Positioning method and positioning system

Abstract

To provide a positioning method, etc. that can suppress a decrease in the positioning accuracy of two components even if the verticality of the head and / or camera changes due to heat.SOLUTION: A positioning method is for positioning a first component P1 held by a head 11 and a second component P2 placed on a stage 12 by imaging them with a camera (vertical field-of-view recognition camera 13). Input data containing recognition results of at least one of the first component P1 and the second component P2 captured by the camera is input, a learning model is used to output a relative positional deviation amount between the first component P1 and the second component P2, and on the basis of the relative positional deviation amount output by the learning model, the position of at least one of the head 11 and the stage 12 is controlled.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to an alignment method and system for aligning two parts. [Background technology]

[0002] A conventional alignment method for aligning two components, such as electronic components, involves using a camera to recognize a first component held by a head and a second component placed on a stage, calculating the amount of misalignment between the first and second components through image processing, and then correcting the misalignment by moving the head or the stage based on the calculation results, thereby aligning the first and second components.

[0003] This type of alignment method is disclosed in Patent Document 1. The alignment method disclosed in Patent Document 1 uses a simultaneous top-bottom observation camera that can simultaneously observe the back surface of the first component, which is the bonding surface between the first component and the second component, and the front surface of the second component. Specifically, the simultaneous top-bottom observation camera is inserted between the first component and the second component, simultaneously capturing images of the positioning mark on the back surface of the first component and the positioning mark on the front surface of the second component, calculating the amount of horizontal misalignment between the first component and the second component, and correcting the misalignment to achieve alignment. By aligning the first component and the second component in this manner in the horizontal direction, the first component can be placed on the second component simply by lowering the head after correcting the misalignment. This minimizes movement errors in the head and stage, enabling highly accurate alignment. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent No. 3569820 Summary of the Invention [Problem to be solved by the invention]

[0005] However, conventional alignment methods do not take into account the perpendicularity of the head from the alignment position to the component placement position, and are therefore unable to correct alignment errors caused by the head descending. In particular, when the head and its components thermally expand due to heat generated by the camera and the actuator of the drive shaft, as well as friction between the drive components, a deviation occurs between the horizontal position of the component during alignment and the horizontal position of the component when the head descends, resulting in reduced alignment accuracy. Similarly, when the perpendicularity of the camera's optical axis (perpendicularity to the component or board) changes due to the effects of heat, a deviation occurs in the component recognition position of the camera, resulting in reduced alignment accuracy.

[0006] As described above, the conventional alignment method has a problem in that the perpendicularity of the head and / or camera changes due to heat, resulting in a decrease in alignment accuracy.

[0007] The present disclosure has been made to solve such problems, and aims to provide an alignment method and alignment system that can prevent a decrease in the alignment accuracy of two parts even if the perpendicularity of the head and / or camera changes due to heat. [Means for solving the problem]

[0008] In order to achieve the above-mentioned object, one aspect of the alignment method disclosed herein is an alignment method that aligns a first part held by a head and a second part placed on a stage by capturing images of them with a camera, and uses a learning model that outputs the amount of relative positional deviation between the first part and the second part when input data including recognition results of at least one of the first part and the second part captured by the camera is input, and the position of at least one of the head and the stage is controlled based on the amount of relative positional deviation output by the learning model.

[0009] Furthermore, one aspect of the alignment system according to the present disclosure is an alignment system that aligns a first part and a second part, and includes: a head that holds the first part; a stage on which the second part is placed; a camera that images the first part and the second part; and a control device that uses a learning model that outputs a relative positional deviation amount between the first part and the second part in response to input data including a recognition result of at least one of the first part and the second part imaged by the camera, and controls the position of at least one of the head and the stage based on the relative positional deviation amount output by the learning model. [Effects of the Invention]

[0010] According to the present disclosure, even if the perpendicularity of the head and / or camera changes due to heat, it is possible to prevent the alignment accuracy of the two components from decreasing. [Brief explanation of the drawings]

[0011] [Figure 1] FIG. 1 is a diagram schematically illustrating the configuration of an alignment system according to the first embodiment. [Figure 2] FIG. 2 is a functional block diagram of the alignment system according to the first embodiment. [Figure 3] FIG. 3 is a diagram showing an operation flow of the alignment method according to the first embodiment. [Figure 4] FIG. 4 is a diagram showing a flow of generating a learning model in the registration system according to the first embodiment. [Figure 5] FIG. 5 is a diagram showing an example of operation parameters acquired by an operation parameter acquisition unit in the control device of the alignment system according to the first embodiment. [Figure 6] FIG. 6 is a diagram for explaining a method for detecting detection data of the alignment mark of the first component. [Figure 7] FIG. 7 is a diagram for explaining the relationship between the posture and edge strength of the alignment mark of the first component in an ideal state where the positional relationship of the alignment device has not changed. [Figure 8] FIG. 8 is a diagram for explaining the relationship between the posture of the alignment mark of the first component and the strength of the edge when the positional relationship of the alignment device has changed due to thermal strain or the like. [Figure 9] FIG. 9 is a diagram schematically illustrating the configuration of an alignment system according to the second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0012] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. Note that each of the embodiments described below represents a specific example of the present disclosure. Therefore, the numerical values, shapes, materials, components, the arrangement and connection of the components, steps (processes), and the order of steps shown in the following embodiments are merely examples and are not intended to limit the present disclosure. Therefore, among the components in the following embodiments, components that are not described in the independent claims that represent the superordinate concept of the present disclosure will be described as optional components.

[0013] In addition, in this specification and drawings, the X-axis, Y-axis, and Z-axis represent the three axes of a three-dimensional Cartesian coordinate system. The X-axis and Y-axis are mutually orthogonal and are both orthogonal to the Z-axis. The direction of rotation around the Z-axis is designated as θ. In this embodiment, the Z-axis direction is the vertical direction. Note that each drawing is a schematic diagram and is not necessarily an exact illustration. Therefore, the scales and the like do not necessarily match in each drawing. In each drawing, substantially identical components are designated by the same reference numerals, and redundant explanations are omitted or simplified.

[0014] Furthermore, in this specification, the terms "above," "up," "below," and "below" do not refer to the upward direction (vertically upward) and downward direction (vertically downward) in absolute spatial recognition, but are used as terms defined by a relative positional relationship. Furthermore, the terms "above," "up," "below," and "below" are used not only when two components are arranged with a gap between them and another component exists between them, but also when two components are arranged in contact with each other.

