Inspection system and inspection method

Abstract

This inspection system (100) comprises: a first illumination unit (31) in which a first polarizing filter (31a) is disposed and that irradiates a resin molded article (200) with illumination light; a first imaging unit (32) in which a second polarizing filter (32a) having a polarization axis angle deviated from that of the first polarizing filter (31a) is disposed and that images the resin molded article (200); and a control unit (41) that performs control to detect, on the basis of a first inspection image (71) of the resin molded article (200) captured using the first illumination unit (31) and the first imaging unit (32), a silver defect (61) that occurs as a silver-colored defect on the resin molded article (200) due to an abnormality during resin molding.

Description

Inspection System and Inspection Method

[0006]

[0001] The present disclosure relates to an inspection system and an inspection method.

[0002] Conventionally, an inspection system for inspecting resin molded products is known. For example, International Publication No. 2013 / 021968 discloses an inspection system for inspecting linear scratches on resin molded products. This inspection system includes an optical microscope and a calculation unit. The optical microscope includes a light source and an imaging unit, irradiates light from the light source onto the resin molded product, and captures an image of the resin molded product by the imaging unit to obtain a microscopic image of the resin molded product. The calculation unit inspects the linear scratches on the resin molded product based on the microscopic image obtained by the optical microscope.

[0003] International Publication No. 2013 / 021968

[0004] Here, in addition to the linear scratches described in International Publication No. 2013 / 021968, there are also defects that appear on the appearance of resin molded products. As one such defect, there is a silver defect that occurs as a silver-colored defect in the resin molded product due to an abnormality during resin molding. Since the silver defect is also a type of defect in the resin molded product, it needs to be inspected. However, the technology described in Patent Document 1 above is a technology for inspecting linear scratches on resin molded products, and thus is not suitable as a technology for inspecting silver defects in resin molded products. In addition, since the silver defect appears as a defect with a small difference in brightness and darkness from the peripheral part of the silver defect, simply irradiating light from a light source and capturing an image of the resin molded product as described in Patent Document 1 may not clearly show the silver defect in the captured image. In this case, it is difficult to accurately detect the silver defect. Therefore, it is desired to accurately detect the silver defect.

[0005] The present disclosure has been made to solve the above problems, and one object of the present disclosure is to provide an inspection system and an inspection method capable of accurately detecting silver defects.

[0006] To achieve the above objective, the inventors of this invention conducted diligent studies and found that by arranging polarizing filters with offset polarization axes in the illumination unit and the imaging unit, it is possible to obtain an image that clearly captures silver defects that occur as silver defects in resin molded products due to abnormalities during resin molding, and thus completed the configuration of this disclosure. Specifically, the first surface inspection system comprises a first illumination unit in which a first polarizing filter is arranged and illumination light is irradiated onto the resin molded product; a first imaging unit in which a second polarizing filter with an offset polarization axis from the first polarizing filter is arranged and images the resin molded product; and a control unit that performs control to detect silver defects that occur as silver defects in resin molded products due to abnormalities during resin molding, based on a first inspection image of the resin molded product captured using the first illumination unit and the first imaging unit.

[0007] The first phase inspection system, as described above, comprises a first illumination unit that irradiates illumination light onto the resin molded product and has a first polarizing filter positioned thereon, and a first imaging unit that images the resin molded product and has a second polarizing filter positioned thereon, with the polarization axis angle shifted from that of the first polarizing filter. As a result, a first inspection image clearly showing silver defects can be obtained using the first illumination unit with the first polarizing filter positioned thereon, and the first imaging unit with the second polarizing filter positioned thereon, with the polarization axis angle shifted from that of the first polarizing filter. The control unit then performs control to detect silver defects that occur as silver defects on the resin molded product due to abnormalities during resin molding, based on the first inspection image of the resin molded product captured using the first illumination unit and the first imaging unit. As a result, silver defects can be detected with high accuracy based on the first inspection image clearly showing the silver defects.

[0008] The second phase of the inspection method comprises: irradiating the resin molded product with illumination light using a first illumination unit on which a first polarizing filter is positioned; imaging the resin molded product using a first imaging unit on which a second polarizing filter, whose polarization axis angle is shifted from that of the first polarizing filter, is positioned; and detecting silver defects that occur as silver defects on the resin molded product due to abnormalities during resin molding, based on a first inspection image of the resin molded product captured using the first illumination unit and the first imaging unit.

[0009] The second phase of the inspection method, as described above, comprises irradiating the resin molded product with illumination light using a first illumination unit where a first polarizing filter is positioned, and imaging the resin molded product using a first imaging unit where a second polarizing filter, whose polarization axis angle is shifted from that of the first polarizing filter, is positioned. This allows for the acquisition of a first inspection image clearly showing silver defects using the first illumination unit with the first polarizing filter and the first imaging unit with the second polarizing filter, whose polarization axis angle is shifted from that of the first polarizing filter. Based on the first inspection image of the resin molded product captured using the first illumination unit and the first imaging unit, silver defects occurring as silver defects in the resin molded product due to abnormalities during resin molding are detected. This provides an inspection method capable of accurately detecting silver defects based on a first inspection image clearly showing the silver defects.

[0010] According to this disclosure, as described above, silver defects can be detected with high accuracy.

[0011] This is a diagram illustrating the schematic of the inspection system according to the first embodiment. This is a block diagram of the inspection system according to the first embodiment. This is a diagram illustrating an example of a signal generated by the inspection system according to the first embodiment. This is a diagram illustrating the relative movement of the inspection unit of the inspection system according to the first embodiment. This is a diagram illustrating the inspection work in response to the relative movement of the inspection unit of the inspection system according to the first embodiment. This is a diagram illustrating the curved relative movement of the inspection unit of the inspection system according to the first embodiment. This is a diagram illustrating a first configuration example of the inspection unit of the inspection system according to the first embodiment. This is a diagram illustrating a second configuration example of the inspection unit of the inspection system according to the first embodiment. This is a diagram illustrating a defect in a resin molded product. This is a diagram illustrating two types of inspection images of the inspection system according to the first embodiment. This is a diagram illustrating the detection of a defect in a resin molded product using a machine learning model of the inspection system according to the first embodiment. This is a diagram illustrating the acquisition of a heatmap using a machine learning model of the inspection system according to the first embodiment. This is a diagram illustrating the schematic of the inspection system according to the second embodiment. This is a block diagram of the inspection system according to the second embodiment. This is a flowchart illustrating the control process of the inspection system according to the second embodiment. This is a diagram illustrating the generation of a movement path of the inspection system according to the second embodiment. This is a diagram illustrating the generation of coordinate transformation information according to the second embodiment. This figure illustrates a coordinate transformation table that associates the robot's movement amount with the coordinate values ​​of the robot coordinate system according to the second embodiment. This figure illustrates a coordinate transformation table that associates the robot's movement amount with the coordinate values ​​of the workpiece coordinate system according to the second embodiment. This figure illustrates an inspection operation on a workpiece according to the second embodiment. This figure illustrates an inspection image according to the second embodiment. This figure illustrates how to show the location of a defect on an actual workpiece according to the second embodiment. This figure illustrates how to show the location of a defect on a three-dimensional image of a workpiece according to the second embodiment. This is a flowchart illustrating the control process of the inspection system according to the third embodiment. This figure illustrates an operation image for displaying an inspection image according to the third embodiment. This is a diagram illustrating the outline of the inspection system according to the fourth embodiment.This is a flowchart illustrating the control process of the inspection system according to the fourth embodiment. This is a diagram illustrating the outline of the inspection system according to the fifth embodiment. This is a diagram illustrating the setting of teaching points and movement paths for the robot teaching device according to the fifth embodiment. This is a diagram illustrating the reflection state of illumination light for the robot teaching device according to the fifth embodiment. This is a diagram illustrating the outline of the inspection system according to the sixth embodiment. This is a diagram illustrating a virtual image displayed on a wearable display device according to the sixth embodiment. This is a diagram illustrating a virtual image displayed on a fixed display device according to the sixth embodiment. This is a flowchart illustrating the control process of the inspection system according to the sixth embodiment. This is a diagram illustrating an inspection system according to a modification of the first embodiment.

[0012] [First Embodiment] The configuration of the inspection system 100 according to the first embodiment will be described.

[0013] As shown in Figure 1, the inspection system 100 is a robotic system that performs inspection work on the object to be inspected, which is a resin molded product 200. The inspection system 100 comprises a robot 10 and a robot control device 20 that controls the robot 10. The inspection system 100 also comprises an inspection unit 30 and an inspection control device 40 that controls the inspection unit 30. The inspection system 100 also comprises a display unit 50 that displays information such as inspection results. The display unit 50 includes a monitor such as a liquid crystal monitor.

[0014] Robot 10 is, for example, an industrial robot. Robot 10 includes multiple joints. For example, robot 10 includes six vertical articulated joints. Robot 10 is powered by AC power supplied from an external source.

[0015] As shown in Figure 2, the robot control device 20 includes a robot control unit 21 and a signal output unit 22. The signal output unit 22 has an enable generation unit 23 and a pulse generation unit 24. The signal output unit 22 is an example of a control unit.

[0016] The robot control unit 21 controls the movement of the robot 10. Specifically, the robot control unit 21 controls the movement of the robot 10 by controlling the power supplied to the motors 14 provided at each joint of the robot 10. The robot control unit 21 also includes a CPU (Central Processing Unit) and memory. The robot control unit 21 controls the operation of the robot 10 by executing a predetermined program. The robot control unit 21 also receives instructions from the user regarding the movement of the robot 10 and controls the robot 10 to perform the movements based on the instructions. Specifically, the robot control unit 21 receives the position and orientation of the control points of the robot 10 and calculates the movement of each joint of the robot 10.

[0017] As shown in Figure 1, the robot 10 includes six joints 12a, 12b, 12c, 12d, 12e, and 12f, and links 13a, 13b, 13c, 13d, and 13e connecting each joint. Each of the six joints 12a through 12f is also provided with a motor 14 consisting of a servo motor and a position detection unit 15 for detecting the rotational position of each joint, as shown in Figure 2. Furthermore, as shown in Figure 1, an inspection unit 30 is attached to one end of the robot 10. The robot 10 also includes a base 11 at the other end, which can be attached to a floor, wall, column, etc.

[0018] Each of the six joints 12a to 12f rotates under the drive of the motor 14.

[0019] The first axis joint 12a is connected to the base 11. Joint 12a rotates link 13a around the rotation axis A1 relative to the base 11. The second axis joint 12b rotates link 13b relative to link 13a around the rotation axis A2, which is perpendicular to the rotation axis A1.

[0020] The third joint 12c rotates link 13c relative to link 13b around a rotation axis A3 that is parallel to the rotation axis A2. The fourth joint 12d rotates link 13d relative to link 13c around a rotation axis A4 that is perpendicular to the rotation axis A3.

[0021] The fifth joint 12e rotates link 13e relative to link 13d around a rotation axis A5 perpendicular to the rotation axis A4. The sixth joint 12f rotates the inspection unit 30 relative to link 13e around a rotation axis A6 perpendicular to the rotation axis A5.

