A hand-eye system calibration method without fixed zero position

Abstract

The application relates to a fixed-zero-free hand-eye system calibration method, which comprises the following steps: S0: obtaining an actuator, a camera, a target A and a target B, the target A and the target B being provided with feature points; S1: in the calibration stage, the target A is fixed at the end of the actuator, the actuator is moved, the target A is recognized through the camera, the image of the calibration feature points on the target A is collected, and a rotation matrix of a camera coordinate system and an actuator end coordinate system is obtained; S2: in the equipment running stage, the target B is fixed at the end of the actuator, when the actuator is started, the target B is recognized through the camera, the image of the feature points on the target B is collected, and a translation vector of the camera coordinate system and the actuator end coordinate system is obtained; and S3: the fixed-zero-free hand-eye system calibration is completed. Compared with the prior art, the application reduces the requirements of the traditional hand-eye calibration method on the hand-eye system, has the advantages of more flexible hand-eye calibration, high calibration precision and low cost and the like.

Description

Technical Field

[0001] This invention relates to the field of calibration technology, and in particular to a calibration method for a hand-eye system without a fixed zero point. Background Technology

[0002] As the automation level of industrial product production and testing processes gradually increases, one of the common automation modes at present is the hand-eye mode, which involves the collaboration between CNC mechanical actuators and vision measurement devices such as line structure light sensors. In this mode, the vision measurement device is responsible for acquiring the spatial feature information of the product, and then converting the data into the measurement space of the mechanical actuator based on the pose conversion parameters obtained from hand-eye calibration. This allows the mechanical actuator to perform specific product processing steps based on the existing spatial information, such as product gripping and cutting.

[0003] Hand-eye calibration is a crucial step in achieving the aforementioned automated process. Conventional hand-eye calibration methods typically establish a base coordinate system using the fixed zero point of the mechanical actuator as a reference point, and then indirectly calculate the hand-eye pose transformation parameters using this base coordinate system as a medium. However, for mechanical actuators without a fixed zero point, conventional hand-eye calibration methods are ineffective due to the lack of a fixed medium. Therefore, there is a need for a hand-eye system calibration method that can solve the problem of calibrating hand-eye pose transformation parameters between mechanical actuators without a fixed zero point and line structured light vision measurement equipment. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art, which is that it is difficult to play an effective role due to the lack of a fixed medium and is not applicable to the calibration of mechanical actuators without a fixed zero position, and to provide a calibration method for hand-eye systems without a fixed zero position.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A method for calibrating a hand-eye system without a fixed zero point includes the following steps:

[0007] S0: Acquire the actuator, camera, target A and target B, wherein target A and target B are provided with feature points;

[0008] S1: During the calibration phase, target A is fixed at the end of the actuator, the actuator is moved, the camera identifies target A and acquires images of the calibrated feature points on target A, and the rotation matrix between the camera coordinate system and the end coordinate system of the actuator is obtained.

[0009] S2: During the equipment operation phase, target B is fixed at the end of the actuator. When the actuator starts, the camera identifies target B and collects feature point images on target B to obtain the translation vector between the camera coordinate system and the coordinate system at the end of the actuator.

[0010] S3: Based on the rotation matrix and translation vector between the camera coordinate system and the actuator end coordinate system, complete the calibration of the hand-eye system without a fixed zero position.

[0011] Furthermore, the target A has a concentric circle pattern, the center of the concentric circle is a feature point, the inner circle of the target A is white, and the outer circle is black; the target B is based on the target A and has multiple feature circles on the ring, the center of the feature circle is a feature point.

[0012] Furthermore, in S1, the identification of target A includes the following steps:

[0013] S101: Acquire an image of target A at the end of the actuator using a camera, and preprocess the image;

[0014] S102: Identify image contours, set corresponding threshold conditions based on the number of points and area of ​​the image contours, filter the contours according to the threshold conditions, and fit the filtered contours to an ellipse.

