A robot hand-eye matrix and tcp joint calibration method and device

Abstract

The application belongs to the technical field of robot measurement, and discloses a robot hand-eye matrix and TCP joint calibration method and equipment, which uniformly models the whole coordinate chain of the calibration process, comprehensively considers hand-eye errors, TCP errors and joint errors of the robot, establishes a joint synchronization model, the model is more in line with the actual situation, can identify multiple source errors synchronously, avoids the influence of robot kinematics errors on the hand-eye calibration and TCP calibration accuracy, and effectively improves the calibration accuracy. Meanwhile, the method uses Lie algebra to solve the problem that the rotation matrix is not closed to addition in the optimization process, realizes synchronous iterative optimization of rotation components and translation components, and finally obtains accurate hand-eye matrix and TCP position relationship.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of robot measurement, and more particularly relates to a robot hand-eye matrix and TCP joint calibration method and device. BACKGROUND

[0002] Visual measurement sensors provide robots with the necessary perception capabilities, enabling them to perform more efficiently and intelligently in automated and intelligent applications. Thanks to the low cost, non-contact and information-rich advantages of visual sensors, visual guidance technology is widely used in robotic automation, providing an effective solution for automated assembly processes and improving assembly quality. However, for six-degree-of-freedom serial industrial robots, due to their low structural stiffness and end-of-life deformation, the motion accuracy of such robots is generally poor, and poor motion accuracy restricts the upper limit of the accuracy of the robot vision system.

[0003] In order to obtain the precise positional relationship between the robot and the visual measurement system, hand-eye calibration is necessary. Existing calibration algorithms assume that the robot has no motion error, which causes error accumulation in the input data stage. SUMMARY

[0004] In view of the above defects or improvement needs of the prior art, the present application provides a robot hand-eye matrix and TCP joint calibration method and device, which aims to solve the problem of low calibration accuracy caused by the assumption that there is no motion error in the existing calibration method.

[0005] To achieve the above-mentioned purpose, according to one aspect of the present application, a robot hand-eye matrix and TCP joint calibration method is provided, which comprises the following steps:

[0006] (1) Install an operating tool at the end of the robot, move the robot to any measurable pose, and record the robot pose data at this time E B T, and measure the three-dimensional coordinates of TCP at this time using a measuring device, and then collect a predetermined number of robot pose and joint rotation angle data;

[0007] (2) The hand-eye matrix is defined as The homogeneous coordinates of TCP in the robot end coordinate system E p and the homogeneous coordinates in the external photogrammetry system S p satisfy the relationship Split the homogeneous matrix into a rotation matrix and a translation matrix and combine them, so For n measurements, use the Kronecker product and matrix vectorization method to construct the least squares linear equation, and solve x based on the pseudo-inverse to obtain andE The coarse identification result of p is denoted as and E p r ;

[0008] (3) Error matrices E are added to the hand-eye matrix, TCP position vector, and each joint of the robot for compensation. Each error matrix is ​​regarded as a left-multiplied perturbation e added to the current value by a six-dimensional Lie algebra vector δξ. δξ^ A joint calibration and fine identification model based on Lie algebra is constructed for hand-eye-TCP-joint joints.

[0009] (4) Based on the precise identification model, the hand-eye, robot joint, and TCP error matrices can be obtained through a nonlinear optimization iteration process. The obtained error matrix is ​​then used to update the hand-eye matrix and TCP vector, resulting in the calibration hand-eye matrix. and TCP position vector

[0010] Furthermore, the least squares linear equation is Ax = b, where,

[0011]

[0012] in and E p represents the hand-eye matrix parameters and TCP position to be determined.