[0022] The inspection unit 30 performs inspection work on the resin molded product 200. The inspection unit 30 includes an illumination unit and an imaging unit.

[0023] The inspection unit 30 performs an inspection on the resin molded product 200 while moving relative to the resin molded product 200. The resin molded product 200 is, for example, an interior part for an automobile. The resin molded product 200 is, for example, unpainted. The resin molded product 200 may or may not have a textured finish.

[0024] As shown in Figure 2, the inspection control device 40 includes an inspection control unit 41 and a storage unit 42. The inspection control unit 41 controls the inspection work performed by the inspection unit 30 on the resin molded product 200. The storage unit 42 includes non-volatile memory and stores a machine learning model 43. The inspection control unit 41 is an example of a control unit.

[0025] Here, the inspection control unit 41 controls the inspection work performed by the inspection unit 30 on the resin molded product 200 based on the signal output unit 22 of the robot control device 20.

[0026] Furthermore, the signal output unit 22 outputs a pulse signal based on the relative movement of the inspection unit 30 relative to the resin molded product 200, for each movement of the inspection unit 30 located at the tip of the robot 10. Specifically, the signal output unit 22 generates a pulse signal based on the relative movement of the inspection unit 30 based on the detection result of the position detection unit 15. More specifically, the robot control unit 21 acquires the detection result of the position detection unit 15 and calculates the relative movement of the inspection unit 30 based on the acquired detection result of the position detection unit 15. The signal output unit 22 acquires the relative movement of the inspection unit 30 calculated by the robot control unit 21 and generates a pulse signal based on the relative movement of the inspection unit 30 based on the acquired relative movement of the inspection unit 30.

[0027] Specifically, the signal output unit 22 outputs a pulse signal based on the relative movement of the inspection unit 30 relative to the resin molded product 200, using a variable frequency pulse signal. For example, the signal output unit 22 generates a pulse enable using the enable generation unit 23. The signal output unit 22 also generates a pulse signal using the pulse generation unit 24 based on the pulse enable generated by the enable generation unit 23.

[0028] Furthermore, the signal output unit 22 outputs a pulse signal corresponding to the relative movement amount of the inspection unit 30 relative to the resin molded product 200 for each relative movement amount of the inspection unit 30. For example, as shown in Figure 3, the signal output unit 22 generates and outputs a pulse signal based on the relative movement amount of the inspection unit 30 at predetermined processing cycles. In other words, the signal output unit 22 acquires the end-effector movement amount as the relative movement amount of the inspection unit 30 relative to the resin molded product 200 at predetermined processing cycles. The signal output unit 22 then generates a number of pulse signals corresponding to the acquired relative movement amount. A pulse signal is generated for every x mm of relative movement. For example, if the relative movement is 5x mm in a predetermined cycle, five pulse signals are generated within the predetermined cycle. A pulse signal is counted as one on the rising edge and one on the falling edge. In other words, a pulse signal is counted as two due to the rising and falling edges. The frequency of the output pulse is variable, for example, in the range from 0 Hz to several MHz. In other words, as the relative displacement increases, the frequency of the output pulse increases, and as the relative displacement decreases, the frequency of the output pulse decreases.

[0029] In the example shown in Figure 3, the control cycle is 2 msec, and the amount of movement is acquired at each control cycle, with a pulse signal output based on the amount of movement. Note that the end-effector movement in Figure 3 represents the cumulative amount of movement from 0 mm. In other words, the difference in end-effector movement from the previous control cycle is acquired as the relative movement in the current control cycle. For example, if the end-effector movement in the previous control cycle was 10 mm and the end-effector movement in the current control cycle is 16 mm, the relative movement in the current control cycle will be acquired as 6 mm. Also, in the example shown in Figure 3, the pulse resolution is set to 1 mm / pulse. In other words, one pulse signal is output for every 1 mm movement. For example, if the movement is 2 mm, the number of output pulses is set to 2, and the pulse frequency is 1 kHz. If the movement is 3 mm, the number of output pulses is set to 3, and the pulse frequency is 1.5 kHz.

[0030] The signal output unit 22 outputs a pulse enable signal from the enable generation unit 23 at the start of a predetermined processing cycle, and the pulse generation unit 24 starts outputting pulses simultaneously with the pulse enable signal output. Furthermore, when the pulse generation unit 24 outputs its last pulse, the signal output unit 22 stops outputting the pulse enable signal from the enable generation unit 23. This prevents a surge in processing at the beginning of a predetermined processing cycle. As a result, there is no need to provide buffer time for calculations.

[0031] The signal output unit 22 may continuously output a pulse enable signal to the pulse generation unit 24 via the enable generation unit 23. Alternatively, the signal output unit 22 may stop outputting the pulse enable signal for a sufficiently small calculation period correction amount relative to the processing cycle via the enable generation unit 23. This ensures sufficient buffer time for calculations. For example, the calculation period correction amount is 40 μsec for a processing cycle of 2 msec.

[0032] Furthermore, the signal output unit 22 may, within the processing cycle, initially pause before generating pulses from the pulse generation unit 24. In other words, the signal output unit 22 outputs a pulse signal corresponding to the relative movement amount of the inspection unit 30 relative to the workpiece 200, so it does not output a pulse when the relative movement of the inspection unit 30 starts, when the relative movement amount is zero.

[0033] The signal output unit 22 includes, for example, an FPGA (Field Programmable Gate Array), and processing is performed by the FPGA.

[0034] If the CPU controlling the robot 10 were to directly control the pulse output function, the CPU load would increase, potentially making it impossible to accurately control high-frequency pulses. Therefore, the pulse output is controlled using a pulse control processing unit, such as an FPGA, which is separate from the CPU controlling the robot 10.

[0035] The CPU controlling the robot 10 calculates the relative movement of the end-effector, and the pulse control processing unit controls the pulse frequency and number of pulses based on the relative movement of the end-effector. By dividing the processing in this way, accurate pulse output is possible. Furthermore, since the pulse output section is controlled by a separately provided processing unit, the pulse output specifications, such as pulse-to-distance conversion and n-multiplied pulses, can be easily changed and expanded by changing the control parameters.

[0036] Furthermore, the signal output unit 22 acquires the relative movement amount of the inspection unit 30 during a predetermined processing cycle and outputs a pulse signal assuming that the relative movement is constant during the predetermined processing cycle. However, since the predetermined processing cycle is sufficiently small, even assuming a constant relative movement, it is not significantly different from the actual relative movement amount of the inspection unit 30.

[0037] Furthermore, the signal output unit 22 may acquire the relative movement amount of the inspection unit 30 based on the actual movement of the inspection unit 30, or it may acquire the relative movement amount of the inspection unit 30 based on the movement command of the robot 10 from the robot control unit 21.

[0038] Furthermore, when the robot 10 is moved by an external moving mechanism, the signal output unit 22 takes into account the movement by the external moving mechanism to obtain the relative movement amount of the inspection unit 30 with respect to the resin molded product 200. The external moving mechanism includes a travel axis and a rotary table that move the base 11 of the robot 10.

[0039] The relative movement of the inspection unit 30 relative to the resin molded product 200 is obtained based on the movement of the control point TCP shown in Figure 4, which controls the movement of the robot 10. The control point TCP for controlling the movement of the robot 10 is set, for example, at the inspection work position of the resin molded product 200 by the inspection unit 30. The control point TCP is set, for example, at the focal point of imaging by the inspection unit 30.

[0040] The inspection control unit 41 controls the inspection work performed by the inspection unit 30 on the resin molded product 200 based on the pulse signals generated by the signal output unit 22. Specifically, the inspection control unit 41 causes the inspection unit 30 to perform inspection work at regular intervals based on pulse signals derived from the relative movement of the inspection unit 30. For example, the inspection control unit 41 counts the pulse signals output from the signal output unit 22 to obtain the relative movement of the inspection unit 30. Then, the inspection control unit 41 causes the inspection unit 30 to perform inspection work on the resin molded product 200 each time the inspection unit 30 moves by a regular amount. In other words, the inspection control unit 41 controls the inspection unit 30 to perform imaging at regular intervals based on the relative movement of the inspection unit 30.

[0041] The robot control unit 21 moves the inspection unit 30 relative to the resin molded product 200 along the surface of the resin molded product 200 using the robot 10. For example, as shown in Figure 4, the robot control unit 21 moves the inspection unit 30 in a curved manner along the curved resin molded product 200 using the robot 10. The inspection control unit 41 controls the inspection unit 30 to perform inspection work for each movement amount L1 of the control point TCP.

[0042] Furthermore, as shown in Figure 6, the robot control unit 21 moves the inspection unit 30 in a curved manner relative to the robot 10 along the inspection work path which has a curved section of the resin molded product 200. The inspection control unit 41 controls the inspection unit 30 to perform work for each movement amount L1 of the control point TCP.

[0043] Specifically, as shown in FIG. 5, for the output of the pulse signal for each movement amount L1, a signal A for causing the inspection unit 30 to perform an inspection operation is turned on. Further, regardless of the moving speed of the inspection unit 30, the inspection control unit 41 turns on the signal A for causing the inspection unit 30 to perform an inspection operation for each movement amount L1 of the inspection unit 30.

[0044] (Configuration of the inspection unit) FIG. 7 shows a first configuration example of the inspection unit 30. The inspection unit 30 of the first configuration example includes a lighting unit 31, an imaging unit 32, and an imaging unit 33. Note that the lighting unit 31, the imaging unit 32, and the imaging unit 33 are examples of a first lighting unit, a first imaging unit, and a second imaging unit, respectively.

[0045] The lighting unit 31 irradiates illumination light onto the resin molded product 200. The lighting unit 31 includes a light source that generates illumination light. The light source is, for example, an LED (Light Emitting Diode) or the like. A polarizing filter 31a is disposed in the lighting unit 31. The polarizing filter 31a has a polarization axis, transmits light vibrating in the same direction as the polarization axis, and attenuates light vibrating in a direction other than the same direction as the polarization axis. The lighting unit 31 irradiates illumination light through the polarizing filter 31a.

[0046] The imaging unit 32 images the resin molded product 200. The imaging unit 32 includes a camera. The camera includes, for example, a line camera. A polarizing filter 32a is disposed in the imaging unit 32. The polarizing filter 32a has a polarization axis, transmits light vibrating in the same direction as the polarization axis, and attenuates light vibrating in a direction other than the same direction as the polarization axis. The polarization axis of the polarizing filter 32a is shifted from that of the polarizing filter 31a. For example, the polarization axes of the polarizing filter 31a and the polarizing filter 32a are shifted by 90 degrees. The imaging unit 32 images through the polarizing filter 32a.

[0047] The imaging unit 33 images the resin molded product 200. The imaging unit 33 includes a camera. The camera includes, for example, a line camera. No polarizing filter is disposed in the imaging unit 33.

[0048] In the first configuration example, the imaging unit 32 and the imaging unit 33 are arranged on both sides of the illumination unit 31. The optical axis of the imaging unit 32 is inclined with respect to the illumination unit 31. The optical axis of the imaging unit 33 is inclined on the opposite side of the imaging unit 32 with respect to the illumination unit 31. When imaging the resin molded product 200, the illumination light of the illumination unit 31 is irradiated perpendicularly to the imaging position on the surface of the resin molded product 200. Also, the optical axis of each of the imaging unit 32 and the imaging unit 33 is inclined with respect to the imaging position on the surface of the resin molded product 200.