[0015] S103: Classify the fitted ellipse into inner and outer attributes, then divide the ellipse into pairs based on the distance between the center points of the inner and outer ellipses, and obtain the ellipse pairs whose center point distance meets the threshold condition.

[0016] S104: For the obtained ellipse pair, the target A is identified by positioning it according to the true projection of the center point of the concentric circle.

[0017] Furthermore, in S102, the equation of the fitted ellipse for the contour is:

[0018] Ax 2 +By 2 +Cxy+Dx+Ey+F=0

[0019] In the formula, A, B, C, D, E, and F are the ellipse parameters to be solved, and the calculation expressions for each ellipse parameter are as follows:

[0020]

[0021] In the formula: a is the length of the circumscribed rectangle of the ellipse, b is the length of the circumscribed rectangle of the ellipse, θ is the tilt angle, and c(x,y) are the coordinates of the center point.

[0022] Furthermore, the identification of target B involves detecting the center of the concentric circles according to the identification steps of target A, obtaining a projected image of target B; performing an affine transformation on the projected image to standardize its shape, resulting in a concentric circle image of a preset size; detecting and fitting the feature circles within the annulus based on the radius of the concentric circles; then converting the image of target B into a projected image to obtain the projected coordinates of the center point of the feature circles, thus completing the identification of target B.

[0023] Furthermore, in step S1, the specific steps for obtaining the rotation matrix between the camera coordinate system and the actuator end effector coordinate system include:

[0024] S111: During the calibration stage, target A is fixed at the end of the actuator, and the actuator is driven to move along the direction of the laser plane so that the center point of target A coincides with the center line of the light stripe on its surface.

[0025] S112: Switch the laser to acquire image pairs of the relatively stationary target A before and after the light stripe projection;

[0026] S113: Identify the target A image before the light stripe projection, and obtain the three-dimensional coordinates of the center point of the concentric circle in the coordinate system of the actuator end; and obtain the three-dimensional coordinates of the center point of the concentric circle in the camera coordinate system based on the target A image after the light stripe projection.

[0027] S114: The actuator end passes through each laser plane in sequence and acquires the corresponding images. The three-dimensional coordinate set of the target A center point in the actuator end coordinate system and the camera coordinate system is obtained. Based on the two three-dimensional coordinate sets, the rotation matrix of the actuator end coordinate system and the camera coordinate system is obtained through the SVD decomposition algorithm.

[0028] Furthermore, in S114, the objective function for obtaining the rotation matrix of the actuator end-effector coordinate system and the camera coordinate system is:

[0029]

[0030] In the formula, R cb t is the rotation matrix from the camera coordinate system to the actuator end effector coordinate system; cb_x ω is the translation vector between the camera coordinate system and the zero position of the actuator end-effector coordinate system at system startup; j P represents the weight of each corresponding point in two point sets; the value is positive and is usually set to 1. j Q represents the three-dimensional coordinates of the center point of target A in the camera coordinate system. j Let be the three-dimensional coordinates of the center point of target A in the coordinate system of the actuator end effector.

[0031] Furthermore, in step S2, the specific steps for obtaining the translation vector between the camera coordinate system and the actuator end effector coordinate system include:

[0032] S201: During the system operation phase, target B is fixed at the end of the actuator. Each time the actuator is started, an image of target B is captured by a camera, and a world coordinate system is established on target B.

[0033] S202: Obtain the sub-pixel coordinates of the center points of each feature circle within the concentric ring on target B, as well as the three-dimensional coordinates of the center points of each feature circle in the world coordinate system and the camera coordinate system.

[0034] S203: Based on the sub-pixel coordinates of the center points of multiple sets of the same feature circles and their 3D coordinates in the world coordinate system and camera coordinate system, the rotation matrix and translation vector between the camera coordinate system and the world coordinate system are obtained through the PnP pose calculation algorithm.