[0015] (Embodiment 1) [Alignment system / alignment device] First, an alignment system 1 and an alignment device 10 according to the first embodiment will be described with reference to Fig. 1 and Fig. 2. Fig. 1 is a diagram schematically showing the configuration of the alignment system 1 according to the first embodiment. Fig. 2 is a functional block diagram of the alignment system 1 according to the first embodiment.

[0016] As shown in FIG. 1, the alignment system 1 includes an alignment device 10 for aligning (positioning) two parts, and a control device 20 for controlling the alignment device 10.

[0017] The alignment device 10 is a processing device that aligns a first component P1 (first member) and a second component P2 (second member). As an example, the alignment device 10 is a component mounter. In this case, for example, when mounting an electronic component, which is the first component P1, on a board, which is the second component, the alignment device 10 aligns the first component P1 and the second component P2 and places the first component P1 on the second component P2.

[0018] As shown in FIG. 1, the alignment device 10 includes a head 11 that holds a first component P1, a stage 12 on which a second component P2 is placed, and an upper and lower view recognition camera 13 that recognizes the first component P1 and the second component P2.

[0019] Alignment marks M1 and M2 are provided on the first part P1. The alignment marks M1 and M2 are printed, for example, on the back surface of the first part P1. The alignment marks M3 and M4 are provided on the second part P2. The alignment marks M3 and M4 are printed, for example, on the front surface of the second part P2. The alignment marks M1, M2, M3, and M4 are, for example, one or more figures. As an example, the shapes of the figures constituting the alignment marks M1, M2, M3, and M4 are a circle, a rectangle, a cross, etc., but are not limited to these.

[0020] The head 11 and the stage 12 are configured to be relatively movable. Therefore, either the head 11 or the stage 12 may be configured to be movable, or both the head 11 and the stage 12 may be configured to be movable.

[0021] In this embodiment, both the head 11 and the stage 12 are configured to be movable. Specifically, the head 11 can move in the Z-axis direction, and the stage 12 can move in the horizontal direction (X-axis direction, Y-axis direction) and rotate around the Z-axis (θ rotation). Therefore, although not shown, the alignment device 10 has a drive unit that can move the head 11 in the Z-axis direction, and a drive unit that can move the stage 12 in the X-axis direction and Y-axis direction and rotate it around the Z-axis.

[0022] The head 11 may be configured to be able to rotate around the Z axis as well, and may be configured to be able to move in the horizontal direction (X-axis direction, Y-axis direction). The stage 12 may be configured to be able to move in the Z axis direction as well.

[0023] The upper and lower view recognition camera 13 is an example of a camera that captures images of the first component P1 and the second component P2. In this embodiment, the upper and lower view recognition camera 13 includes a camera body 13a having an imaging element, an optical system 13b including a lens, a lighting device 13c, and a computing device (not shown) that performs various computational processes. The camera body 13a has an imaging element. The lens of the optical system 13b focuses light onto the imaging element of the camera body 13a. The lighting device 13c has a light source and emits white light as illumination light. The illumination light of the lighting device 13c is not limited to white light, and may be monochromatic light such as red light or blue light, or invisible light such as infrared light (IR light).

[0024] The upper and lower view recognition camera 13 can recognize an upper view and a lower view by capturing images with the camera body 13a. Possible structures for recognizing the upper view and the lower view include a structure using two cameras, a first camera that recognizes the upper view and a second camera that recognizes the lower view, as the camera body 13a, or a structure using two mirrors, one that captures the upper view and one that captures the lower view, as one of the optical systems 13b, and using one camera as the camera body 13a.

[0025] The top and bottom view recognition camera 13 recognizes the first part P1 by capturing an image of the first part P1 located above. Specifically, the top and bottom view recognition camera 13 recognizes the first part P1 by capturing an image of and recognizing an alignment mark M1 or M2 on the first part P1. The top and bottom view recognition camera 13 also recognizes the second part P2 by capturing an image of the second part P2 located below. Specifically, the top and bottom view recognition camera 13 recognizes the second part P2 by capturing an image of and recognizing an alignment mark M3 or M4 on the second part P2.

[0026] When capturing images of the first component P1 and the second component P2, the lighting device 13c may be turned on to create a light-dark inversion state to capture images of the first component P1 and the second component P2. The alignment device 10 may also have a mechanism for adjusting the focal positions of the camera body 13a and the lens of the optical system 13b. For example, a mechanism for adjusting the position of the camera body 13a or the lens may be considered.

[0027] The upper and lower view recognition camera 13 configured in this manner is configured to be movable in the X-axis direction and the Y-axis direction. Therefore, although not shown, the alignment device 10 has a drive unit that can move the upper and lower view recognition camera 13 in the X-axis direction and the Y-axis direction.

[0028] When aligning the first component P1 and the second component P2, the positions of the head 11 and the stage 12 are moved so that the first component P1 and the second component P2 are roughly aligned in the horizontal direction, and the top and bottom view recognition camera 13 is inserted between the first component P1 held by the head 11 and the second component P2 placed on the stage 12 (specifically, at the intermediate position). The top and bottom view recognition camera 13 then captures images of the back surface of the first component P1 and the front surface of the second component P2, thereby recognizing the first component P1 and the second component P2. Specifically, the top and bottom view recognition camera 13 captures an image of the first component P1, thereby recognizing the alignment mark M1 or M2 printed on the first component P1. The top and bottom view recognition camera 13 also captures an image of the second component P2, thereby recognizing the alignment mark M3 or M4 printed on the second component P2. The recognition of the first component P1 and the second component P2 by the upper and lower view recognition cameras 13 may be performed simultaneously or at different times. After the first component P1 and the second component P2 are recognized by the upper and lower view recognition cameras 13, the first component P1 and the second component P2 are aligned based on the recognition results. After the alignment is complete, the upper and lower view recognition cameras 13 are retracted to a position where they will not interfere with the lowering of the head 11.

[0029] The control device 20 is configured by a computer system or the like, and performs various controls on the alignment device 10. As shown in Fig. 2, the control device 20 controls the head 11, stage 12, and upper and lower view recognition camera 13 of the alignment device 10 as control objects.

[0030] In this embodiment, the control device 20 includes an operation parameter acquisition unit 21, a data processing unit 22, a learning model generation unit 23, a positional deviation amount estimation unit 24, an operation command unit 25, an operation control unit 26, and a memory unit 27.

[0031] The operation parameter acquisition unit 21 acquires operation parameters during the alignment operation that include at least the recognition results of the first part P1 and the second part P2 by the upper and lower view recognition camera 13. The recognition results of the first part P1 included in the operation parameters are, for example, detection data of the alignment marks M1 and M2 on the first part P1. Furthermore, the recognition results of the second part P2 are, for example, detection data of the alignment marks M3 and M4 on the second part P2.