[0049] FIG. 8 shows a second configuration example of the inspection unit 30. The inspection unit 30 of the second configuration example includes an illumination unit 31, an imaging unit 32, and an illumination unit 34. Note that the illumination unit 34 is an example of a second illumination unit.

[0050] The illumination unit 34 irradiates illumination light onto the resin molded product 200. The illumination unit 34 includes a light source that generates illumination light. The light source is, for example, an LED (Light Emitting Diode) or the like. A polarizing filter is not arranged in the illumination unit 34.

[0051] In the second configuration example, the illumination unit 31 and the illumination unit 34 are arranged on both sides of the imaging unit 32. The optical axis of the illumination unit 31 is inclined with respect to the imaging unit 32. The optical axis of the illumination unit 34 is inclined on the opposite side of the illumination unit 31 with respect to the imaging unit 32. When imaging the resin molded product 200, the illumination light of each of the illumination unit 31 and the illumination unit 34 is irradiated obliquely with respect to the imaging position on the surface of the resin molded product 20(). Also, the optical axis of the imaging unit 32 is perpendicular to the imaging position on the surface of the resin molded product 200.

[0052] (Inspection operation) In the first embodiment, the robot 10 relatively moves the illumination unit 31 and the imaging unit 32 with respect to the resin molded product 200. When the inspection unit 30 is in the first configuration example, the robot 10 relatively moves the illumination unit 31, the imaging unit 32, and the imaging unit 33 with respect to the resin molded product 200. When the inspection unit 30 is in the second configuration example, the robot 10 relatively moves the illumination unit i31, the imaging unit 32, and the illumination unit 34 with respect to the resin molded product 200.

[0053] The signal output unit 22 generates a pulse signal based on the relative movement of the illumination unit 31 and the imaging unit 32 relative to the resin molded product 200 for each relative movement of the illumination unit 31 and the imaging unit 32, and the inspection control unit 41 controls the imaging operation of the resin molded product 200 by the illumination unit 31 and the imaging unit 32 based on the pulse signal generated by the signal output unit 22. In the first configuration example, the inspection unit 30 generates a pulse signal based on the relative movement of the illumination unit 31, the imaging unit 32 and the imaging unit 33 relative to the resin molded product 200 for each relative movement of the illumination unit 31, the imaging unit 32 and the imaging unit 33, and the inspection control unit 41 controls the imaging operation of the resin molded product 200 by the illumination unit 31, the imaging unit 32 and the imaging unit 33 based on the pulse signal generated by the signal output unit 22. In the second configuration example, the inspection unit 30 generates a pulse signal based on the relative movement of the illumination unit 31, imaging unit 32, and illumination unit 34 relative to the resin molded product 200 for each relative movement of the illumination unit 31, imaging unit 32, and illumination unit 34, and the inspection control unit 41 controls the imaging operation of the resin molded product 200 by the illumination unit 31, imaging unit 32, and illumination unit 34 based on the pulse signal generated by the signal output unit 22.

[0054] (Detection of defects in resin molded products) The inspection control unit 41 performs control to detect defects in the resin molded product 200 based on the image captured by the inspection unit 30. The defects in the resin molded product 200 that are detected are defects that appear on the appearance of the resin molded product 200. As shown in Figure 9, defects that appear on the appearance of the resin molded product 200 include silver streaks 61, glossiness 62, scratches 63, and color unevenness 64. Silver streaks 61 are an example of a silver defect.

[0055] Silver streaks 61 are defects that occur on the resin molded product 200 as silvery defects due to abnormalities during resin molding. Silver streaks 61 occur when gas generated within the resin is stretched within the mold during resin molding, causing streaky flow marks to appear as silvery streaks on the surface of the resin molded product 200. In other words, silver streaks 61 are defects in which internal abnormalities of the resin molded product 200 manifest on the surface. Glossiness 62 is a glossy defect that occurs on the surface of the resin molded product 200. Scratches 63 are linear defects that occur on the surface of the resin molded product 200. Color unevenness 64 is a color defect that occurs on the surface of the resin molded product 200. Unlike silver streaks 61, glossiness 62, scratches 63, and color unevenness 64 are defects that occur on the surface and are unrelated to internal issues in the resin molded product 200.

[0056] Here, the silver streak 61 appears as a defect with a small difference in brightness between the silver streak 61 and its surrounding area. Therefore, simply illuminating the resin molded product 200 with light from a light source and taking an image may not clearly capture the silver streak 61 in the image. In this case, it is difficult to detect the silver streak 61 with high accuracy.

[0057] Therefore, after diligent research by the inventors of the present invention, they found that by arranging polarizing filters 31a and 32a, which have polarizing axes offset from each other, in the illumination unit 31 and the imaging unit 32, an image clearly showing the silver streak 61 can be obtained.

[0058] The exact mechanism is unclear, but it is speculated to be as follows: When illumination light is shone on the resin molded product 200, the illumination light not only reflects off the surface of the resin molded product 200 but also penetrates into the interior of the resin molded product 200. The illumination light specularly reflected off the surface of the resin molded product 200 is polarized by the action of the polarizing filter 31a, and is therefore attenuated by the polarizing filter 32a, which has a polarization axis angle shifted from that of the polarizing filter 31a. On the other hand, in the area where the silver streak 61 appears, the illumination light is diffusely reflected inside the resin molded product 200 due to an abnormality inside the resin molded product 200 that manifested as the silver streak 61. Since the diffusely reflected illumination light is unpolarized, it is less attenuated by the polarizing filter 32a. As a result, the difference in brightness between the silver streak 61 and its periphery is increased, making it possible to capture the silver streak 61 brightly relative to its periphery. Therefore, it is speculated that the silver streak 61 can be clearly captured in the image. Furthermore, the closer the angle difference between the polarization axes of polarizing filter 31a and polarizing filter 32a approaches 90 degrees, the clearer the silver streak 61 can be captured in the image.

[0059] Therefore, in the first embodiment, as shown in Figure 10, the inspection control unit 41 performs control to detect silver streaks 61 based on the inspection image 71 of the resin molded product 200 captured using the illumination unit 31 and the imaging unit 32. That is, the inspection control unit 41 performs control to detect silver streaks 61 based on the inspection image 71 captured through both the polarizing filter 31a and the polarizing filter 32a. The inspection image 71 is an example of a first inspection image.

[0060] Furthermore, the inventors of the present invention have newly discovered that in images captured using the illumination unit 31 and the imaging unit 32, silver streaks 61 are clearly visible, while defects in the resin molded product 200 that are different from silver streaks 61, such as gloss 62, scratches 63, or color unevenness 64, are not clearly visible.

[0061] The mechanism is not entirely clear, but it is speculated that the following occurs: Defects in the resin molded product 200 that are different from silver streaks 61, such as glossiness 62, scratches 63, or color unevenness 64, occur on the surface and are unrelated to the interior of the resin molded product 200. Therefore, unlike silver streaks 61, the illumination light does not scatter inside the resin molded product 200 due to abnormalities inside the resin molded product 200. For this reason, it is speculated that defects in the resin molded product 200 that are different from silver streaks 61 do not appear clearly in the captured image taken using the illumination unit 31 and the imaging unit 32.

[0062] Therefore, in the first embodiment, the inspection control unit 41 performs the following: a control to detect silver streaks 61 on the resin molded product 200 based on the inspection image 71, and a control to detect defects in the resin molded product 200 that are different from the silver streaks 61 based on an inspection image 72 of the resin molded product 200 captured in a polarization state different from that of the inspection image 71. Specifically, the inspection control unit 41 performs the control to detect defects in the resin molded product 200 that are different from the silver streaks 61 based on an inspection image 72 captured through one of the polarizing filters 31a and 32a but not through the other. Based on the inspection image 72, the inspection control unit 41 performs the control to detect at least one of gloss 62, scratches 63 and color unevenness 64 as defects in the resin molded product 200 that are different from the silver streaks 61. The inspection control unit 41 may perform the control to detect all of gloss 62, scratches 63 and color unevenness 64, or it may perform the control to detect one or some of these. The inspection image 72 is an example of a second inspection image.

[0063] Note that in Figure 10, for ease of understanding, examples are shown in inspection image 71 and inspection image 72 where all of the silver streaks 61, gloss 62, scratches 63, and color unevenness 64 are visible. However, it is not necessary for all of the silver streaks 61, gloss 62, scratches 63, and color unevenness 64 to be visible. Also, in Figure 10, for ease of understanding, the silver streaks 61, gloss 62, scratches 63, and color unevenness 64 that are clearly visible in the captured image are represented by solid lines, and the ones that are not clearly visible in the captured image are represented by dashed lines.

[0064] In the first configuration example, the inspection unit 30 performs control to detect defects in the resin molded product 200 that are different from the silver streak 61, based on the inspection image 72 captured using the illumination unit 31 and the imaging unit 33. That is, the inspection control unit 41 performs control to detect defects in the resin molded product 200 that are different from the silver streak 61, based on the inspection image 72 captured via the polarizing filter 31a but without the polarizing filter 32a. In the first configuration example, since imaging can be performed without the polarizing filter 32a, the illumination light that has become polarized due to the action of the polarizing filter 31a is not attenuated by the polarizing filter 32a. Also, defects in the resin molded product 200 that are different from the silver streak 61, such as gloss 62, scratches 63, or color unevenness 64, appear as defects with a large difference in brightness between the defect and the surrounding area. For these reasons, defects in the resin molded product 200 that are different from silver streaks 61, such as glossiness 62, scratches 63, or color unevenness 64, can be clearly captured in the inspection image 72 taken using the illumination unit 31 and the imaging unit 33.

[0065] Furthermore, if the inspection unit 30 is the first configuration example, the inspection control unit 41 controls the imaging unit 32 and imaging unit 33 to simultaneously image the resin molded product 200. The inspection control unit 41 controls the imaging unit 32 and imaging unit 33 to image the resin molded product 200 at the same timing while irradiating the resin molded product 200 with illumination light using the illumination unit 31. Based on the pulse signal output from the signal output unit 22, the inspection control unit 41 controls the imaging unit 32 and imaging unit 33 to simultaneously image the resin molded product 200 at regular intervals of a certain relative movement amount of the illumination unit 31, imaging unit 32 and imaging unit 33 relative to the resin molded product 200. Furthermore, if the imaging unit 32 and imaging unit 33 are line cameras, the inspection control unit 41 controls the imaging unit 32 and imaging unit 33 to scan and image the resin molded product 200 while the illumination unit 31, imaging unit 32 and imaging unit 33 are moved relative to the resin molded product 200. In this case, the inspection image 71 is obtained by stitching together the images captured by the imaging unit 32 at each of the multiple imaging positions of the scan imaging. Similarly, the inspection image 72 is obtained by stitching together the images captured by the imaging unit 33 at each of the multiple imaging positions of the scan imaging.