[0035] S204: Using the three-dimensional coordinates of the center point of the concentric circle of target B in the coordinate system of the actuator end effector as the origin coordinates of the world coordinate system, the current translation vector between the camera coordinate system and the world coordinate system is calculated based on the rotation matrix and translation vector between the camera coordinate system and the world coordinate system.

[0036] Furthermore, in S204, the objective function for obtaining the rotation matrix and translation vector between the camera coordinate system and the world coordinate system is:

[0037]

[0038] In the formula, ρ is the scaling factor, M is the camera's intrinsic parameter matrix, and both ρ and M are known quantities; R cw t is the rotation matrix between the camera coordinate system and the world coordinate system; cw This is the translation vector between the camera coordinate system and the world coordinate system; The sub-pixel coordinates of the center point of the feature circle. The three-dimensional coordinates of the center point of the feature circle in the camera coordinate system. The coordinates of the center point of the characteristic circle in the world coordinate system are given.

[0039] Furthermore, in S201, the origin of the world coordinate system is the center point of the concentric circles on target B, and the concentric circles are the world coordinate system Z. W The plane = 0, the number of characteristic circles on the ring is at least 4.

[0040] Compared with the prior art, the present invention has the following advantages:

[0041] (1) In this scheme, the direction of the coordinate system at the end of the mechanical actuator remains unchanged, so the rotation matrix between the coordinate system at the end of the actuator and the camera coordinate system remains unchanged. The translation vector between the two coordinate systems is not fixed each time the task is executed because the zero point is not fixed. The rotation matrix and translation vector of the hand-eye system are solved separately. By collecting the feature points at the end of the actuator during the calibration stage, the rotation matrix between the camera coordinate system and the coordinate system at the end of the actuator is obtained. At each system startup, the initial position of the end of the actuator and the translation vector of the camera coordinate system are determined to complete the calibration of the hand-eye system without a fixed zero point. This realizes the collaboration between the line structured light vision measurement equipment and the mechanical actuator without a fixed zero point, reduces the requirements of the traditional hand-eye calibration method on the hand-eye system, makes the hand-eye calibration more flexible, has high calibration accuracy and low cost, and provides strong support for the automation upgrade and transformation of industrial product production, testing and other processes.

[0042] (2) This scheme calculates the rotation matrix and translation vector of the hand-eye system separately and settles them in stages. Before the system is used, the end of the actuator is calibrated using a target A with concentric circles to obtain the fixed value of the rotation matrix of the coordinate system of the end of the actuator relative to the coordinate system of the camera. Then, before each system start-up, the translation vector of the end of the actuator relative to the coordinate system of the camera is determined by using a target B set at the end of the actuator and multiple feature circles on the target B, and a picture before the actuator moves, thus completing the calibration of the hand-eye system.

[0043] Taking into full account that the mechanical actuator will not return to a fixed position after each task, the zero position of its end point is not fixed each time it starts. Thus, the parallel vector between the actuator and the camera coordinate system before each movement is obtained independently, the actuator is calibrated, and the rotation matrix is ​​confirmed separately. This avoids the tediousness of repeated calibration in subsequent use, simplifies the process, and improves calibration efficiency. Attached Figure Description

[0044] Figure 1 This is a schematic diagram illustrating the application scenario architecture of a hand-eye system calibration method without a fixed zero position according to an embodiment of the present invention.

[0045] Figure 2 The diagram shown is a schematic diagram of target A provided for solving the rotation matrix in a hand-eye system calibration method without a fixed zero position according to the present invention.

[0046] Figure 3 The diagram shown is a schematic diagram of target B provided for solving the translation vector in a hand-eye system calibration method without a fixed zero position according to the present invention.

[0047] Figure 4The diagram shows a specified motion designed for solving a rotation matrix in one embodiment of a hand-eye system calibration method without a fixed zero position according to the present invention.

[0048] Figure 5 The image shows the effect of a hand-eye system calibration method without a fixed zero position according to an embodiment of the present invention, in which the center line of the light stripe passes through the center of the target A.