[0032] The data processing unit 22 formats and processes the data of the operation parameters during the positioning operation acquired by the operation parameter acquisition unit 21.

[0033] The learning model generation unit 23 receives as input data operation parameters during the alignment operation that include at least the recognition results of the first part P1 and the second part P2 by the upper and lower view recognition camera 13, and generates a learning model that has been machine-learned to output the amount of relative positional deviation between the first part P1 and the second part P2.

[0034] The positional deviation estimation unit 24 inputs operation parameters during the alignment operation, which include at least the recognition results of the first part P1 and the second part P2 by the upper and lower view recognition camera 13, into the learning model generated by the learning model generation unit 23, and estimates the relative positional deviation amount between the first part P1 and the second part P2 from the relative positional deviation amount between the first part P1 and the second part P2 output from the learning model.

[0035] The operation command unit 25 sends an operation command for the operation of aligning the first part P1 and the second part P2 (correction operation) to the alignment device 10 based on the amount of misalignment (estimated value) estimated by the misalignment amount estimation unit 24. Specifically, the control device 20 is connected to the head 11 and the stage 12, and sends an operation command to the head 11 to move the head 11, and sends an operation command to the stage 12 to move the stage 12. The control device 20 is also connected to the upper and lower view recognition camera 13, and sends an operation command to the upper and lower view recognition camera 13 to move the upper and lower view recognition camera 13.

[0036] The operation control unit 26 executes the alignment operation by controlling the operation of each component of the alignment device 10. Specifically, the operation control unit 26 controls the operation of the head 11, the operation of the stage 12, and the operation of the upper and lower view recognition camera 13.

[0037] The memory unit 27 stores the data processed by the data processing unit 22, the learning model generated by the learning model generation unit 23, and the positional deviation amount estimated by the positional deviation amount estimation unit 24. The control device 20 may send operation commands to the alignment device 10 and control the operation of the alignment device 10 in accordance with the production program stored in the memory unit 27.

[0038] [Alignment method (alignment operation)] Next, the alignment method in the alignment system 1 will be described using Fig. 3 while referring to Fig. 1 and Fig. 2. Fig. 3 is a diagram showing an operation flow of the alignment method according to the first embodiment.

[0039] 3, in the alignment method according to this embodiment, first, images of the first component P1 and the second component P2 are captured by the upper and lower view recognition camera 13 (step SA1). Specifically, as shown in FIG. 1, the upper and lower view recognition camera 13 is moved between the first component P1 held by the head 11 and the second component P2 placed on the stage 12, and the upper and lower view recognition camera 13 captures images of the back surface of the first component P1 and the front surface of the second component P2. In this embodiment, the upper and lower view recognition camera 13 is moved to an imaging position where the alignment mark M1 printed on the first part P1 and the alignment mark M3 printed on the second part P2 can be recognized, and images of the alignment mark M1 and the alignment mark M3 are taken by the upper and lower view recognition camera 13. Thereafter, the upper and lower view recognition camera 13 is moved to an imaging position where the alignment mark M2 printed on the first part P1 and the alignment mark M4 printed on the second part P2 can be recognized, and images of the alignment mark M2 and the alignment mark M4 are taken by the upper and lower view recognition camera 13. There are no particular limitations on the number of times and order in which the first part P1 and the second part P2 are imaged.

[0040] Next, as shown in FIG. 3, the component positions of the first component P1 and the second component P2 are detected (step SA2). Specifically, the component positions of the first component P1 and the second component P2 can be detected by calculating the images captured by the upper and lower view recognition camera 13 (camera body 13a) in step SA1 using a calculation device of the upper and lower view recognition camera 13. For example, to determine the component position of the first component P1, alignment marks M1 and M2 are detected from the images captured by the upper and lower view recognition camera 13 in step SA1, and the component position of the first component P1 is calculated from the representative coordinates of the alignment marks M1 and M2. At this time, the angle of the first component P1 is also calculated. Similarly, to determine the component position of the second component P2, alignment marks M3 and M4 are detected from the images captured by the upper and lower view recognition camera 13 in step SA1, and the component position and angle of the second component P2 are calculated from the representative coordinates of the alignment marks M3 and M4.

[0041] In this embodiment, the positions of the first component P1 and the second component P2 are detected using a recognition image of an alignment mark provided for component position detection, but this is not limiting. For example, the positions of the first component P1 and the second component P2 may be detected using a recognition image in which a portion of the first component P1 and / or a portion of the second component P2 is captured. As the portions of the first component P1 and the second component P2, portions of the outer peripheries of the first component P1 and the second component P2 (e.g., corners of the components) or electrodes of the first component P1 and the second component P2 may be used.

[0042] Next, the amount of misalignment between the first part P1 and the second part P2 is estimated by the control device 20 (step SA3). In this embodiment, the amount of misalignment between the first part P1 and the second part P2 is estimated using a learning model.

[0043] Specifically, the operation parameter acquisition unit 21 acquires operation parameters during the alignment operation including the recognition results (detection data of the alignment marks) of at least one of the first part P1 and the second part P2, the data processing unit 22 generates input data for the learning model by shaping and / or processing the acquired operation parameters, and the positional deviation amount estimation unit 24 inputs the input data generated by the data processing unit 22 to the learning model stored in the storage unit 27 and estimates the positional deviation amount between the first part P1 and the second part P2 based on the relative positional deviation amount between the first part P1 and the second part P2 output from the learning model. Note that the operation parameters acquired by the operation parameter acquisition unit 21 are the same as the input data acquired when generating the learning model, which will be described later.

[0044] Next, in step SA3, it is determined whether or not a positioning operation (correction operation) for aligning the first part P1 and the second part P2 is necessary, depending on the positional deviation amount between the first part P1 and the second part P2 estimated by the learning model (step SA4). This determination is made, for example, by the operation command unit 25.

[0045] As a method for determining whether or not an alignment operation is necessary, a method may be used in which the positional deviation tolerance amount set in the production parameters is used as a threshold value, and if the estimated positional deviation amount is equal to or greater than the threshold value, it is determined that an alignment operation is necessary, and if it is equal to or less than the threshold value, it is determined that an alignment operation is not necessary, or a method may be used in which an alignment operation is performed according to the number of corrections set in the production parameters, or a method in which a correction operation is not necessary if either of the above conditions is met.

[0046] Next, if it is determined in step SA4 that an alignment operation is necessary (YES in step SA4), the process proceeds to the alignment operation and the alignment operation is executed (step SA5). Specifically, the position of at least one of the head 11 and the stage 12 is controlled based on the relative positional deviation amount output by the learning model.