[0066] In the case of the second configuration example, the inspection control unit 41 performs control to detect defects in the resin molded product 200 that are different from silver streaks 61, based on the inspection image 72 of the resin molded product 200 captured using the illumination unit 34 and the imaging unit 32. That is, the inspection control unit 41 performs control to detect defects in the resin molded product 200 that are different from silver streaks 61, based on the inspection image 72 captured via the polarizing filter 32a but without the polarizing filter 31a. In the second configuration example, since the illumination unit 34 emits unpolarized illumination light, the illumination light is less attenuated by the polarizing filter 32a. Also, defects in the resin molded product 200 that are different from silver streaks 61, such as glossiness 62, scratches 63, or color unevenness 64, appear as defects with a large difference in brightness between the defect and the surrounding area. For these reasons, defects in the resin molded product 200 that are different from silver streaks 61, such as glossiness 62, scratches 63, or color unevenness 64, can be clearly captured in the inspection image 72 taken using the illumination unit 34 and the imaging unit 32.

[0067] Furthermore, in the case of the second configuration example, the inspection control unit 41 controls the imaging unit 32 to image the resin molded product 200 while alternately turning on the illumination unit 31 and the illumination unit 34. The inspection control unit 41 controls the imaging unit 32 to image the resin molded product 200 while alternately irradiating the resin molded product 200 with illumination light using the illumination unit 31 and the illumination unit 34. While one of the illumination unit 31 or illumination unit 34 is lit, the other is turned off. Images taken with illumination light irradiated by the illumination unit 31 and images taken with illumination light irradiated by the illumination unit 34 are obtained alternately. Note that the same location on the resin molded product 200 is imaged under different illumination conditions of the illumination unit 31 and the illumination unit 34.

[0068] The inspection control unit 41 controls the illumination unit 31 and illumination unit 34 to alternately light up at regular intervals of a certain relative movement of the illumination unit 31, imaging unit 32, and illumination unit 34 relative to the resin molded product 200, based on the pulse signal output from the signal output unit 22. The inspection control unit 41 also controls the imaging unit 32 to image the resin molded product 200 at regular intervals of a certain relative movement of the illumination unit 31, imaging unit 32, and illumination unit 34 relative to the resin molded product 200, based on the pulse signal output from the signal output unit 22. Furthermore, if the imaging unit 32 and imaging unit 33 are line cameras, the inspection control unit 41 controls the imaging unit 32 to scan and image the resin molded product 200 while the illumination unit 31, imaging unit 32, and illumination unit 34 are moved relative to the resin molded product 200. In this case, the inspection image 71 is an image created by stitching together the images captured by the imaging unit 32 at each of the multiple imaging positions where illumination light is irradiated by the illumination unit 31 during the scan imaging. Furthermore, in the scanning imaging process, the images captured by the imaging unit 32 at each of the multiple imaging positions illuminated by the illumination unit 34 are combined to form the inspection image 72.

[0069] Furthermore, in the first embodiment, as shown in Figure 11, the inspection control unit 41 uses a machine learning model 43 that has been trained to detect the silver streaks 61 of the resin molded product 200 using the inspection image 71 as input to perform control for detecting the silver streaks 61 of the resin molded product 200. The machine learning model 43 is generated by machine learning. Machine learning can be, for example, supervised learning using training data, or unsupervised learning without training data. The machine learning model 43 may also be generated by a learning device separate from the robot control device 20 and the inspection control device 40, or it may be generated by the robot control device 20 or the inspection control device 40. The generated machine learning model 43 is stored in the storage unit 42 of the inspection control device 40.

[0070] The inspection control unit 41 inputs the inspection image 71 to the machine learning model 43 and controls the acquisition of output results from the machine learning model 43. The machine learning model 43 may output a judgment result of whether the inspection is good or bad based on the presence or absence of silver streaks 61, or it may output an image with silver streaks 61 emphasized, or it may output both. If the machine learning model 43 outputs an image with silver streaks 61 emphasized, the inspection control unit 41 may determine whether the inspection is good or bad based on the presence or absence of silver streaks 61 based on the image from the machine learning model 43. The inspection control unit 41 controls the display unit 50 to display at least one of the judgment result of whether the inspection is good or bad based on the presence or absence of silver streaks 61 and the image with silver streaks 61 emphasized. This allows the user to check either the judgment result of whether the inspection is good or bad or the image with silver streaks 61 emphasized.

[0071] For example, as shown in Figure 12, the machine learning model 43 outputs a heat map 73 that is colored to change according to the accuracy of the silver streaks 61 on the resin molded product 200. The heat map 73 is colored, for example, by varying the hue according to the accuracy of the silver streaks 61, or by varying the brightness of a single hue according to the accuracy of the silver streaks 61. The inspection control unit 41 uses the machine learning model 43 to control the acquisition of the heat map 73. The inspection control unit 41 also determines the quality of the inspection based on the presence or absence of silver streaks 61 based on the heat map 73. For example, the inspection control unit 41 acquires the degree of abnormality based on the heat map 73 and determines the quality of the inspection based on the presence or absence of silver streaks 61 by comparing the degree of abnormality with a threshold value.

[0072] Furthermore, the machine learning model 43 may be trained to detect defects other than silver streaks 61 on the resin molded product 200, such as glossiness 62, scratches 63, or color unevenness 64, using the inspection image 72 as input. In this case, a single machine learning model 43 can detect both silver streaks 61 and defects other than silver streaks 61, eliminating the need to generate multiple machine learning models. Also, when a single machine learning model 43 detects both silver streaks 61 and defects other than silver streaks 61, the machine learning model 43 may be configured not to distinguish between silver streaks 61 and defects other than silver streaks 61. In this case, since the model does not distinguish between silver streaks 61 and defects other than silver streaks 61, the number of images required for machine learning can be reduced, thus reducing the effort required for machine learning.

[0073] Furthermore, when detecting defects other than the silver streak 61 of the resin molded product 200, the machine learning model 43 can be configured in the same way as when detecting the silver streak 61. That is, the machine learning model 43 may output a judgment result of whether the inspection is good or bad based on the presence or absence of defects other than the silver streak 61, or it may output an image that highlights defects other than the silver streak 61, or it may output both. In addition, the machine learning model 43 may output a heat map that is colored to change according to the accuracy of the defects other than the silver streak 61.

[0074] (Effects of the First Embodiment) In the first embodiment, the following effects can be obtained.

[0075] The inspection system 100 includes an illumination unit 31 that irradiates illumination light onto the resin molded product 200 and has a polarizing filter 31a, and an imaging unit 32 that images the resin molded product 200 and has a polarizing filter 32a that is offset in polarization axis angle from the polarizing filter 31a. As a result, an inspection image 71 clearly showing the silver streak 61 can be obtained using the illumination unit 31 with the polarizing filter 31a and the imaging unit 32 with the polarizing filter 32a that is offset in polarization axis angle from the polarizing filter 31a.The inspection control unit 41 then performs control to detect the silver streak 61 that occurs as a silvery defect on the resin molded product 200 due to an abnormality during resin molding, based on the inspection image 71 of the resin molded product 200 captured using the illumination unit 31 and the imaging unit 32.As a result, the silver streak 61 can be detected with high accuracy based on the inspection image 71 that clearly shows the silver streak 61.

[0076] The inspection control unit 41 performs the following: a control to detect silver streaks 61 on the resin molded product 200 based on the inspection image 71, and a control to detect defects in the resin molded product 200 other than silver streaks 61 based on the inspection image 72 of the resin molded product 200 captured in a polarization state different from that of the inspection image 71. Here, the inventors of the present invention have newly discovered that in the inspection image 71 captured using the illumination unit 31 and the imaging unit 32, silver streaks 61 are clearly visible, while defects in the resin molded product 200 other than silver streaks 61 are not clearly visible. Therefore, as described above, the inspection control unit 41 performs the following: a control to detect silver streaks 61 on the resin molded product 200 based on the inspection image 71, and a control to detect defects in the resin molded product 200 other than silver streaks 61 based on the inspection image 72 of the resin molded product 200 captured in a polarization state different from that of the inspection image 71. As a result, defects in the resin molded product 200 that are different from the silver streak 61 can be clearly captured in the inspection image 72, which is captured under a different polarization state than the inspection image 71, and defects in the resin molded product 200 that are different from the silver streak 61 can be accurately detected based on the inspection image 72 which clearly captures defects in the resin molded product 200 that are different from the silver streak 61. As a result, both the silver streak 61 and defects in the resin molded product 200 that are different from the silver streak 61 can be accurately detected.

[0077] The inspection control unit 41 performs control to detect defects in the resin molded product 200 that are different from the silver streak 61, based on the inspection image 72 captured through one of the polarizing filters 31a and 32a but not through the other. This makes it easy and clear to capture defects in the resin molded product 200 that are different from the silver streak 61 in the inspection image 72, and allows for easy and accurate detection of defects in the resin molded product 200 that are different from the silver streak 61 based on the inspection image 72 that clearly captures these defects.

[0078] An imaging unit 33 is provided to image the resin molded product 200, and the inspection control unit 41 performs control to detect defects in the resin molded product 200 that are different from the silver streak 61, based on the inspection image 72 captured using the illumination unit 31 and the imaging unit 33. As a result, the illumination unit 31 can be used in common with both the imaging unit 32 and the imaging unit 33, so there is no need to provide an independent illumination unit for each of the imaging unit 32 and the imaging unit 33. As a result, both the silver streak 61 and defects in the resin molded product 200 that are different from the silver streak 61 can be detected with high accuracy while suppressing structural complexity and size increase. Furthermore, if the illumination unit 31, imaging unit 32 and imaging unit 33 are arranged at the tip of the robot 10, if the structure of the tip of the robot 10 becomes large, it may become difficult to approach the resin molded product 200 and difficult to inspect the resin molded product 200. In contrast, since the size of the structure at the tip of the robot 10 can be suppressed, it becomes difficult to approach the resin molded product 200, and it becomes difficult to inspect the resin molded product 200.

[0079] The inspection control unit 41 controls the imaging unit 32 and imaging unit 33 to simultaneously image the resin molded product 200. As a result, the imaging unit 32 and imaging unit 33 can simultaneously image the resin molded product 200, thus reducing the time required for the imaging unit 32 and imaging unit 33 to image the resin molded product 200.

[0080] An illumination unit 34 is provided to irradiate the resin molded product 200 with illumination light, and the inspection control unit 41 performs control to detect defects in the resin molded product 200 that are different from the silver streak 61, based on the inspection image 72 captured using the illumination unit 34 and the imaging unit 32. As a result, the imaging unit 32 can capture both the inspection image 71 and the inspection image 72, so there is no need to provide two separate imaging units to capture the inspection image 71 and the inspection image 72. As a result, both the silver streak 61 and defects in the resin molded product 200 that are different from the silver streak 61 can be detected with high accuracy while suppressing structural complexity and size increase. Furthermore, when the illumination unit 31, illumination unit 34 and imaging unit 32 are arranged at the tip of the robot 10, if the structure at the tip of the robot 10 becomes large, it may become difficult to approach the resin molded product 200 and difficult to inspect the resin molded product 200. In contrast, since the size of the structure at the tip of the robot 10 can be suppressed, it becomes difficult to approach the resin molded product 200, and it becomes difficult to inspect the resin molded product 200.