[0049] Figure 6 This is an example of a pair of relatively still images after the light stripes are projected, which is a hand-eye system calibration method without a fixed zero position according to an embodiment of the present invention.

[0050] Figure 7 This is an example of a pair of relatively still images collected before the projection of light stripes, which is used in one embodiment of the hand-eye system calibration method without a fixed zero position according to the present invention.

[0051] Figure 8 This is a schematic diagram showing the world coordinate system established on the target B plane fixed to the end of the actuator in one embodiment of the hand-eye system calibration method without fixed zero position according to the present invention.

[0052] Figure 9 The flowchart shown is a hand-eye system calibration method without a fixed zero position according to the present invention.

[0053] Figure 10 The flowchart shown is for identifying target A according to the present invention;

[0054] In the diagram: 1. Guide rail, 2. Actuator, 3. Camera, 4. Laser, 5. Laser plane. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0056] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0057] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0058] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship in which the product of this invention is usually placed during use. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0059] It should be noted that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0060] Furthermore, terms such as "horizontal" and "vertical" do not imply that components must be absolutely horizontal or suspended, but rather that they can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0061] Example 1

[0062] This embodiment provides a hand-eye system calibration method without a fixed zero position, used to calculate hand-eye pose conversion parameters between a mechanical actuator without a fixed zero position and a line structured light vision sensor measurement device. In this specification, [the method is described in the original text]. Figure 1 The calibration method for a hand-eye system without a fixed zero position is illustrated using an example application scenario. The application scenario includes a guide rail 1 and actuator 2 of a CNC execution device, a camera 3, a laser 4, and a laser plane 5 generated by the laser. The internal parameters, distortion coefficients of the camera, and the plane equation coefficients of the laser plane have all been precisely calibrated. If there are multiple cameras, the pose transformation parameters between the coordinate systems of each camera have also been precisely calibrated. The actuator 2 has no fixed zero position; the position of its end point at each startup is the current zero position of the mechanical execution device.

[0063] This invention provides two different targets, A and B, which are fixed to the end of the mechanical actuator 2 during the solution of the rotation matrix and translation vector parameters of the hand-eye system to increase different types of feature information;

[0064] The target A is fixed to the end of the mechanical actuator 2, and the mechanical actuator 2 is controlled to perform a specified movement to collect images of the end-point calibration feature points. The rotation matrix between the camera coordinate system and the end-point coordinate system of the actuator is calculated by a preset algorithm.

[0065] The target B is fixed to the end of the mechanical actuator 2. When the actuator is started, the camera 3 acquires feature point images of the end of the actuator 2 at the starting position. The translation vector between the camera coordinate system and the coordinate system of the actuator end is calculated using a preset algorithm. Thus, the calculation of the pose transformation parameters of the hand-eye system without a fixed zero position is realized.

[0066] like Figure 9 As shown, a hand-eye system calibration method without a fixed zero point according to the present invention includes the following steps:

[0067] Step S11: Provide target A and target B, as well as the corresponding identification method.

[0068] Specifically, this invention provides corresponding targets A and B for the different features required by the rotation matrix solution method and translation vector solution method included in the hand-eye system calibration method without a fixed zero position, such as... Figure 2 and Figure 3 As shown. Both targets use concentric circles (white inside, black outside) as their basic unit for easy identification and localization. Target A has a simpler structure and is mainly used for solving the rotation matrix in the hand-eye system, while target B adds a series of feature circles in the annular region to provide geometric information for calculating the translation vector. The identification of both targets is based on concentric circle detection, and the specific identification method is as follows.

[0069] like Figure 10 As shown, for target A, before the rotation matrix solution begins, it is fixed to the end of actuator 2 and moves together with actuator 2. Each camera captures the image of target A fixed to the end of actuator. The image is preprocessed by Gaussian filtering and other operations. Then, the contour is identified by the Devernay subpixel edge detection algorithm based on Canny. The contour is filtered according to the number of points, area and other conditions. The filtered contour is used as the object to fit an ellipse.