[0013] Furthermore, the formula for compensation is:

[0014]

[0015] Furthermore, for m measurements, the following optimization function is constructed:

[0016]

[0017] Furthermore, the mathematical expression for the precise identification model is:

[0018]

[0019] Where δξ=[δρδφ] T ∈se(3) and δξ bs ,δξ 1～n ,δξ me These represent the six-dimensional Lie algebra error vectors corresponding to the hand-eye relationship, TCP position, and each joint of the robot, respectively. This represents the transformation matrix from the robot's base coordinate system to the vision sensor coordinate system; denotes the transformation matrix of the robot from the j th joint to the j-1 th joint; the robot joint and TCP error matrices are respectively

[0020] Further, the measuring device is a laser tracker-reflection target ball, a photogrammetry system-target point, or a monocular camera-chessboard.

[0021] The application further provides a robot hand-eye matrix and TCP joint calibration system, which comprises a memory and a processor, the memory stores a computer program, and the processor executes the computer program to execute the robot hand-eye matrix and TCP joint calibration method as described above.

[0022] The application further provides a computer readable storage medium, which stores machine executable instructions, and the machine executable instructions, when called and executed by a processor, cause the processor to implement the robot hand-eye matrix and TCP joint calibration method as described above.

[0023] Overall, compared with the prior art, the robot hand-eye matrix and TCP joint calibration method and device provided by the application mainly have the following beneficial effects:

[0024] 1. The entire coordinate chain of the calibration process is uniformly modeled, the hand-eye error, the TCP error and the joint error of the robot are comprehensively considered, a joint synchronization model is established, the model is more in line with the actual situation, multi-source errors can be identified synchronously, the influence of the robot kinematic error on the hand-eye calibration and TCP calibration accuracy is avoided, and the calibration accuracy is effectively improved.

[0025] 2. The identification strategy combining coarse calibration and fine calibration is adopted, the result of coarse calibration is used as the initial value of fine calibration, the search domain is effectively reduced, the iteration time is saved, and the optimization efficiency is improved.

[0026] 3. The homogeneous error matrix is mapped to the Lie algebra space, the problem of non-closed addition in the rotation matrix iteration process is solved, the translation vector and the rotation matrix are considered in the iteration at the same time, and the error transmission and accumulation caused by the traditional step-by-step optimization are avoided.

[0027] 4. In the error distribution of the measured 30 groups of sampling point three-dimensional coordinates (X, Y, Z) after coarse and fine calibration, the average error of coarse calibration in the x direction is 0.692 mm, the average error of fine calibration is 0.096 mm; the average error of coarse calibration in the y direction is 1.941 mm, the average error of fine calibration is 0.104 mm; the average error of coarse calibration in the z direction is 0.176 mm, and the average error of fine calibration is 0.132 mm. It can be seen that the calibration accuracy is high. BRIEF DESCRIPTION OF DRAWINGS

[0028] Figure 1 is a schematic diagram of coordinate systems of a robot vision system and conversion relationship thereof;

[0029] Figure 2 is a flowchart of a robot hand-eye matrix and TCP joint calibration method provided by the present application;

[0030] Figure 3 is an X-direction error schematic diagram obtained by an embodiment of the present application;

[0031] Figure 4 is a Y-direction error schematic diagram obtained by an embodiment of the present application;

[0032] Figure 5 is a Z-direction error schematic diagram obtained by an embodiment of the present application. DETAILED DESCRIPTION

[0033] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application. In addition, the technical features involved in each embodiment of the present application described below can be combined with each other as long as they do not conflict with each other.

[0034] The vision measurement sensor provides the necessary sensing capability for the robot, so that it can perform higher efficiency and intelligence level in automation and intelligent application. For the robot vision measurement sensor system, in order to accurately convert the pose information from the measurement system to the robot end, it is necessary to determine the pose parameters between the measurement sensor and the robot and between the robot and the tool center point (TCP) through the calibration process.

[0035] The present application provides a robot hand-eye matrix and TCP joint calibration method, which uses Lie algebra to solve the problem of non-closed addition of rotation matrix in the optimization process, realizes synchronous iterative optimization of rotation component and translation component, and finally obtains accurate hand-eye matrix and TCP position relationship. At the same time, the calibration method can effectively eliminate the calibration error caused by the joint error of the robot, and further obtain higher calibration accuracy of the hand-eye matrix and TCP position vector, which has important significance.