[0081] The inspection control unit 41 controls the imaging unit 32 to image the resin molded product 200 while alternately illuminating the illumination unit 31 and illumination unit 34. This makes it easy for the imaging unit 32 to capture both the inspection image 71 and the inspection image 72.

[0082] The inspection control unit 41 performs control based on the inspection image 72 to detect at least one of glossiness 62, scratches 63, and color unevenness 64 as defects of the resin molded product 200 that are different from the silver streak 61. This makes it possible to accurately detect at least one of glossiness 62, scratches 63, and color unevenness 64 as defects of the resin molded product 200 that are different from the silver streak 61.

[0083] The inspection control unit 41 uses a machine learning model 43, which has been trained to detect silver streaks 61 on the resin molded product 200, as input to the inspection image 71, to perform control for detecting silver streaks 61 on the resin molded product 200. This makes it easy and quick to perform the process of detecting silver streaks 61 on the resin molded product 200 using the machine learning model 43.

[0084] The inspection control unit 41 uses a machine learning model 43 to control the acquisition of a heat map 73 that is colored according to the accuracy of the silver streaks 61 on the resin molded product 200. This makes it possible to intuitively and easily grasp the position of the silver streaks 61 on the resin molded product 200 based on the color of the heat map 73.

[0085] Polarizing filters 31a and 32a have polarization axes that are offset by 90 degrees. This makes it easy to obtain an inspection image 71 that clearly shows the silver streak 61.

[0086] A robot 10 is provided to move the illumination unit 31 and the imaging unit 32 relative to the resin molded product 200. This allows the resin molded product 200 to be inspected while the illumination unit 31 and the imaging unit 32 are moved relative to the resin molded product 200 by the robot 10, thereby enabling inspection of various parts of the resin molded product 200.

[0087] The signal output unit 22 generates a pulse signal based on the relative movement of the illumination unit 31 and the imaging unit 32 relative to the resin molded product 200 for each relative movement of the illumination unit 31 and the imaging unit 32. The inspection control unit 41 controls the imaging operation of the resin molded product 200 by the illumination unit 31 and the imaging unit 32 based on the generated pulse signal. This allows the relative movement of the illumination unit 31 and the imaging unit 32 relative to the resin molded product 200 to be acquired for each relative movement, and the operation by the illumination unit 31 and the imaging unit 32 to be controlled. As a result, the operation of the resin molded product 200 can be performed without pre-setting all work positions. Consequently, when the robot 10 performs work while moving the illumination unit 31 and the imaging unit 32 relative to the resin molded product 200, the complexity of the setting operation can be suppressed.

[0088] [Second Embodiment] The configuration of the inspection system 100a according to the second embodiment will be described.

[0089] As shown in Figures 13 and 14, the inspection system 100a includes a robot 10, a robot control device 20a, an inspection unit 30, an inspection control device 40, a display unit 50, an instruction unit 130, an image processing device 150, and a result display device 160. The robot control device 20a includes a signal output unit 22 and a robot control unit 140.

[0090] The indicator unit 130 is, for example, located in the inspection unit 30 and indicates the location of the defect 201, which will be described later, obtained by inspection, on the resin molded product 200. The indicator unit 130 is a laser irradiation unit and irradiates laser light to indicate the location of the defect 201 on the resin molded product 200. Note that the indicator unit 130 may be located in a part other than the inspection unit 30 if it moves along with the movement of the robot 10.

[0091] The robot control unit 140 includes a processing unit 141 and a storage unit 142. The processing unit 141 includes a processor and performs various processes related to the operation of the robot 10. The storage unit 142 includes non-volatile memory and stores coordinate transformation information 171 and 172, which will be described later.

[0092] The image processing device 150 performs image processing on the image captured by the inspection unit 30. The image processing device 150 also controls the imaging timing performed by the inspection unit 30. The image processing device 150 includes a processing unit 151 and a storage unit 152. The processing unit 151 includes a processor and performs various processing related to the image captured by the inspection unit 30 and the imaging timing performed by the inspection unit 30. The storage unit 152 includes a non-volatile memory and stores inspection images 121, which will be described later. The image processing device 150 may also be configured with an inspection control device 40.

[0093] The result display device 160 displays the inspection results of the resin molded product 200. The result display device 160 includes a processing unit 161, a storage unit 162, a display unit 163, and an operation unit 164. The processing unit 161 includes a processor and performs various processes related to displaying the inspection results of the resin molded product 200. The storage unit 162 includes non-volatile memory and stores coordinate transformation information 172, a three-dimensional image of the resin molded product 200, etc. The display unit 163 includes a monitor such as a liquid crystal monitor and displays a screen showing the inspection results of the resin molded product 200. The operation unit 164 includes input devices such as a mouse and keyboard and accepts user input operations. The display unit 163 and the operation unit 164 may be integrated. That is, the display unit 163 and the operation unit 164 may be configured as an operation unit / display unit such as a touch panel. The display unit 163 may be configured as a display unit 50.

[0094] (Control processing of the inspection system) The control processing of the inspection system 100a will be explained below.

[0095] As shown in Figure 15, in step S1, the processing unit 141 of the robot control unit 140 generates a movement path 113 for the robot 10 when the robot 10 moves the inspection unit 30 relative to the resin molded product 200 and has the inspection unit 30 perform the inspection work. As shown in Figure 16, the movement path 113 is a path for operating the robot 10, and multiple paths are generated. For example, the processing unit 141 receives instructions from the user on the operation of the robot 10 and generates a movement path 113 for the robot 10 based on the received instructions.

[0096] In step S2, the processing unit 141 of the robot control unit 140 performs the process of generating coordinate transformation information 171 shown in Figure 18 and coordinate transformation information 172 shown in Figure 19 based on the generated movement path 113. The coordinate transformation information 171 and 172 is information that converts the coordinate values ​​of the inspection coordinate system of the inspection image 121, which will be described later, obtained by inspecting the resin molded product 200 by the inspection unit 30, into coordinate values ​​of a three-dimensional coordinate system that can represent the coordinate values ​​of the resin molded product 200 in three dimensions. The inspection coordinate system is a two-axis orthogonal coordinate system that is orthogonal to each other, and the three-dimensional coordinate system is a three-axis orthogonal coordinate system that is orthogonal to each other. Details of the coordinate transformation using the coordinate transformation information 171 and 172 will be described later.

[0097] As shown in Figure 17, the processing unit 141 performs a process to acquire coordinate values ​​of the three-dimensional coordinate system at first distance intervals D1 along the movement path 113 and generate coordinate transformation information 171 and 172. At this time, the processing unit 141 actually moves the inspection unit 30 along the movement path 113 relative to the resin molded product 200 using the robot 10 and performs a process to acquire coordinate values ​​of the three-dimensional coordinate system at first distance intervals D1. The first distance interval D1 is the distance interval of the control points 114a. The processing unit 141 performs a process to acquire the coordinate values ​​of the three-dimensional coordinate system of the control points 114a at first distance intervals D1. The control points 114a are set at the focal position of the inspection unit 30 for imaging. The focal position of the inspection unit 30 for imaging is set near the surface of the resin molded product 200. The control points 114a are provided for the process of acquiring coordinate values ​​of the three-dimensional coordinate system. Furthermore, for example, the first distance interval D1 is larger than the second distance interval D2, which will be described later, when inspecting the resin molded product 200.

[0098] As shown in Figures 18 and 19, the coordinate transformation information 171 and 172 are coordinate transformation tables that associate the amount of movement of the robot 10 along the movement path 113 with the coordinate values ​​of the three-dimensional coordinate system. In Figures 18 and 19, the path number represents the number of the movement path 113, the position number represents the number of the control point, the amount of movement represents the amount of movement of the control point 114a of the robot 10 along the movement path 113, and the coordinate value represents the coordinate value of the control point 114a in the three-dimensional coordinate system. In other words, in the coordinate transformation information 171 and 172, for each movement path 113, the amount of movement of the robot 10 for each control point 114a is associated with the coordinate value of the control point 114a in the three-dimensional coordinate system.

[0099] As shown in Figure 18, in the coordinate transformation information 171, the three-dimensional coordinate system is the robot coordinate system relating to the robot 10. The coordinate transformation information 171 is a coordinate transformation table that associates the amount of movement of the robot 10 with the coordinate values ​​of the robot coordinate system. In the coordinate transformation information 171, the coordinate values ​​used are those that indicate the position and orientation of the control point 114a in the robot coordinate system.

[0100] As shown in Figure 19, in the coordinate transformation information 172, the three-dimensional coordinate system is the work coordinate system related to the resin molded product 200. The work coordinate system is a coordinate system based on the resin molded product 200. The coordinate transformation information 172 is a coordinate transformation table that associates the amount of movement of the robot 10 with the coordinate values ​​of the work coordinate system. In the coordinate transformation information 172, the coordinate values ​​used are those that indicate the position of the control point 114a in the work coordinate system.

[0101] Furthermore, the processing unit 141 stores the coordinate transformation information 171 and 172 in the storage unit 142, and also outputs the coordinate transformation information 172 to the processing unit 161 of the result display device 160. The processing unit 161 stores the coordinate transformation information 172 in the storage unit 162.

[0102] In step S3, the processing unit 141 of the robot control unit 140 operates the robot 10 based on the movement path 113 and performs inspection of the resin molded product 200 by the inspection unit 30. Then, the processing unit 151 of the image processing device 150 performs the process of acquiring the inspection image 121 shown in Figure 21 based on the output result of the inspection unit 30. The inspection image 121 is an image of the surface of the resin molded product 200 captured by the inspection unit 30. The inspection image 121 may also be inspection image 71 or inspection image 72.

[0103] As shown in Figure 20, the robot control unit 140 operates the inspection unit 30 to inspect the resin molded product 200 at second distance intervals D2 along the movement path 113 and acquires inspection images 121. Specifically, the processing unit 151 operates the inspection unit 30 to image the resin molded product 200 at second distance intervals D2 and scans the resin molded product 200. More specifically, the processing unit 141 outputs a pulse signal to the processing unit 151 at second distance intervals D2. Based on the pulse signal from the processing unit 141, the processing unit 151 outputs a trigger signal to the inspection unit 30 at second distance intervals D2. Based on the trigger signal, the inspection unit 30 images the resin molded product 200 at second distance intervals D2. The second distance interval D2 is the distance interval of the control points 114b. The control points 114b are set to the focal position for imaging by the inspection unit 30.

[0104] In step S4, the processing unit 151 of the image processing device 150 performs a process to detect defects 201 in the resin molded product 200 in the inspection image 121 shown in Figure 21. The processing unit 151 performs a process to detect defects 201 in the inspection image 121 by performing predetermined image processing on the inspection image 121. Defects 201 include silver streaks 61, glossiness 62, scratches 63, or color unevenness 64. The processing unit 151 performs the process to detect defects 201 in all inspection images 121.