[0070] The general equation of an ellipse is given as:

[0071] Ax 2 +By 2 +Cxy+Dx+Ey+F=0

[0072] Where A, B, C, D, E, and F are the ellipse parameters to be solved. During fitting, the ellipse parameters are determined based on its circumscribed rectangle. Assuming the length of the circumscribed rectangle is *a*, the width is *b*, the inclination angle is *θ*, and the center point coordinates are *c(x,y)*, then the parameters of the ellipse equation can be calculated using the following formula:

[0073]

[0074] The solved ellipse is classified into inner and outer attributes. Since the inner circle of the target pattern provided by this invention is white and the outer circle is black, it can be determined as an inner or outer ellipse based on the gray values ​​inside and outside the ellipse outline. After the attribute classification is completed, the ellipse pairs are divided according to the distance between the center points of the inner and outer ellipses. The inner and outer ellipses whose center point distance meets the threshold condition are paired and retained.

[0075] For the detected ellipse pairs, the identification of target A can be completed by locating the true projection point of the center of the concentric circle;

[0076] It is important to note that concentric circles often exhibit perspective deviation during projective transformation. That is, concentric circles whose center points coincide in real space appear as two separate ellipses on the image plane after being captured from a non-frontal angle. This is due to the existence of eccentricity error. Therefore, relevant eccentricity error correction algorithms are needed to obtain the true projection center.

[0077] At this point, the identification of target A is complete.

[0078] For target B, its basic unit is also a concentric circle with a white inner circle and a black outer circle, so its recognition process is basically the same as that for target A. The difference is that the annular region of target B also has a series of feature circles to be identified. The specific method is as follows:

[0079] Based on the target A recognition process described above, the center of the basic concentric circle unit of target B is detected. Then, an affine transformation is performed on the image of target B to standardize the projected shape and restore it to a concentric circle image of preset specifications. Since the preset size of the target is known, the feature circles within the annular region are directly detected and fitted based on the annular radius value. Then, the projected image of the target is restored, thereby obtaining the projected coordinate values ​​of the feature circle center.

[0080] At this point, target B can be identified.

[0081] Step S12: Fix target A to the end of mechanical actuator 02, control it to make a specified movement, collect the image of the center feature point of the concentric circle at the end, and calculate the rotation matrix between the camera coordinate system and the coordinate system of the end of the actuator through a preset algorithm.

[0082] Specifically, the objective of step S12 is to calculate the rotation matrix from the camera coordinate system to the end effector coordinate system before the mechanical actuator formally performs its task. It's important to note that the mechanical actuator does not have a fixed zero point. Conventional calibration methods that establish a base coordinate system with a fixed zero point as the origin and use this as a medium to calculate hand-eye pose parameters are no longer applicable in this scenario. The solution to this problem is to calculate the rotation matrix and translation vector of the hand-eye system separately. Since the direction of the end effector coordinate system remains unchanged, its rotation matrix with the camera coordinate system remains constant. Only the non-fixed zero point causes the translation vector between the two coordinate systems to change with each task execution. The method for calculating the fixed rotation matrix value is described below.

[0083] The specified motion to be performed by the mechanical actuator 2 fixed to the target A is as follows: The control system of the mechanical actuator controls the end of the actuator 2 fixed to the target A to move along the direction of the laser plane, as shown below. Figure 4 As shown, this makes the center of target A at the end of actuator 2 coincide with the center line of the light stripes on its surface, achieving the effect shown. Figure 5 As shown. After the two coincide, the laser is switched on and off, and images of the relatively still light stripes before and after projection are acquired, as shown. Figure 6 and Figure 7 As shown.