[0047] In this embodiment, the operation command unit 25 sends an operation command to align the first part P1 and the second part P2 to the stage 12 according to the amount of positional deviation estimated by the learning model in step SA3, and the operation control unit 26 drives the stage 12 in the X-axis direction, Y-axis direction, and θ direction, thereby aligning the first part P1 and the second part P2.

[0048] In this embodiment, the first component P1 and the second component P2 are aligned by driving only the stage 12, but this is not limiting. Specifically, as long as the first component P1 and the second component P2 can be aligned relative to each other, the first component P1 and the second component P2 may be aligned by driving only the head 11, or the first component P1 and the second component P2 may be aligned by driving both the head 11 and the stage 12. After the alignment operation, the process returns to step SA1.

[0049] On the other hand, if it is determined in step SA4 that alignment is not required (NO in step SA4), the first part P1 and the second part P2 are aligned (step SA6). In this embodiment, the first part P1 is aligned with the second part P2. Specifically, the head 11 is lowered in the Z-axis direction to place the first part P1 on the second part P2.

[0050] After placing the first component P1 on the second component P2, the head 11 may release its hold on the first component P1 to separate the head 11 from the first component P1, or the head 11 may press the first component P1 against the second component P2 while still holding the first component P1 to transfer the shape of the first component P1.

[0051] [Generating learning models] Next, a method for generating a learning model by the learning model generation unit 23 in the control device 20 will be described with reference to Fig. 4. Fig. 4 is a diagram showing a flow for generating a learning model in the alignment system 1 according to the first embodiment.

[0052] As shown in Fig. 4, first, the operation parameter acquisition unit 21 acquires the operation parameters of the alignment device 10 (step SL1). An example of the operation parameters acquired by the operation parameter acquisition unit 21 is shown in Fig. 5. As shown in Fig. 5, the operation parameters acquired by the operation parameter acquisition unit 21 include input data and output data. The input data is a group of parameters for estimating the amount of misalignment between the first part P1 and the second part P2, and the output data is a group of parameters indicating the amount of misalignment between the first part P1 and the second part P2.

[0053] The operation parameter acquisition unit 21 acquires, as input data, operation parameters for estimating the amount of misalignment between the first part P1 and the second part P2 during the alignment operation from the first part P1, the second part P2, the head 11, the stage 12, the upper and lower view recognition camera 13, and the alignment device 10. Specifically, as shown in Fig. 5, the operation parameters include part position detection data for the first part P1 and the second part P2, device operation data for the alignment device 10, and production parameters.

[0054] The component position detection data includes the component positions of the first component P1 and the second component P2, and parameters related to component position calculation, such as detection data for the alignment marks M1 and M2 of the first component P1 and detection data for the alignment marks M3 and M4 of the second component P2.

[0055] The detection data of the alignment marks M1 and M2 of the first part P1 is one of the recognition results of the first part P1. Similarly, the detection data of the alignment marks M3 and M4 of the second part P2 is one of the recognition results of the second part P2. That is, the alignment marks M1 and M2 of the first part P1 are characteristic shapes used to calculate the recognition result of the first part P1. Similarly, the alignment marks M3 and M4 of the second part P2 are characteristic shapes used to calculate the recognition result of the second part P2.

[0056] Here, a method for acquiring detection data for the alignment marks M1 and M2 on the first part P1 and detection data for the alignment marks M3 and M4 on the second part P2 will be described. Note that, although a method for acquiring detection data for the alignment mark M1 on the first part P1 will be described below using Fig. 6 as an example, the detection data for the alignment mark M2 on the first part P1 and the detection data for the alignment marks M3 and M4 on the second part P2 can also be acquired using a similar method.

[0057] 6 is a diagram for explaining a method for detecting detection data of the alignment mark M1 of the first component P1. As shown in FIG. 6, in this embodiment, the alignment mark M1 has a circular shape.

[0058] The detection data of the alignment mark M1 of the first component P1 can be obtained by edge detection. In this case, first, an image including at least a portion of the alignment mark M1 is captured by the upper and lower view recognition camera 13, and the edge of the alignment mark M1 is detected by performing edge detection calculations on the captured image by the calculation device of the upper and lower view recognition camera 13. Specifically, an edge detection point cloud 30 consisting of multiple edge detection points along the circle of the alignment mark M1 is obtained.

[0059] Then, by fitting these edge detection point groups 30 to a circle equation, the center coordinates of the edge detection point group 30 are calculated. These center coordinates become the recognition position Pc of the first part P1. In this embodiment, the recognition position Pc is the representative coordinate of the alignment mark M1. In other words, the recognition position Pc becomes one of the representative coordinates of the first part P1.

[0060] The edge detection point of the alignment mark M1 can be calculated from the edge strength of the alignment mark M1. Edge strength refers to the brightness gradient value between a pixel in an image and its surrounding pixels (the difference value between two adjacent pixels). The edge detection point refers to the XY coordinates of a pixel in the image where the edge strength reaches a maximum and exceeds a threshold value, resulting in the pixel being determined to be an edge. Edge intensity refers to the edge strength at the edge detection point.

[0061] For example, when calculating an edge detection point from the brightness gradient of an image of the alignment mark M1, the XY coordinates of the pixel where the brightness gradient is at a maximum value can be determined as the edge detection point of the alignment mark M1. For example, the difference in brightness between two adjacent pixels can be calculated, and the difference value is detected as the edge strength. The pixel with the largest difference value can be detected as the edge detection point, thereby obtaining the edge detection point group 30. For example, in FIG. 6, points Pl and Pr are the points with the minimum and maximum X coordinates in the edge detection point group 30. In this case, when the focus of the upper and lower view recognition camera 13 is aligned with Pl or Pr, the edge strength is maximized. If the focus of the upper and lower view recognition camera 13 is shifted from Pl or Pr, the boundary of the alignment mark M1 becomes blurred and the edge strength weakens. Therefore, the positional deviation in the Z-axis direction between the upper and lower view recognition camera 13 and Pl or Pr can be estimated from the edge strength.

[0062] In this way, the recognition position Pc of the alignment mark M1 can be acquired as one of the detection data of the alignment mark M1. Furthermore, the detection data of the alignment mark M1 may include position coordinates (XY coordinates) of multiple edge detection points in the edge detection point group 30 in addition to the recognition position Pc (representative coordinates) of the alignment mark M1 of the first component P1.