[0105] As shown in Figure 21, the inspection coordinate system of the inspection image 121 is a two-dimensional coordinate system in which the direction along the movement path 113 is the Y-axis direction and the direction perpendicular to the movement path 113 is the X-axis direction. The processing unit 151 performs the process of acquiring the coordinate values ​​of the inspection coordinate system of the defect 201. That is, the processing unit 151 performs the process of acquiring the coordinate values ​​of the X axis and Y axis of the inspection coordinate system of the defect 201.

[0106] In step S5, the processing unit 141 of the robot control unit 140 performs a process to convert the coordinate values ​​of the inspection coordinate system of the defect 201 to the coordinate values ​​of the robot coordinate system based on the coordinate transformation information 171. Also in step S5, the processing unit 161 of the result display device 160 performs a process to convert the coordinate values ​​of the inspection coordinate system of the defect 201 to the coordinate values ​​of the three-dimensional coordinate system based on the coordinate transformation information 172.

[0107] In step S6, as shown in Figure 22, the processing unit 141 of the robot control unit 140 performs a process to indicate the location of the defect 201 on the actual resin molded product 200 based on the converted coordinate values ​​of the defect 201 in the three-dimensional coordinate system. Specifically, the processing unit 141 operates the robot 10 based on the coordinate values ​​of the defect 201 converted to robot coordinate system coordinate values, and performs a process to indicate the location of the defect 201 on the actual resin molded product 200 using the indicator unit 130. That is, the processing unit 141 operates the robot 10 to move the indicator unit 130 to a predetermined position where the location of the defect 201 can be indicated. Then, with the indicator unit 130 positioned in the predetermined position, the processing unit 141 irradiates laser light from the indicator unit 130 to indicate the location of the defect 201 on the actual resin molded product 200.

[0108] In step S7, as shown in Figure 23, the processing unit 161 of the result display device 160 performs a process to indicate the location of the defect 201 on the 3D image of the resin molded product 200 based on the coordinate values ​​of the converted defect 201 in the 3D coordinate system. Specifically, the processing unit 161 performs a process to indicate the location of the defect 201 on the 3D image of the resin molded product 200 based on the coordinate values ​​of the defect 201 converted to coordinate values ​​in the work coordinate system. That is, the processing unit 161 performs a process to superimpose an image indicating the location of the defect 201 onto the 3D image of the resin molded product 200. Then, the processing unit 161 performs a process to display the 3D image of the resin molded product 200 with the superimposed image indicating the location of the defect 201 on the display unit 163. For example, a round mark is displayed as the image indicating the location of the defect 201. The 3D image of the resin molded product 200 with the superimposed image indicating the location of the defect 201 can be enlarged, reduced, or rotated based on user operation using the operation unit 164.

[0109] (Effects of the second embodiment) Based on the generated coordinate transformation information 171 and 172, the following processes are performed: converting the coordinate values ​​of the inspection coordinate system of the defect 201 to coordinate values ​​of the three-dimensional coordinate system; and showing the position of the defect 201 on the actual resin molded product 200 or on a three-dimensional image of the resin molded product 200 based on the converted coordinate values ​​of the three-dimensional coordinate system of the defect 201. As a result, the position of the defect 201 can be shown on the actual resin molded product 200 or on a three-dimensional image of the resin molded product 200. Unlike when the position of the defect 201 is shown on a two-dimensional image of the resin molded product 200, the position of the defect 201 can be shown with high accuracy even on curved surfaces or complex curved surfaces of the resin molded product 200.

[0110] [Third Embodiment] The configuration of the inspection system 100b according to the third embodiment will be described.

[0111] The configuration of the inspection system 100b is the same as that of the inspection system 100a of the second embodiment shown in Figures 13 and 14. Specifically, the inspection system 100b comprises a robot 10, a robot control device 20a, an inspection unit 30, an inspection control device 40, a display unit 50, an instruction unit 130, an image processing device 150, and a result display device 160.

[0112] (Control processing of the inspection system) The control processing of the inspection system 100b will be explained below.

[0113] As shown in Figure 24, the operations from step S1 to S5 are the same as in the second embodiment described above.

[0114] In step S11, as shown in Figure 25, the processing unit 161 of the result display device 160 performs a process to indicate the location of the defect 201 on the 3D image of the resin molded product 200 based on the coordinate values ​​of the converted defect 201 in the 3D coordinate system. Specifically, the processing unit 161 performs a process to indicate the location of the defect 201 on the 3D image of the resin molded product 200 based on the coordinate values ​​of the defect 201 converted to coordinate values ​​in the work coordinate system. That is, the processing unit 161 performs a process to superimpose an image indicating the location of the defect 201 onto the 3D image of the resin molded product 200. Then, the processing unit 161 performs a process to display the 3D image of the resin molded product 200 with the superimposed image indicating the location of the defect 201 on the display unit 163. For example, a round mark is displayed as the image indicating the location of the defect 201. The 3D image of the resin molded product 200 with the superimposed image indicating the location of the defect 201 can be enlarged, reduced, or rotated based on user operation using the operation unit 164.

[0115] Furthermore, the processing unit 161 performs a process to display the inspection results of the resin molded product 200 in list format. The inspection results of the resin molded product 200 represent the results of a process in which defects 201 in the inspection images 121 are detected for all inspection images 121. The inspection results of the resin molded product 200 include the number of the detected defect 201 and the type of the detected defect 201. The types of defects 201 include silver streaks 61, glossiness 62, scratches 63, or color unevenness 64.

[0116] The processing unit 161 performs the process of displaying the inspection results of the resin molded product 200 along with the operation images 271 and 272 in list format. Operation image 271 is an image used to operate the robot 10 so that the location of the defect 201 (described later) is shown on the actual resin molded product 200. Operation image 272 is an image used to display the inspection image 121 of the defect 201 (described later). Operation images 271 and 272 are displayed in list format along with the inspection results of the resin molded product 200 so that the correspondence between the detected defect 201 and its number and type can be identified. The processing unit 161 performs the process of displaying the list-format inspection results of the resin molded product 200 and the operation images 271 and 272 on the display unit 163. The processing unit 161 also performs the process of displaying the 3D image of the resin molded product 200, the list-format inspection results of the resin molded product 200, and the operation images 271 and 272 within the same frame.

[0117] When the operation image 271 is operated, the processing unit 161 performs the process in step S12. In step S12, the processing unit 161 operates the robot 10 to show the location of the defect 201 corresponding to the operated operation image 271 on the actual resin molded product 200. Specifically, the processing unit 161 outputs identification information to the processing unit 141 to identify the defect 201 corresponding to the operated operation image 271. The identification information is, for example, the defect number 201. Based on the identification information from the processing unit 161, the processing unit 141 identifies the defect 201 and operates the robot 10 to show the location of the identified defect 201 on the actual resin molded product 200 as the location of the defect 201 corresponding to the operated operation image 271. Operation of the operation image 271 means, for example, clicking the operation image 271 with a mouse. If the display unit 163 is a touch panel, operation of the operation image 271 means that the operation image 271 is touched by the user.

[0118] More specifically, as shown in Figure 22, the processing unit 141 operates the robot 10 to indicate the location of the defect 201 on the actual resin molded product 200 using the indicator unit 130. That is, the processing unit 141 operates the robot 10 to move the indicator unit 130 to a predetermined position where it can indicate the location of the defect 201. Then, with the indicator unit 130 positioned in the predetermined location, the processing unit 141 irradiates a laser beam from the indicator unit 130 to indicate the location of the defect 201 on the actual resin molded product 200.

[0119] When the operation image 272 is operated, the processing unit 161 performs the process in step S13. In step S13, the processing unit 161 performs the process of displaying the inspection image 121 of the defect 201 corresponding to the operated operation image 272. Specifically, the processing unit 161 performs the process of displaying the inspection image 121 shown in Figure 21 in a frame different from the frame in which the 3D image of the resin molded product 200, the list-format inspection results of the resin molded product 200, and the operation images 271 and 272 are displayed. Alternatively, the inspection image 121 may be displayed in the same frame as the frame in which the 3D image of the resin molded product 200, the list-format inspection results of the resin molded product 200, and the operation images 271 and 272 are displayed. Furthermore, when the user has completed the processing of the defect 201, they may use the operation unit 164 to check the checkbox 273. Note that operation of the operation image 272 refers to clicking the operation image 272 with a mouse, etc. Furthermore, if the display unit 163 is a touch panel, the operation of the operation image 272 refers to the user touching the operation image 272.

[0120] (Effects of the third embodiment) The process involves displaying the location of the detected defect 201 on a three-dimensional image of the resin molded product 200, and, if input is made regarding the displayed defect 201, performing processing on the actual resin molded product 200 with respect to the input defect 201. As a result, the user can not only confirm the location of the defect 201 on the resin molded product 200 using the three-dimensional image of the resin molded product 200, but also perform processing on the actual resin molded product 200 with respect to the input defect 201 by making input regarding the defect 201. As a result, the user's convenience regarding the image of the resin molded product 200 can be improved compared to simply displaying the location of the defect 201 on an image of the resin molded product 200.

[0121] [Fourth Embodiment] The configuration of the inspection system 100c according to the fourth embodiment will be described.

[0122] As shown in Figure 19, the inspection system 100c is equipped with multiple robots 10. Furthermore, the inspection unit 30, the instruction unit 130, and the robot control unit 140 shown in Figure 16 are arranged for each robot 10. The other configurations of the inspection system 100c are the same as those of the inspection system 100a in the second embodiment described above. Inspection work may be performed in parallel on a single resin molded product 200 by multiple inspection units 30, or inspection work may be performed in parallel on multiple resin molded products 200 placed on a belt conveyor or turntable by multiple inspection units 30.

[0123] (Control processing of the inspection system) The control processing of the inspection system 100c will be explained below.

[0124] As shown in Figure 27, the operations from step S1 to S5 are the same as in the second embodiment described above, but in the fourth embodiment, the operations from step S1 to S5 are performed in parallel for each robot 10.

[0125] In step S21, the processing unit 161 of the result display device 160 performs a process to integrate the locations of defects 201 detected from multiple inspection images 121 acquired by inspection units 30 located in each of the multiple robots 10 as data. Specifically, the processing unit 161 integrates the coordinate values ​​of the work coordinate system of the defects 201 acquired from the multiple inspection images 121 acquired by the multiple inspection units 30 into a single three-dimensional data. The integrated data is stored, for example, in a single file. The storage unit 162 of the result display device 160 stores the integrated data. The operations in the subsequent steps S6 and S7 are the same as in the second embodiment described above.

[0126] (Effects of the fourth embodiment) The inspection system 100c includes a processing unit 161 that performs processing to integrate the locations of defects 201 detected from multiple inspection images 121 acquired from inspection units 30 located in each of the multiple robots 10 into data. As a result, the locations of defects 201 of the resin molded product 200 inspected by multiple inspection units 30 are integrated into a single data, so that the locations of all defects 201 of the resin molded product 200 can be referenced by simply referring to this integrated single data once from another computer, for example. As a result, even when the resin molded product 200 is inspected by multiple inspection units 30, the handling of defect 201 data of the resin molded product 200 can be made easier.

[0127] [Fifth Embodiment] The configuration of the inspection system 100d according to the fifth embodiment will be described.