[0084] The image of target A before the projection of the light stripes is processed using a concentric circle center detection algorithm to extract the center point of the concentric circles. Since this point coincides with the light stripes, it simultaneously satisfies the visual measurement model. Based on this, the three-dimensional coordinates of the center point in the camera coordinate system can be calculated. Following the above process, the end of actuator 2 sequentially passes through each laser plane to acquire corresponding image pairs and reconstruct the center points. According to actual needs, actuator 02 can be controlled to reciprocate to obtain a sufficient number of center points.

[0085] Assume the camera coordinate system is O c -X c Y c Z c The set of center points of target A in the camera coordinate system, obtained by the above operations and calculations, is P = {P1, P2, Λ, P...} n The set of three-dimensional coordinates corresponding to it in the coordinate system of the actuator end effector is Q = {Q1, Q2, Λ, Q}. n}, where n represents the number of center points collected. The two sets of n-dimensional point sets P = {P1, P2, Λ, P} n} and Q = {Q1,Q2,Λ,Q n The three-dimensional spatial points P1 and Q1, P2 and Q2, ..., P contained in} n With Q nThese correspond to each other, meaning they describe the coordinate values ​​of the same center point in different spatial coordinate systems. The least squares method can be used to obtain the coordinates relative to the camera coordinate system O. c -X c Y c Z c and the coordinate system O of the actuator end effector b -X b Y b Z b The pose transformation relationship parameters between [R] cb t cb_x Related functions:

[0086]

[0087] Among them, R cb That is, the fixed value of the rotation matrix from the camera coordinate system to the actuator end effector coordinate system; t cb_x ω represents the translation vector between the camera coordinate system and the zero position of the actuator 2 end at the current system startup. This has no practical meaning here because the zero position of the actuator 2 end is not fixed each time the system starts; j This represents the weight of each corresponding point in two point sets, and is usually set to 1, with a positive value.

[0088] Using the above equation as the objective, the rotation matrix between the camera coordinate system and the end coordinate system of the actuator 2 can be calculated using a preset algorithm. The preset algorithm is the SVD decomposition algorithm, mainly used to solve for the rotation matrix and translation vector between two spatial coordinate systems.

[0089] Step S13: Fix the target B to the end of the mechanical actuator 2. Each time the actuator is started, the camera collects the feature point image of the end of the actuator 2 at the starting position, and calculates the current translation vector between the camera coordinate system and the coordinate system of the end of the actuator through a preset algorithm.

[0090] Specifically, since the mechanical actuator 2 does not return to a fixed position after each task is completed, the zero position of its end point is not fixed each time it starts. That is, the translation vector between its end execution coordinate system and the camera coordinate system is changing. Therefore, the present invention designs a corresponding solution method.

[0091] The method can be described as follows: Target B is fixed to the end of actuator 2. Each time the control system starts, a camera captures an image of target B at its end. The center point of target B is the current system zero position, with it as the origin, and its plane is Z. w Establish a world coordinate system O on the plane where 0 = 0. w -X w Y w Z w ,like Figure 8 As shown. The center of the feature circle of the annular region of target B is the usable feature point. The sub-pixel coordinates of each feature point in the annular region of the image are detected according to the method described in step S11. Its three-dimensional coordinates in the world coordinate system are The three-dimensional coordinates in the camera coordinate system are: The coordinates have the following transformation relationship:

[0092]

[0093] Where ρ is the scaling factor, M is the camera's intrinsic parameter matrix, and M is a known quantity; [R cw t cw ] is the rotation matrix and translation vector between the camera coordinate system and the world coordinate system. Its specific value can be calculated by applying a preset algorithm through the corresponding points of the center of the feature circle of the 2D-3D annular region. The preset algorithm is the PnP pose calculation algorithm.

[0094] Based on the rotation matrix and translation vector between the camera coordinate system and the world coordinate system, the three-dimensional coordinates of the center point of the concentric circle of target B in the camera coordinate system can be calculated. Its three-dimensional coordinates in the actuator end coordinate system are the origin coordinates. Based on this, the current translation vector from the camera coordinate system to the actuator end coordinate system can be directly calculated.