[0063] The recognition position Pc (representative coordinates) of the alignment mark M1 does not have to be the center coordinates of a circle. The recognition position Pc of the first component P1 may be any position coordinates related to a mark or reference position detected from an image. For example, if the alignment mark M1 is a cross-shaped mark, the center point of the cross mark may be the recognition position Pc of the first component P1. Alternatively, the recognition position Pc may be the intersection of multiple lines detected from multiple feature points (such as electrodes) of the first component P1. Alternatively, the recognition position Pc may be a part of the outline of the first component P1 (e.g., a corner of the component). In the case of a cross mark, edge detection points may be calculated on each side of the cross mark, the obtained edge detection points may be fitted to a line, and the recognition position Pc may be calculated from the intersection and midpoint of the line. The edge detection method for the alignment mark M1 is not limited to the above method. Any method may be used as long as it is possible to calculate the edge detection points and the edge strength at the edge detection points from the edge strength of the alignment mark M1. For example, edge detection points may be detected by applying any filter processing to the captured image, or the brightness of each pixel may be fitted to any function to determine the feature points of the function as edge detection points, or edge detection points may be calculated by pattern matching and the pattern match rate may be determined as edge strength, or edge detection points may be calculated by differentiating the obtained brightness graph to determine the gradient, or optimal edge detection points may be calculated using a learning model based on brightness information in the captured image.

[0064] Returning to FIG. 5 , the device operation data of the alignment device 10 includes operating time data of the alignment device 10. The operating time data may include the elapsed time since the alignment device 10 was powered on, the elapsed time since the start of the alignment operation, the number of cycles of the alignment operation, the time required for one cycle of the alignment operation, the time required for each step of the alignment operation, the elapsed time from the time when a predetermined condition in the alignment operation was satisfied, and the timing at which each step of the alignment operation was executed. The input data to the learning model may include at least one of these. Note that the head 11, stage 12, and upper and lower view recognition camera 13 of the alignment device 10 may deform over time. For example, the measured position of the first component P1 held by the head 11 may differ between the case where the position of the first component P1 held by the head 11 is measured immediately after the alignment device 10 is powered on and the case where the position of the first component P1 held by the head 11 is measured a certain time after the power is turned on, even if the position of the first component P1 held by the head 11 has not changed. This is because even temperature changes at levels that are difficult to measure with a temperature sensor cause each component of the alignment device 10 to thermally expand or contract, and in an alignment device 10 made up of many components, this slight thermal expansion or contraction accumulates, causing changes over time in the head 11, stage 12, and upper and lower field of view recognition cameras 13 of the alignment device 10. Therefore, by including operation time data of the alignment device 10 in the detection data for the operating status of the alignment device 10, it is possible to estimate an appropriate amount of misalignment under each condition even if changes over time occur in the head 11, stage 12, and upper and lower field of view recognition cameras 13.

[0065] The method for measuring the amount of misalignment is not particularly limited as long as it can measure the amount of misalignment caused by changes over time due to temperature changes inside the alignment device 10. For example, it may be a method of estimating the amount of misalignment from time, a method of measuring the amount of misalignment at certain points on the head 11, stage 12, and upper and lower field of view recognition cameras 13 with a laser displacement meter and estimating the amount of misalignment of the component from changes over time in the measurement results, or a method of measuring strain at certain points on the head 11, stage 12, and upper and lower field of view recognition cameras 13 with a strain gauge and measuring the amount of misalignment from changes over time in the measurement results.

[0066] The device operation data of the alignment device 10 may also include data on temperature changes over time detected at at least one detection point of the alignment device 10. The data on temperature changes over time may include temperature data from a certain time ago at the temperature detection point, difference data between the temperature data from a certain time ago and the current temperature data, and an integrated value of the time and temperature change since the temperature change at a certain detection point. For example, if there is a temperature change in a heat-generating component such as the upper and lower field of view recognition camera 13, the head 11, or an actuator attached to the stage 12, the temperature inside the alignment device 10 may increase over time. Therefore, even if there is no change in the temperature at the detection point, the measured component position of the first component P1 held by the head 11 may differ between a case where the component position of the first component P1 held by the head 11 is measured immediately after the temperature change at the detection point and a case where the component position of the first component P1 held by the head 11 is measured a certain time after the temperature change. This is due to thermal expansion or contraction of each component due to temperature changes in locations where no temperature sensor is attached or in each component constituting the alignment device 10. Therefore, since the detection data on the operating status of the alignment device 10 includes data on temperature changes over time, even if there is a temperature change inside the alignment device 10, it is possible to estimate an appropriate amount of misalignment under each condition.

[0067] The production parameters also include the axis positions, speeds, and acceleration / decelerations for each step of the alignment operation of the drive unit of the alignment device 10 (head 11, stage 12, and upper and lower view recognition camera 13); movement conditions for each step of the alignment operation, such as the axis positions, axis movement speeds, accelerations, axis movement distances, and axis movement directions of the head 11, stage 12, and upper and lower view recognition camera 13; imaging conditions, such as the shutter speed, gain, and output (illumination output) of the camera body 13a set when the upper and lower view recognition camera 13 captures images of the first component P1 and the second component P2; and workpiece conditions, such as the shape, material, refractive index, and alignment mark shape of the workpieces (first component P1 and second component P2) targeted by the alignment device 10. These production parameters may change depending on the product type or production system. By including the production parameters of the alignment device 10 in the input data of the operation parameter acquisition unit 21, even if the production parameters change depending on the product type or production system, the amount of misalignment can be estimated appropriately under each condition.

[0068] 5 are parameters related to alignment accuracy and indicate the amount of misalignment between the first component P1 and the second component P2 after the first component P1 is placed on the second component P2 in step SA6 (component placement operation) of FIG. 3. In this embodiment, the amount of misalignment between the first component P1 and the second component P2 is measured by placing the first component P1 on the second component P2, capturing and recognizing images of the first component P1 and the second component P2 using upper and lower view recognition cameras 13, and calculating the amount of misalignment between the recognized positions of the first component P1 and the second component P2 using a calculation device of the upper and lower view recognition cameras 13. The images of the first component P1 and the second component P2 captured by the upper and lower view recognition cameras 13 include part or all of the first component P1 and the second component P2, and the amount of misalignment between the recognized positions is calculated from alignment marks or the contours (external shapes) of the components.

[0069] In this case, it is not necessary to measure the amount of misalignment while the first component P1 is placed on the second component P2. The first component P1 may be pressed against the second component P2 to transfer the shape of the first component P1, and the amount of misalignment between the transferred shape and the target position may be calculated to determine the amount of misalignment of the recognized position. Furthermore, the amount of misalignment does not have to be measured by the alignment device 10; it may be measured by a separate measuring device after the alignment operation. In this case, the measured amount of misalignment may be stored in the memory unit 27 and processed by the data processing unit 22.