[0128] As shown in Figure 28, in the inspection system 100d, the inspection unit 300 includes an imaging unit 301 for imaging the resin molded product 200 and an illumination unit 302 for irradiating the resin molded product 200 with illumination light. The inspection system 100d also includes a robot teaching device 310. The other configurations of the inspection system 100d are the same as those of the inspection system 100 in the first embodiment described above. The inspection unit 300 may also be composed of an inspection unit 30. The imaging unit 301 may be composed of an imaging unit 32 or an imaging unit 33. The illumination unit 302 may be composed of an illumination unit 31 or an illumination unit 34.

[0129] The robot teaching device 310 is a device for offline teaching of the movements of a robot 10 for visual inspection of a resin molded product 200. The robot teaching device 310 performs offline teaching by simulating the movements of the robot 10 on a display screen without using an actual machine. The robot teaching device 310 is, for example, a personal computer. The robot teaching device 310 includes a display unit 311, an operation unit 312, a processing unit 313, and a storage unit 314. The display unit 311 includes a monitor such as an LCD monitor and displays a screen. The operation unit 312 includes input devices such as a mouse and a keyboard and accepts user input operations. The processing unit 313 includes a processor and performs various processes in the robot teaching device 310. The storage unit 314 includes non-volatile memory and stores the model M1 of the imaging unit 301, the model M2 of the illumination unit 302, the model M3 of the robot 10, and the model M4 of the resin molded product 200, which are used in the simulation. The detailed configuration of offline teaching using the robot teaching device 310 will be described later.

[0130] The imaging unit 301, the illumination unit 302, and the robot 10 are a device that actually performs visual inspection of the resin molded product 200 based on the results of the robot teaching device 310 teaching the robot 10's movements. The imaging unit 301 is a camera that images the resin molded product 200. The illumination unit 302 irradiates the resin molded product 200 with illumination light. The robot 10 integrally holds the imaging unit 301 and the illumination unit 302 at its tip.

[0131] In the visual inspection of the resin molded product 200, the resin molded product 200 is illuminated by the illumination unit 302 while the surface of the resin molded product 200 is imaged by the imaging unit 301. Furthermore, the imaging unit 301 and illumination unit 302 are moved relative to the resin molded product 200 by the robot 10 while the surface of the resin molded product 200 is imaged by the imaging unit 301. Since the location of defects in the resin molded product 200 is unknown, the resin molded product 200 is generally imaged multiple times to cover its entire surface. Based on the image results of the imaging unit 301, it is then inspected whether or not defects exist in the resin molded product 200.

[0132] (Offline Teaching) The processing unit 313 performs offline teaching by simulation based on the model M1 of the imaging unit 301, the model M2 of the illumination unit 302, the model M3 of the robot 10, and the model M4 of the resin molded product 200 stored in the memory unit 314. The processing unit 313 displays at least one of the imaging unit 301, the illumination unit 302, the robot 10, and the resin molded product 200 on the display unit 311 and performs offline teaching by simulation. In offline teaching, as shown in Figure 29, the processing unit 313 displays the teaching points P of the robot 10's movement and the movement path PA defined by the teaching points P on the display unit 311 based on user input operations using the operation unit 312.

[0133] The processing unit 313 acquires a movement path PA by the robot 10 along the curved surface of the resin molded product 200, which is approximately perpendicular to the curved surface of the resin molded product 200, and displays the acquired movement path PA on the display unit 311. Specifically, when a line L is input by the user to the surface of the resin molded product 200 displayed on the display unit 311, the processing unit 313 creates a normal vector V that is approximately perpendicular to the surface of the resin molded product 200 at the location where the line L passes. The tip of the normal vector V is the teaching point P. As a result, the processing unit 313 acquires multiple teaching points P along the curved surface of the resin molded product 200, which is approximately perpendicular to the curved surface of the resin molded product 200. Furthermore, based on the acquired multiple teaching points P, the processing unit 313 acquires a movement path PA along the curved surface of the resin molded product 200, which is approximately perpendicular to the curved surface of the resin molded product 200.

[0134] Then, when teaching the robot 10's movements on the display unit 311, the processing unit 313 simulates the reflection state of the illumination light on the resin molded product 200 and displays an image related to the acquired reflection state of the illumination light on the display unit 311. As shown in Figure 30, for example, the processing unit 313 displays an image representing the reflection state of the illumination light on the resin molded product 200 displayed on the display unit 311. Also, for example, the processing unit 313 changes the image representing the reflection state of the illumination light on the resin molded product 200 displayed on the display unit 311 in accordance with the movement of the imaging unit 301 and illumination unit 302 by the robot 10. In Figure 30, the high-brightness area HB, where the amount of illumination light reflected is large and the brightness is large, is represented by hatching.

[0135] (Effects of the Fifth Embodiment) When performing a visual inspection of the resin molded product 200, the entire range of the image captured by the imaging unit 301 does not contribute to the visual inspection of the resin molded product 200. Rather, the high-brightness area HB, which has a large amount of reflected illumination light and high brightness, contributes to the visual inspection of the resin molded product 200. Therefore, by identifying the high-brightness area HB, it is possible to accurately determine which part of the resin molded product 200 is being inspected, and thus it is possible to teach the robot 10 to perform an inspection that does not miss or minimizes the inspection of the appearance of the resin molded product 200. Thus, as described above, by displaying an image related to the state of reflection of illumination light on the display unit 311, the high-brightness area HB that actually contributes to the visual inspection of the resin molded product 200 can be easily identified based on the image related to the state of reflection of illumination light displayed on the display unit 311, and the robot 10 can be taught to perform an inspection that does not miss or minimizes the inspection of the appearance of the resin molded product 200.

[0136] [Sixth Embodiment] The configuration of the inspection system 100e according to the sixth embodiment will be described.

[0137] As shown in Figure 31, in the inspection system 100e, the inspection unit 400 includes an imaging unit 401 for imaging the resin molded product 200 and an illumination unit 402 for irradiating the resin molded product 200 with illumination light. The inspection system 100e also includes a wearable display device 410 and a fixed display device 420. The other configurations of the inspection system 100e are the same as those of the inspection system 100 of the first embodiment described above. The inspection unit 400 may also be composed of an inspection unit 30. The imaging unit 401 may be composed of an imaging unit 32 or an imaging unit 33. The illumination unit 402 may be composed of an illumination unit 31 or an illumination unit 34.

[0138] As shown in Figure 32, the wearable display device 410 overlays a virtual image G generated by computer graphics onto the real-world image seen by the user U. In other words, the wearable display device 410 is a display unit that displays mixed reality. The wearable display device 410 is worn by the user U.

[0139] As shown in Figure 33, the fixed display device 420 is fixedly positioned and not attached to the user U. The fixed display device 420 displays an image of a three-dimensional model of the resin molded product 200. The image of the three-dimensional model of the resin molded product 200 is, for example, an image of the resin molded product 200 created using CAD (Computer-Aided Design). The fixed display device 420 is, for example, a liquid crystal display or an organic EL display.

[0140] (Function to display the imaging range) The inspection system 100e generates a virtual image G of the illumination light irradiated onto the resin molded product 200 from the illumination unit 402 using computer graphics, and displays the virtual image G of the illumination light on the display unit 411 of the wearable display device 410 so as to overlap with the real image seen by the user U. Also, on the display unit 421 of the fixed display device 420, displays the virtual image G of the illumination light so as to overlap with the image of the 3D model of the resin molded product 200. Furthermore, the function to display the imaging range is used, for example, when the user U is teaching the robot 10 what to do. The control method of the inspection system 100e for displaying the imaging range will be described in detail below.

[0141] As shown in Figure 34, in step S31, the robot control unit 21 receives an operation of the teaching pendant by the user U. Based on the received operation, the robot control unit 21 moves the robot 10 to move the imaging unit 401, including the illumination unit 402, relative to the resin molded product 200.

[0142] In step S32, the imaging unit 401 acquires the coordinates of the irradiated light in the image captured by the imaging unit 401, which is moved relative to the resin molded product 200. Specifically, while the imaging unit 401 is ON, the imaging unit 401 acquires the image captured by the imaging unit 401.

[0143] In step S33, the imaging unit 401 acquires the coordinates of the irradiated light in the image captured by the imaging unit 401, which is moved relative to the resin molded product 200.

[0144] In step S34, the wearable display device 410 and the fixed display device 420 each superimpose a virtual image G of the irradiated light, generated by computer graphics based on the coordinates of the acquired irradiated light and the control point TCP of the robot 10, onto the image of the molded resin product 200, and display it on the display unit 411 of the wearable display device 410 and the display unit 421 of the fixed display device 420. For the wearable display device 410, the image of the molded resin product 200 is the actual image seen by the user U. For the fixed display device 420, the image of the molded resin product 200 is an image of the three-dimensional model of the molded resin product 200.

[0145] In step S35, the robot control unit 21 determines whether the user U has finished moving the robot 10. If the answer in step S35 is no, the operations from step S31 to step S34 are repeated.

[0146] (Effects of the Sixth Embodiment) The wearable display device 410 and the fixed display device 420 each acquire the coordinates of the irradiated light in the image captured by the imaging unit 401, which is moved relative to the resin molded product 200. Based on the acquired coordinates of the irradiated light and the control point TCP of the robot 10, a virtual image G of the irradiated light generated by computer graphics is superimposed on the image of the resin molded product 200 and displayed on the display unit 411 of the wearable display device 410 and the display unit 421 of the fixed display device 420. As a result, the virtual image G of the irradiated light is displayed on the image of the resin molded product 200 based on the actual coordinates of the irradiated light and the actual control point TCP of the robot 10. In other words, since the shape and position are based on actual information, the errors in shape and position are small. Therefore, by visually observing the virtual image G of the irradiated light on the image of the resin molded product 200, the user U can confirm how the irradiated light hit the resin molded product 200, and thus accurately recognize the imageable range of the resin molded product 200 captured by the imaging unit 401.

[0147] (Variations) The embodiments disclosed herein should be considered in all respects to be illustrative and not restrictive. The scope of this disclosure is indicated by the claims rather than the description of the embodiments above, and further includes all modifications (variations) in the sense and scope equivalent to the claims.

[0148] In the first to sixth embodiments described above, an example was shown in which an inspection unit is provided at the tip of a robot and the inspection unit is moved by the robot to move the inspection unit relative to the resin molded product. However, the present disclosure is not limited thereto. In this disclosure, as shown in the example in Figure 35, a resin molded product 200 may be placed at the tip of the robot 10, and the inspection unit 30 may be moved relative to the resin molded product 200 by the robot 10 to move the resin molded product 200. In this case, the inspection unit 30 may perform work on the resin molded product 200 at predetermined movement amounts L2. Furthermore, when a resin molded product 200 is placed at the tip of the robot 10, an end effector may be provided at the tip of the robot 10, and the resin molded product 200 may be gripped by the end effector.