[0095] At this point, the pose parameters of the hand-eye system without a fixed zero position are calculated, specifically including the rotation matrix and translation vector from the camera coordinate system to the coordinate system of the mechanical actuator end effector. Based on this, collaboration between the line structured light vision measurement device and the mechanical actuator without a fixed zero position can be achieved. Taking a hand-eye system without a fixed zero position in an industrial setting as an example, the hand-eye pose parameters are calculated using the method proposed in this invention. The results are shown in Table 1. According to the parameters obtained in Table 1, the data measured by the vision measurement device can be successfully converted to the spatial coordinate system of the actuator, thereby enabling it to automatically perform relevant operations based on the measured data.

[0096] Compared with traditional hand-eye calibration methods, this invention can better adapt to the actual situation that CNC execution equipment in industrial sites does not have a fixed zero position. It reduces the high requirements of traditional hand-eye calibration methods on the hand-eye system, making hand-eye calibration more flexible and cost-effective. It provides strong support for the automation upgrade and transformation of industrial product production, testing and other processes, and has high industrial application value.

[0097] Table 1

[0098]

[0099] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A method for calibrating a hand-eye system without a fixed zero point, characterized in that, Includes the following steps: S0: Acquire the actuator, camera, target A and target B, wherein target A and target B are provided with feature points; S1: During the calibration phase, target A is fixed at the end of the actuator, the actuator is moved, the camera identifies target A and acquires images of the calibrated feature points on target A, and the rotation matrix between the camera coordinate system and the end coordinate system of the actuator is obtained. S2: During the equipment operation phase, target B is fixed at the end of the actuator. When the actuator starts, the camera identifies target B and collects feature point images on target B to obtain the translation vector between the camera coordinate system and the coordinate system at the end of the actuator. S3: Based on the rotation matrix and translation vector between the camera coordinate system and the actuator end coordinate system, complete the calibration of the hand-eye system without a fixed zero position.

2. The hand-eye system calibration method without a fixed zero position according to claim 1, characterized in that, The target A has a concentric circle pattern, the center of the concentric circle is a feature point, the inner circle of the target A is white and the outer circle is black; the target B is based on the target A and has multiple feature circles on the ring, the center of the feature circle is a feature point.

3. The hand-eye system calibration method without a fixed zero position according to claim 1, characterized in that, In step S1, the identification of target A includes the following steps: S101: Acquire an image of target A at the end of the actuator using a camera, and preprocess the image; S102: Identify image contours, set corresponding threshold conditions based on the number of points and area of ​​the image contours, filter the contours according to the threshold conditions, and fit the filtered contours to an ellipse. S103: Classify the fitted ellipse into inner and outer attributes, then divide the ellipse into pairs based on the distance between the center points of the inner and outer ellipses, and obtain the ellipse pairs whose center point distance meets the threshold condition. S104: For the obtained ellipse pair, the target A is identified by positioning it according to the true projection of the center point of the concentric circle.

4. The hand-eye system calibration method without a fixed zero position according to claim 3, characterized in that, In step S102, the equation for the fitted ellipse of the contour is: In the formula, A, B, C, D, E, and F are the ellipse parameters to be solved, and the calculation expressions for each ellipse parameter are as follows: In the formula: Let be the length of the rectangle circumscribed by the ellipse. The width of the rectangle circumscribed in the ellipse. The angle of inclination. The coordinates of the center point are given.

5. The hand-eye system calibration method without a fixed zero position according to claim 3, characterized in that, The identification of target B is performed by detecting the center of the concentric circles according to the identification steps of target A, and obtaining the projected image of target B; the projected image is subjected to affine transformation to standardize its shape and obtain a concentric circle image of a preset size; the feature circles within the annulus are detected and fitted according to the radius of the concentric circles, and then the image of target B is converted into a projected image to obtain the projected coordinates of the center point of the feature circle, thus completing the identification of target B.

