Step 7: Process and analyze the measurement data of the laser tracker and the binocular vision system at each sampling point, and calculate the rotational transformation relationship between the coordinate systems of the laser tracker and the binocular vision system. Step 7 includes: taking the first sampling position as the initial reference position, and recording the coordinates of the reflective target ball and the visual target ball at the initial reference position, measured by the laser tracker and the binocular vision system, as follows: and The coordinates of the reflective target sphere and the visual target sphere at other positions i, measured by the laser tracker and the binocular vision system, are denoted as follows: and ,in Subtract the corresponding measurement results at the initial reference position from the measurement results of the reflective target ball and visual target ball at other sampling positions to obtain the translation vectors of the reflective target ball and visual target ball relative to the initial position at each sampling position, denoted as follows: and Based on the correspondence between each translation vector, the rotational transformation relationship between the laser tracker coordinate system and the binocular vision coordinate system is calculated; Step 8: Calculate the spatial error at each sampling point based on the rotation transformation relationship obtained in Step 7 and the measurement data at each sampling point saved in Step 6. Step 8 includes: based on the translation vector of the reflective target ball at sampling position i relative to the initial position. The rotational transformation relationship R between the laser tracker coordinate system and the binocular vision coordinate system is obtained, with reference to the measured values ​​of the laser tracker. Calculate the standard reference value of the translation vector of the visual target ball at position i relative to the initial position visual target ball in the binocular vision coordinate system. ,in The standard reference value for the visual target ball at position i is Therefore, the measurement error of the binocular vision system at position i ,in The measurement value of the visual target point at this location by the binocular vision system is given; the spatial error of the binocular vision system at each sampling point location is calculated. .

2. The method according to claim 1, characterized in that, In step 1, the target base, reflective target ball, and visual target ball are all standard parts, and their dimensions are all standard sizes of 1.5 inches. The diameter of the marking circle of the visual target ball is 12 mm.

3. The method according to claim 2, characterized in that, In step 1, the metal plate has a thickness greater than 5 mm and an area of ​​not less than 80 mm × 100 mm.

4. The method according to claim 3, characterized in that, In step 2, the laser tracker is positioned at a distance of no more than 2m from the collaborative robot.

5. The method according to claim 4, characterized in that, In step 6, the controlled robot moves sequentially to each point, and the motion is a translational motion relative to the TCP coordinate system.

6. The method according to claim 5, characterized in that, In step 7, based on the least squares algorithm, the rotational transformation relationship between the laser tracker coordinate system and the binocular vision coordinate system is calculated, and the calculation expression is: , where R is the obtained rotational transformation relationship.

7. A storage medium, characterized in that, It stores a computer program or instructions that, when executed, implement the method as described in any one of claims 1 to 6.

Images