[0036] Specifically, the method uniformly models the coordinate chain of the calibration process, comprehensively considers the joint error, hand-eye error and TCP error of the robot, establishes a joint calibration model, and proposes a corresponding parameter synchronous identification method, which can obtain high-precision calibration results, and has great significance for promoting the application of robot equipment, improving operation efficiency and quality.

[0037] The system to which the present application is directed is a robot system under visual guidance, which is conventionally arranged as shown in the figure, comprising a robot, a 3D measuring device and an operating tool. The 3D measuring device is kept relatively fixed with the robot, and the operating tool is fixed at the end of the robot, the three-dimensional position of which can be directly measured by the 3D measuring device or obtained through image processing means, and the measured origin is denoted as TCP (Tool Center Point). Figure 1

[0038] The calibration coordinate system involves a robot base coordinate system {B}, a robot end coordinate system {E}, an operating tool coordinate system {M} and a measuring device coordinate system {S}, and the relationship of each coordinate system is shown in the figure. Figure 1

[0039] The method mainly comprises the following steps:

[0040] Step one: install the operating tool at the end of the robot, move the robot to any measurable pose, and record the robot pose data at this time E B T.

[0041] Step two: measure the three-dimensional coordinates of TCP at this time by using the measuring device.

[0042] Step three: repeat steps one to two, and collect a predetermined number of robot poses and the rotation angle data of each joint.

[0043] Among them, attention should be paid to ensuring that the robot poses of each group of data are different axes during calibration; the number of collected data is at least 15 groups.

[0044] Step four: the hand-eye matrix is defined as The homogeneous coordinates of TCP in the robot end coordinate system E p and the homogeneous coordinates of in the external photogrammetry system S p satisfy the relationship The homogeneous matrix is decomposed into a rotation matrix and a translation matrix and combined, and there are For n times of measurement, the Kronecker product and matrix vectorization method are used to construct the least square linear equation Ax = b, and the following is obtained:

[0045]

[0046] Among them and E p represent the hand-eye matrix parameters and TCP position to be solved, and the pseudo-inverse solution x can be obtained and E The coarse recognition result of p is denoted as and E p r . ​​

[0047] Step five, add error matrix E to each of hand-eye matrix, TCP position vector and each joint of robot respectively to compensate, the corresponding formula is:

[0048]

[0049] For m measurements, the following optimization function can be constructed:

[0050]

[0051] Step six, each error matrix is regarded as a left perturbation e δξ^ based on Lie algebra, a joint calibration precision identification model is constructed, and the mathematical expression of the precision identification model is:

[0052]

[0053] Where δξ=[δρδφ] T ∈se(3) and δξ bs ,δξ 1～n ,δξ me respectively represent the corresponding six-dimensional Lie algebra error vectors of hand-eye relationship, TCP position and each joint of robot; represents the transformation matrix from the base coordinate system of the robot to the vision sensor coordinate system; represents the transformation matrix from the jth joint to the j-1th joint of the robot; the robot joint and TCP error matrices are respectively

[0054] Step seven, based on the precision identification model, the hand-eye matrix, robot joint and TCP error matrix can be obtained through a nonlinear optimization iteration process, and then the obtained error matrix is used to update the hand-eye matrix and TCP vector to obtain the calibration result hand-eye matrix and TCP position vector

[0055] Where the applicable measurement equipment type is all instruments that can obtain the three-dimensional coordinates of the object, such as laser tracker-reflection target ball, photogrammetry system-target point, monocular camera-chessboard, etc.

[0056] Please refer to Figure 3 , Figure 4 and Figure 5In one embodiment, in the error distribution of the measured 30 groups of three-dimensional coordinate (X, Y, Z) of the sampling points after coarse calibration and fine calibration, the average error of coarse calibration in the x direction is 0.692 mm, the average error of fine calibration is 0.096 mm; the average error of coarse calibration in the y direction is 1.941 mm, the average error of fine calibration is 0.104 mm; the average error of coarse calibration in the z direction is 0.176 mm, the average error of fine calibration is 0.132 mm, and it can be seen that the calibration accuracy is higher.