[0070] Returning to FIG. 4, after step SL1, a training dataset for the training model is generated (step SL2). Specifically, the data processing unit 22 shapes and processes the data group of the operational parameters of the alignment device 10 acquired by the operational parameter acquisition unit 21, and stores the input data and output data in the storage unit 27 as a single training dataset. Examples of processing the data group include calculating representative values ​​such as maximum, minimum, median, and average values ​​from multiple pieces of detection data, and calculating statistical values ​​such as the difference value, variance, and standard deviation between two points of data. By performing such preprocessing, it is possible to maintain or improve the accuracy of the training model while reducing the number of data points in the training dataset.

[0071] Next, as shown in Fig. 4, a learning model is generated using the learning dataset generated in step SL2 (step SL3). Specifically, the learning model generation unit 23 reads the learning dataset stored in the storage unit 27 from the storage unit 27, and generates a learning model using a learning algorithm such as machine learning, in which the parameters of the input data are used as input and the parameters of the output data are used as output. As the learning algorithm, a neural network (including deep learning using a multi-layer neural network), genetic programming, a decision tree, a Bayesian network, a support vector machine (SVM), linear regression, or the like can be used.

[0072] Furthermore, the number of learning models generated does not need to be one for each alignment device 10. Furthermore, a learning model may be added or updated at any timing, taking into consideration the number of learning data sets of operation parameters acquired by the operation parameter acquisition unit 21 as training data, the estimation accuracy of the learning model, the time required to generate the learning model, and the like. For example, multiple learning models may be generated according to the types of the first part P1 and the second part P2. Furthermore, after the operation parameter acquisition unit 21 acquires new operation parameters, the learning model may be updated based on the added learning data set. By adding or updating a learning model in this way, the accuracy of the learning model can be maintained or improved.

[0073] Next, as shown in FIG. 4, the learning model generated in step SL3 is stored in the storage unit 27 (step SL4).

[0074] [Effects, etc.] As described above, in the alignment method and alignment system 1 according to this embodiment, a learning model is used that outputs the amount of relative positional deviation between the first part P1 and the second part P2 in response to input data including the recognition results of at least one of the first part P1 and the second part P2 captured by the upper and lower view recognition camera 13, and the position of at least one of the head 11 and the stage 12 is controlled based on the amount of relative positional deviation output by this learning model.

[0075] Specifically, the operation parameter acquisition unit 21 of the control device 20 acquires operation parameters during the alignment operation, including the recognition results of at least one of the first part P1 and the second part P2 (e.g., detection data of the alignment mark M1 of the first part P1). The data processing unit 22 then shapes and processes the operation parameters to generate input data for the learning model. The misalignment amount estimation unit 24 inputs the input data generated by the data processing unit 22 into the learning model stored in the memory unit 27 and estimates the misalignment amount between the first part P1 and the second part P2 based on the relative misalignment amount between the first part P1 and the second part P2 output from the learning model. This allows for accurate estimation of the misalignment amount between the first part P1 and the second part P2. The position of at least one of the head 11 holding the first part P1 and the stage 12 on which the second part P2 is placed is then controlled based on the estimated misalignment amount between the first part P1 and the second part P2. This allows for accurate correction of the misalignment between the first part P1 and the second part P2. Therefore, even if the perpendicularity of the head 11 and / or the upper and lower view recognition camera 13 changes due to heat, it is possible to prevent a decrease in the alignment accuracy between the first part P1 and the second part P2. In other words, it is possible to improve the alignment accuracy between the first part P1 and the second part P2.

[0076] The input data to the learning model may include a calculation result of the relative positional deviation between the first part P1 and the second part P2. The input data to the learning model may also include a calculation result of the edge strength of the characteristic shape (alignment mark) for calculating the recognition result of at least one of the first part P1 and the second part P2.

[0077] In addition, the input data to the learning model may include at least one of the external dimensions of the characteristic shape (alignment mark) and the recognition position of the characteristic shape for calculating the recognition result of at least one of the first part P1 and the second part P2, or may include the brightness of the characteristic shape (alignment mark) for calculating the recognition result of at least one of the first part P1 and the second part P2, or may include the temperature at one or more measurement points of the alignment device 10.

[0078] Here, the effect of including, for example, detection data of alignment mark M1 as a parameter related to component position calculation in the input data included in the operation parameters acquired by operation parameter acquisition unit 21, i.e., the input parameters of the generated learning model, will be described below.

[0079] When the perpendicularity of the optical axis of the upper and lower view recognition camera 13 changes due to the influence of heat, the distance between the upper and lower view recognition camera 13 and the alignment mark M1 changes. This may change the edge strength and recognition position of the alignment mark M1. FIGS. 7 and 8 are diagrams for explaining the relationship between the posture and edge strength of the alignment mark M1 of the first part P1. The detection data shown in FIG. 7 represents the detection data of the alignment mark M1 in a state where the first part P1 is positioned perpendicular to the optical axis of the upper and lower view recognition camera 13 (an ideal state). On the other hand, the detection data shown in FIG. 8 represents the detection data of the alignment mark M1 in a state where the perpendicularity of the first part P1 to the optical axis of the upper and lower view recognition camera 13 has changed due to thermal distortion of the head 11 holding the first part P1, the stage 12 on which the second part P2 is placed, or the upper and lower view recognition camera 13.

[0080] 7 and 8, among the edge detection point group 30 of the alignment mark M1 detected by the upper and lower view recognition camera 13, if the difference ΔE is the edge strength El at the detection point Pl with the smallest X coordinate and the edge strength Er at the detection point Pr with the largest X coordinate, ΔE in Fig. 8 may be larger than ΔE in Fig. 7. That is, in Fig. 7, the heights of the detection points Pl and Pr are approximately the same, but in Fig. 8, the heights of the detection points Pl and Pr do not match, so ΔE in Fig. 8 is larger than ΔE in Fig. 7. For example, if the position of the Pr side is deviated from the focal position of the upper and lower view recognition camera 13, the image will be out of focus and blurred, resulting in a decrease in the edge strength of Er, and therefore ΔE in Fig. 8 will be larger than ΔE in Fig. 7.

[0081] Therefore, by using the edge strength of the alignment mark M1 as one of the input data for the learning model, even if the positional relationship between the upper and lower view recognition camera 13 and the first part P1 or the second part P2, particularly the perpendicularity of the first part P1 or the second part P2 relative to the optical axis of the upper and lower view recognition camera 13, changes due to thermal strain of the head 11 holding the first part P1, the stage 12 on which the second part P2 is placed, and the upper and lower view recognition camera 13, it is possible to estimate an appropriate amount of misalignment between the first part P1 and the second part P2 under each condition.