6. The hand-eye system calibration method without a fixed zero position according to claim 1, characterized in that, In step S1, the specific steps for obtaining the rotation matrix between the camera coordinate system and the actuator end effector coordinate system include: S111: During the calibration stage, target A is fixed at the end of the actuator, and the actuator is driven to move along the direction of the laser plane so that the center point of target A coincides with the center line of the light stripe on its surface. S112: Switch the laser to acquire image pairs of the relatively stationary target A before and after the light stripe projection; S113: Identify the target A image before the light stripe projection, and obtain the three-dimensional coordinates of the center point of the concentric circle in the coordinate system of the actuator end; and obtain the three-dimensional coordinates of the center point of the concentric circle in the camera coordinate system based on the target A image after the light stripe projection. S114: The actuator end passes through each laser plane in sequence and acquires the corresponding images. The three-dimensional coordinate set of the target A center point in the actuator end coordinate system and the camera coordinate system is obtained. Based on the two three-dimensional coordinate sets, the rotation matrix of the actuator end coordinate system and the camera coordinate system is obtained through the SVD decomposition algorithm.

7. The hand-eye system calibration method without a fixed zero position according to claim 6, characterized in that, In step S114, the objective function for obtaining the rotation matrix of the actuator end-effector coordinate system and the camera coordinate system is: In the formula, This is the rotation matrix from the camera coordinate system to the actuator end effector coordinate system; This is the translation vector between the camera coordinate system and the zero position of the actuator end coordinate system at the current startup of the system; This represents the weight of each corresponding point in the two point sets; the value is positive and is set to 1. Let A be the three-dimensional coordinates of the center point of target A in the camera coordinate system. Let be the three-dimensional coordinates of the center point of target A in the coordinate system of the actuator end effector.

8. The hand-eye system calibration method without a fixed zero position according to claim 1, characterized in that, In step S2, the specific steps for obtaining the translation vector between the camera coordinate system and the actuator end effector coordinate system include: S201: During the system operation phase, target B is fixed at the end of the actuator. Each time the actuator is started, an image of target B is captured by a camera, and a world coordinate system is established on target B. S202: Obtain the sub-pixel coordinates of the center points of each feature circle within the concentric ring on target B, as well as the three-dimensional coordinates of the center points of each feature circle in the world coordinate system and the camera coordinate system. S203: Based on the sub-pixel coordinates of the center points of multiple sets of the same feature circles and their 3D coordinates in the world coordinate system and camera coordinate system, the rotation matrix and translation vector between the camera coordinate system and the world coordinate system are obtained through the PnP pose calculation algorithm. S204: Using the three-dimensional coordinates of the center point of the concentric circle of target B in the coordinate system of the actuator end effector as the origin coordinates of the world coordinate system, the current translation vector between the camera coordinate system and the world coordinate system is calculated based on the rotation matrix and translation vector between the camera coordinate system and the world coordinate system.

9. A hand-eye system calibration method without a fixed zero position according to claim 8, characterized in that, In step S203, the objective function for obtaining the rotation matrix and translation vector between the camera coordinate system and the world coordinate system is: In the formula, As a scaling factor, This is the camera's intrinsic parameter matrix. , All are known quantities; This is the rotation matrix between the camera coordinate system and the world coordinate system; This is the translation vector between the camera coordinate system and the world coordinate system; The sub-pixel coordinates of the center point of the feature circle. The three-dimensional coordinates of the center point of the feature circle in the camera coordinate system. The coordinates of the center point of the characteristic circle in the world coordinate system are given.

10. A hand-eye system calibration method without a fixed zero position according to claim 8, characterized in that, In S201, the origin of the world coordinate system is the center point of the concentric circles on target B, and the concentric circles are the world coordinate system Z. W The plane is equal to 0, and the number of characteristic circles on the ring is at least 4.

Images