[0057] The application further provides a robot hand-eye matrix and TCP joint calibration system, which comprises a memory and a processor, the memory stores a computer program, and the processor executes the computer program to execute the robot hand-eye matrix and TCP joint calibration method as described above.

[0058] The application further provides a computer readable storage medium, which stores machine executable instructions, and the machine executable instructions, when called and executed by a processor, make the processor realize the robot hand-eye matrix and TCP joint calibration method as described above.

[0059] Those skilled in the art can easily understand that the above description is only a preferred embodiment of the application, and is not intended to limit the application, and any modification, equivalent replacement and improvement made within the spirit and principle of the application shall be included in the protection scope of the application.

Claims

1. A method for joint calibration of robot hand-eye matrix and TCP, characterized in that: The method comprises the following steps: (1) Install an operating tool at the end of the robot, move the robot to any measurable position, record the robot position data at this time and measure the three-dimensional coordinates of the TCP at this time using a measuring device, thereby collecting a predetermined number of robot position and rotation angle data of each joint; (2) The hand-eye matrix is defined as , the homogeneous coordinates of TCP in the robot end coordinate system and the homogeneous coordinates of TCP in the external photogrammetry system satisfy the relationship , the homogeneous matrix is decomposed into the combination of the rotation matrix and the translation matrix, and there is , for times of measurement, the Kronecker product and matrix vectorization method are used to construct the least square linear equation, and the pseudo-inverse is solved based on , the coarse recognition results of and can be obtained, and are recorded as and ; (3) Add error matrix at hand-eye matrix, TCP position vector and each joint of robot respectively Each error matrix is regarded as a six-dimensional Lie algebra vector A left disturbance added to the current value Construct a hand-eye-TCP-joint joint calibration precision identification model based on Lie algebra (4) Based on the fine recognition model, the hand-eye, robot joint and TCP error matrix can be obtained through a nonlinear optimization iteration process , and then the obtained error matrix is used to update the hand-eye matrix and TCP vector to obtain the calibration result hand-eye matrix and TCP position vector , .

2. The robot hand-eye matrix and TCP joint calibration method of claim 1, wherein: The least squares linear equation is wherein, wherein and denote the hand-eye matrix parameters and TCP position to be solved.

3. The robot hand-eye matrix and TCP joint calibration method of claim 2, wherein: The compensation corresponding formula is: 。 4. The robot hand-eye matrix and TCP joint calibration method of claim 3, wherein: For For the secondary measurement, the following optimization function is constructed: 。 5. The robot hand-eye matrix and TCP joint calibration method of claim 4, wherein: The mathematical expression of the fine recognition model is: in and , These represent the six-dimensional Lie algebra error vectors corresponding to the hand-eye relationship, TCP position, and each joint of the robot, respectively. This represents the transformation matrix from the robot's base coordinate system to the vision sensor coordinate system; Indicates that the robot starts from the first joint to the The transformation matrix of the joint; The hand-eye error matrix, robot joint, and TCP error matrix are respectively .

6. The robot hand-eye matrix and TCP joint calibration method of any one of claims 1-5, wherein: The measuring device is a laser tracker-reflective target ball, a photogrammetry system-target point or a monocular camera-chessboard.

7. A robot hand-eye matrix and TCP joint calibration system, characterized in that: The system comprises a memory and a processor, the memory stores a computer program, and the processor executes the computer program to execute the robot hand-eye matrix and TCP joint calibration method in any one of claims 1-6.

8. A computer-readable storage medium, characterized in that: The computer readable storage medium stores machine executable instructions, when the machine executable instructions are called and executed by the processor, the machine executable instructions cause the processor to implement the robot hand-eye matrix and TCP joint calibration method in any one of claims 1-6.

Images