[0082] It is desirable that the number of edge detection points of the alignment mark M1 be selected appropriately based on the estimation accuracy of the learning model, the estimation speed, the required alignment accuracy, and the like.

[0083] Furthermore, the edge intensity at the edge detection point of the alignment mark M1 is calculated, for example, from the brightness of each pixel in the captured image of the alignment mark M1. Therefore, the detection data of the alignment mark M1 may include the brightness of any pixel in the captured image of the alignment mark M1, or values ​​calculated from a certain region, such as the average or variance. Furthermore, if the positional relationship between the upper and lower view recognition camera 13 and the alignment mark M1 changes, the detected shape of the alignment mark M1 may change. For example, if the positional relationship between the upper and lower view recognition camera 13 and the first component P1 changes due to thermal strain or the like, the diameter of the detection circle of the alignment mark M1 may be smaller than the diameter of the detection circle of the alignment mark M1 when the positional relationship between the upper and lower view recognition camera 13 and the first component P1 is ideal. Therefore, the detection data of the alignment mark M1 may include detection shape data. The detection shape data may change depending on the shape of the alignment mark M1. For example, if the alignment mark M1 is circular, the detection data may include the diameter of the circle, the lengths of the major and minor axes, the circularity, etc. Furthermore, the captured image may be used as input data because it includes the recognized position of the alignment mark M1, the edge intensity distribution, the brightness distribution, the detected shape of the alignment mark M1, etc. The same applies to the alignment marks M2, M3, and M4.

[0084] Furthermore, the device operation data of the alignment device 10 may include detection data of the alignment operation status detected at at least one detection point of the alignment device 10, such as temperature, humidity, displacement, etc. The detected temperature of the alignment device 10 when thermal strain is occurring in the alignment device 10 may be higher than the detected temperature of the alignment device 10 when thermal strain is occurring in the alignment device 10 in an ideal state where thermal strain is not occurring in the alignment device 10. Therefore, by including temperature measurement data of the alignment device 10 in the detection data of the operation status of the alignment device 10, even if the positional relationship between the upper and lower view recognition camera 13 and the first component P1 or the second component P2 changes due to thermal strain of the head 11 holding the first component P1, the stage 12 on which the second component P2 is placed, or the upper and lower view recognition camera 13, it is possible to estimate the appropriate amount of misalignment under each condition. Possible temperature detection points include the upper and lower field of view recognition camera 13 (e.g., camera body 13a), near heat-generating parts such as actuators in components constituting alignment device 10 such as head 11 and stage 12, the surface of head 11, the surface of stage 12, and the ambient atmosphere of alignment device 10. Temperature data does not need to be acquired at one point per detection point, but may be acquired at multiple points. Temperature data may also be acquired at each step of the alignment method, or may be acquired continuously at regular intervals regardless of the step. Temperature data may also be acquired as image data using thermography or the like.

[0085] (Embodiment 2) Next, a second embodiment will be described with reference to Fig. 9. Fig. 9 is a diagram schematically showing the configuration of an alignment system 1A according to the second embodiment.

[0086] As shown in FIG. 9, an alignment system 1A according to this embodiment includes an alignment device 10A and a control device 20. The alignment device 10A is a control device 20. The control device 20 controls the alignment device 10A.

[0087] The alignment device 10A in this embodiment is different from the alignment device 10 in the above-described embodiment 1 in that the component recognition camera 13A is used. That is, in this embodiment, the component recognition camera 13A is used instead of the upper and lower view recognition camera 13.

[0088] Like the upper and lower field of view recognition camera 13 in the first embodiment, the component recognition camera 13A includes a camera body 13a, an optical system 13b including a lens, a lighting device 13c, and a computing device (not shown). The component recognition camera 13A further includes a focus adjustment mechanism (not shown). The focus adjustment mechanism adjusts the focus position when capturing images of the alignment marks M1 to M4.

[0089] Furthermore, the component recognition camera 13A in this embodiment is positioned differently relative to the head 11 and the stage 12 when capturing images of the first component P1 and the second component P2 compared to the top and bottom view recognition camera 13 in the first embodiment. Specifically, in the first embodiment, the top and bottom view recognition camera 13 was positioned between the head 11 holding the first component P1 and the stage 12 on which the second component P2 is placed, but the component recognition camera 13A in this embodiment is positioned above the head 11 holding the first component P1.

[0090] When performing an alignment method (alignment operation) using alignment system 1A configured in this manner, the alignment method can be performed in the same manner as in the alignment method of the first embodiment shown in FIG.

[0091] 3, the first component P1 and the second component P2 are imaged using the component recognition camera 13A. Specifically, the component recognition camera 13A images the first component P1 and the second component P2 from directly above the first component P1 and the second component P2, with the focus adjusted to be on the alignment marks M1 and M2 of the first component P1. Thereafter, the component recognition camera 13A images the first component P1 and the second component P2, with the focus adjusted to be on the alignment marks M3 and M4 of the second component P2.

[0092] Fig. 9 shows an example of a recognition image K captured by the component recognition camera 13A. The recognition image K in Fig. 9 shows the positional relationship between the head 11 and the first and second components P1 and P2 when captured by the component recognition camera 13A. Note that the recognition image K does not show blurring due to focus deviation.

[0093] Next, the positions of the first component P1 and the second component P2 are detected in the same manner as in step SA2 of FIG. 3. In this case, the position of the first component P1 is calculated from an image captured so that the alignment marks M1 and M2 of the first component P1 are in focus, and the position of the second component P2 is calculated from an image captured so that the alignment marks M3 and M4 of the second component P2 are in focus. Note that in this embodiment, the positions of the first component P1 and the second component P2 are detected separately from the two images, but this is not limiting. Specifically, the positions of the first component P1 and the second component P2 may be detected from a single image obtained by combining the two images.

[0094] Next, the control device 20 estimates the amount of misalignment between the first part P1 and the second part P2 in the same manner as in step SA3 of Fig. 3, and then determines whether or not an alignment operation to align the first part P1 and the second part P2 is necessary in the same manner as in step SA4 of Fig. 3, and if it is determined that an alignment operation is necessary, the alignment operation is performed in the same manner as in step SA5 of Fig. 3. Also in this embodiment, a learning model is generated in the same manner as in the first embodiment.

[0095] In this embodiment, among the operation parameters acquired by the operation parameter acquisition unit 21 in step SA5 (performing the alignment operation) in the alignment method of FIG. 3 and step SL1 (acquiring operation parameters) in the learning model generation method of FIG. 4, the production parameters may include imaging conditions such as the focus position of the component recognition camera 13A and the shutter speed, gain, or output amount (illumination output amount) of the camera body 13a set when the component recognition camera 13A images the first component P1, the second component P2, etc.

[0096] As described above, in the alignment method and alignment system 1A according to this embodiment, similar to the first embodiment, a learning model is used that outputs the amount of relative positional deviation between the first component P1 and the second component P2 in response to input data including the recognition results of at least one of the first component P1 and the second component P2 captured by the component recognition camera 13A, and the position of at least one of the head 11 and the stage 12 is controlled based on the amount of relative positional deviation output by this learning model.

[0097] As a result, similar to the first embodiment, even if the perpendicularity of the head 11 and / or the component recognition camera 13A changes due to heat, it is possible to prevent a decrease in the alignment accuracy between the first component P1 and the second component P2.

[0098] Furthermore, in this embodiment, the alignment marks M1 and M2 of the first component P1 and the alignment marks M3 and M4 of the second component P2 are imaged from above the first component P1. As a result, in this embodiment, unlike in the first embodiment, it is not necessary to insert the upper and lower view recognition camera 13 between the first component P1 and the second component P2 when imaging the first component P1 and the second component P2 during alignment. This shortens the distance between the first component P1 and the second component P2 in the Z-axis direction, minimizing horizontal misalignment between the first component P1 and the second component P2 caused by the head 11 descending. This improves the alignment accuracy between the first component P1 and the second component P2.

[0099] (Variation) The alignment method and alignment system according to the present disclosure have been described above based on the first and second embodiments, but the present disclosure is not limited to the first and second embodiments.

[0100] For example, in the first and second embodiments, the alignment devices 10 and 10A are component mounters, but are not limited to this. Specifically, the alignment devices 10 and 10A may be devices that use a first component P1 as a mold to press it against an accurate position on a second component P2 for processing. In this case, the alignment devices 10 and 10A may be, for example, imprint devices in which the first component P1 is an imprint mold having a concave-convex structure and the second component P2 is a workpiece to be imprinted. Furthermore, the alignment devices 10 and 10A may not be processing devices such as imprint devices, but may be inspection devices in which the first component P1 is an inspection probe, the second component P2 is an electronic component, and the first component P1 is accurately positioned relative to electrodes of the second component P2.

[0101] In the above description, the elements constituting the control device 20 (such as the operation parameter acquisition unit 21) may be circuits. These circuits may form a single circuit as a whole, or each may be a separate circuit. Furthermore, each of these circuits may be a general-purpose circuit or a dedicated circuit. Furthermore, the processes described as the operation of the elements constituting the control device 20 may be executed by a computer. For example, a computer executes a program using hardware resources such as a processor (CPU), memory, and input / output circuits to execute each of the above processes. Specifically, the processor executes each process by acquiring data to be processed from a memory (such as the storage unit 27) or an input / output circuit, etc., calculating the data, and outputting the calculation results to the memory or the input / output circuit, etc.

[0102] Furthermore, the alignment method (operation method) in the above embodiment can be used as a processing method for positioning the first part P1 and the second part P2, or a manufacturing method for a product including the first part P1 and the second part P2.

[0103] The present disclosure also includes forms obtained by applying various modifications to the above-described embodiments that would occur to a person skilled in the art, and forms realized by arbitrarily combining the components and functions of the embodiments within the scope of the present disclosure. The present disclosure also includes any combination of two or more claims from the claims set forth in the claims at the time of filing, provided that there is no technical contradiction. For example, when a dependent claim set forth in the claims at the time of filing is made into a multiple claim or multiple multiple claims that cite all of the superordinate claims within the scope of the technical contradiction, the present disclosure also includes any combination of all claims included in the multiple claim or multiple multiple claims. [Industrial Applicability]

[0104] The techniques of the present disclosure can be used to align two components. [Explanation of symbols]

[0105] 1. 1A Alignment System 10, 10A Alignment device 11 heads 12 stages 13. Up and down view recognition camera 13a Camera body 13b Optical system 13c lighting equipment 13A Part Recognition Camera 20 Control device 21 Operation parameter acquisition unit 22 Data processing section 23 Learning model generation unit 24 Position deviation estimation unit 25 Operation command section 26 Operation control section 27 Memory section 30 Edge detection point cloud P1 First part P2 2nd part M1, M2, M3, M4 alignment marks PC recognition position Pl, Pr detection points K Recognition Image

Claims

1. A method for aligning a first component held by a head and a second component placed on a stage by capturing images of the first component and the second component with a camera, the method comprising: a learning model that outputs a relative positional deviation amount between the first part and the second part when input data including a recognition result of at least one of the first part and the second part captured by the camera is input; controlling the position of at least one of the head and the stage based on the amount of relative positional deviation output by the learning model; Alignment method.

2. the input data includes a calculation result of a relative positional deviation amount between the first component and the second component. The alignment method according to claim 1 .

3. the input data includes calculated edge strengths of characteristic shapes for calculating a recognition result of at least one of the first part and the second part. The alignment method according to claim 1 .

4. the input data includes at least one of an outer dimension of a characteristic shape and a recognition position of the characteristic shape for calculating a recognition result of at least one of the first part and the second part; The alignment method according to claim 1 .

5. the input data includes luminance of a feature shape for calculating a recognition result of at least one of the first part and the second part. The alignment method according to claim 1 .

6. the input data includes a recognition image in which a portion of one or both of the first part and the second part is captured; The alignment method according to claim 1 .

7. the input data includes at least one of temperature, humidity, displacement, and strain at one or more measurement points of the alignment device; The alignment method according to claim 1 .

8. The input data includes at least one of the following: the elapsed time since the alignment device was powered on; the elapsed time since the alignment operation started; the number of cycles of the alignment operation; the time required for one cycle of the alignment operation; the time required for each step of the alignment operation; and the elapsed time from the time when any predetermined condition in the alignment operation was satisfied. The alignment method according to claim 1 .

9. the input data includes data relating to at least one or more production parameters of an axis position, an axis movement speed, and an acceleration for each step of an alignment operation in a drive unit of the alignment device, a shutter speed, a gain, and an amount of illumination output in a camera of the alignment device, and shapes, materials, and refractive indices of the first part and the second part; The alignment method according to claim 1 .

10. An alignment system for aligning a first part and a second part, comprising: a head for holding the first component; a stage on which the second component is placed; a camera that captures images of the first component and the second component; a control device that uses a learning model that outputs a relative positional deviation amount between the first part and the second part when input data including a recognition result of at least one of the first part and the second part imaged by the camera is input, and controls the position of at least one of the head and the stage based on the relative positional deviation amount output by the learning model. Alignment system.

Images