[0149] Furthermore, while the first embodiment described above shows an example in which a robot is provided to move the illumination unit and imaging unit relative to the resin molded product, the present disclosure is not limited thereto. In this disclosure, a moving device other than a robot may be provided to move the illumination unit and imaging unit relative to the resin molded product. In this case, the moving device may be, for example, an XY table or a rotary table. Also, in this disclosure, it is not necessary to provide a configuration in which the illumination unit and imaging unit are moved relative to the resin molded product. In this case, a stationary resin molded product may be imaged using a stationary illumination unit and imaging unit.

[0150] Furthermore, although the first embodiment described above shows an example in which the imaging unit includes a line camera, this disclosure is not limited thereto. In this disclosure, the imaging unit may include area cameras other than line cameras.

[0151] Furthermore, while the first embodiment described above showed an example where the silver defect was a streaky silver streak that occurs as a silver defect on a resin molded product due to an abnormality during resin molding, the present disclosure is not limited to this. In this disclosure, the silver defect is not limited to streaks, as long as it occurs as a silver defect on a resin molded product due to an abnormality during resin molding. The silver defect may also be spotted.

[0152] Furthermore, while the first embodiment described above shows an example where the angle of the polarizing axes of the two polarizing filters is offset by 90 degrees, the present disclosure is not limited to this. In the present disclosure, the angle of the polarizing axes of the two polarizing filters may be offset by an angle other than 90 degrees. However, from the viewpoint of clearly capturing silver streaks in the captured image, it is preferable that the angle of the polarizing axes of the two polarizing filters be close to 90 degrees.

[0153] Furthermore, although the first embodiment described above shows an example in which a polarizing filter is not provided in the imaging unit as the second imaging unit, the present disclosure is not limited thereto. In this disclosure, a polarizing filter may be provided in the second imaging unit if it is possible to clearly capture defects other than silver streaks in the captured image. In this case, it is preferable that the angle difference of the polarizing axis between the polarizing filter of the first illumination unit and the polarizing filter of the second imaging unit is smaller than that between the polarizing filter of the first illumination unit and the polarizing filter of the first imaging unit.

[0154] Furthermore, although the first embodiment described above shows an example in which a polarizing filter is not provided in the illumination unit as the second illumination unit, the present disclosure is not limited thereto. In this disclosure, a polarizing filter may be provided in the second illumination unit if it is possible to clearly capture defects other than silver streaks in the captured image. In this case, from the viewpoint of clearly capturing defects other than silver streaks in the captured image, it is preferable that the polarizing filter of the second illumination unit and the polarizing filter of the first imaging unit have a smaller deviation in the angle of the polarization axis compared to the polarizing filter of the first illumination unit and the polarizing filter of the first imaging unit.

[0155] Furthermore, while the first embodiment described above shows an example of acquiring an inspection image as a first inspection image and an inspection image as a second inspection image using either one illumination unit and two imaging units, or two illumination units and one imaging unit, the disclosure is not limited thereto. In this disclosure, from the viewpoint of clearly capturing defects other than silver streaks in the captured image, two illumination units and two imaging units may be used to acquire an inspection image as a first inspection image and an inspection image as a second inspection image.

[0156] Furthermore, while the first embodiment described above shows an example of detecting silver streaks and defects other than silver streaks, this disclosure is not limited thereto. In this disclosure, only silver streaks may be detected. In this case, there may be only one imaging unit and one illumination unit.

[0157] Furthermore, while the first embodiment described above shows an example of detecting at least one of gloss, scratches, and color unevenness as a defect different from silver streaks, the disclosure is not limited thereto. In this disclosure, defects other than gloss, scratches, and color unevenness may be detected as defects different from silver streaks.

[0158] Furthermore, while the first embodiment described above demonstrates an example of detecting silver streaks, gloss, scratches, or color unevenness using a machine learning model, this disclosure is not limited thereto. In this disclosure, silver streaks, gloss, scratches, or color unevenness may be detected using image processing other than machine learning models.

[0159] Furthermore, although the first embodiment described above shows an example in which the robot control unit, signal output unit, and inspection control unit are arranged separately, the disclosure is not limited thereto. In this disclosure, the robot control unit, signal output unit, and inspection control unit may be included in a common control unit. In this case, the common control unit may have separate processing units such as CPUs as the robot control unit, signal output unit, and inspection control unit, or it may have a common processing unit such as a CPU.

[0160] Furthermore, while the first embodiment described above shows an example of a robot configuration including six vertical joints, the disclosure is not limited thereto. In this disclosure, the robot may include five or fewer joints, or seven or more joints. The robot may also be a horizontal robot.

[0161] Furthermore, while the first embodiment described above shows an example of a configuration in which the relative movement amount of the inspection unit with respect to the resin molded product is obtained based on the movement of the robot's control point, the disclosure is not limited thereto. In this disclosure, the relative movement amount of the inspection unit with respect to the resin molded product may be obtained based on the movement of any position of the robot.

[0162] Furthermore, while the first embodiment described above shows an example of a configuration in which the robot control unit and the signal output unit are located in a common control device, the disclosure is not limited thereto. In this disclosure, the robot control unit and the signal output unit may be located in separate control devices. In addition, the signal output unit may be located in a common control device with the robot control unit by adding hardware, or by adding software.

[0163] Furthermore, while the first embodiment described above shows an example of a configuration in which a pulse signal based on the relative movement of the inspection unit relative to the resin molded product is output in accordance with the relative movement of the inspection unit relative to the resin molded product, the present disclosure is not limited thereto. In this disclosure, the relative position of the inspection unit relative to the resin molded product may be output in real time based on the movement of the resin molded product or the inspection unit provided at the tip of the robot. In this case, the position coordinates of the tip of the robot may be output. In this case, the robot may be moved at a low speed in advance to obtain the position coordinates of the tip of the robot, and then, when the robot is moved along the same path, a signal based on the relative movement of the inspection unit may be output in conjunction with the position coordinates of the tip of the robot, according to the relative movement of the sewing unit relative to the resin molded product.

[0164] Furthermore, while the second embodiment described above shows an example where the result display device is a monitor such as a liquid crystal monitor, the present disclosure is not limited to this. In the present disclosure, the result display device may be a wearable display device that displays mixed reality.

[0165] Furthermore, in the third embodiment described above, an example was shown in which, as processing related to an actual resin molded product, both a process of irradiating the actual resin molded product with laser light from an indicator unit to indicate the position of the target, and a process of displaying a three-dimensional image of the resin molded product, the inspection results of the resin molded product in list format, and the inspection image are performed. However, this disclosure is not limited to this. In this disclosure, only one of the above two processes may be performed.

[0166] Furthermore, while the sixth embodiment described above shows an example in which a virtual image of the irradiated light is displayed on both a wearable display device and a fixed display device, the present disclosure is not limited thereto. In this disclosure, the virtual image of the irradiated light may be displayed on only one of the wearable display device or the fixed display device.

[0167] The functions of the elements disclosed herein can be performed using circuits or processing circuits, including general-purpose processors, dedicated processors, integrated circuits, ASICs (Application Specific Integrated Circuits), conventional circuits, and / or combinations thereof, configured or programmed to perform the disclosed functions. A processor is considered a processing circuit or circuit because it includes transistors and other circuits. In this disclosure, a circuit, unit, or means is hardware that performs the enumerated functions, or hardware programmed to perform the enumerated functions. The hardware may be hardware disclosed herein, or other known hardware that is programmed or configured to perform the enumerated functions. If the hardware is a processor, which is considered a type of circuit, then the circuit, means, or unit is a combination of hardware and software, and the software is used to configure the hardware and / or the processor.

[0168] [Embodiments] The exemplary embodiments described above will be understood by those skilled in the art to be specific examples of the following embodiments.

[0169] (Aspect 1) An inspection system comprising: a first illumination unit which has a first polarizing filter and irradiates illumination light onto a resin molded product; a first imaging unit which has a second polarizing filter which has a polarization axis angle shifted from that of the first polarizing filter and images the resin molded product; and a control unit which performs control to detect silver defects that occur as silver defects on the resin molded product due to abnormalities during resin molding, based on a first inspection image of the resin molded product captured using the first illumination unit and the first imaging unit.

[0170] (Aspect 2) The inspection system according to aspect 1, wherein the control unit performs the following: control to detect the silver defect of the resin molded product based on the first inspection image; and control to detect a defect in the resin molded product other than the silver defect based on a second inspection image of the resin molded product captured in a polarization state different from the first inspection image.

[0171] (Aspect 3) The inspection system according to aspect 2, wherein the control unit performs control to detect defects in the resin molded product that are different from the silver defects, based on the second inspection image captured through one of the first polarizing filter and the second polarizing filter without passing through the other.

[0172] (Aspect 4) The inspection system according to aspect 3, further comprising a second imaging unit for imaging the resin molded product, wherein the control unit performs control to detect defects in the resin molded product that are different from the silver defects, based on the second inspection image captured using the first illumination unit and the second imaging unit.

[0173] (Aspect 5) The inspection system according to aspect 4, wherein the control unit controls the first imaging unit and the second imaging unit to simultaneously image the resin molded product.

[0174] (Aspect 6) The inspection system according to aspect 3, further comprising a second illumination unit that irradiates illumination light onto the resin molded product, wherein the control unit performs control to detect defects in the resin molded product that are different from the silver defects, based on the second inspection image captured using the second illumination unit and the first imaging unit.

[0175] (Aspect 7) The inspection system according to aspect 6, wherein the control unit controls the first imaging unit to image the resin molded product while alternately lighting the first illumination unit and the second illumination unit.

[0176] (Aspect 8) The inspection system according to any one of aspects 2 to 7, wherein the control unit performs control to detect at least one of gloss, scratches, and color unevenness as defects of the resin molded product that are different from the silver defects, based on the second inspection image.

[0177] (Aspect 9) The inspection system according to any one of aspects 1 to 8, wherein the control unit performs control to detect the silver defect in the resin molded product using a machine learning model that has been trained to detect the silver defect in the resin molded product with the first inspection image as input.

[0178] (Aspect 10) The inspection system according to aspect 9, wherein the control unit uses the machine learning model to perform control to acquire a heat map that is colored to change according to the accuracy of the silver defect in the resin molded product.

[0179] (Aspect 11) The inspection system according to any one of the aspects 1 to 10, wherein the first polarizing filter and the second polarizing filter have polarity axes offset by 90 degrees.

[0180] (Aspect 12) The inspection system according to any one of aspects 1 to 11, further comprising a robot for moving the first illumination unit and the first imaging unit relative to the resin molded product.

[0181] (Aspect 13) The inspection system according to aspect 12, wherein the control unit generates a pulse signal based on the relative movement amount of the first illumination unit and the first imaging unit with respect to the resin molded product for each relative movement amount of the first illumination unit and the first imaging unit, and controls the imaging operation of the resin molded product by the first illumination unit and the first imaging unit based on the generated pulse signal.

[0182] (Aspect 14) An inspection method comprising: irradiating a resin molded product with illumination light using a first illumination unit on which a first polarizing filter is arranged; imaging the resin molded product using a first imaging unit on which a second polarizing filter is arranged with a polarization axis angle shifted from that of the first polarizing filter; and detecting silver defects that occur as silver defects in the resin molded product due to abnormalities during resin molding, based on a first inspection image of the resin molded product captured using the first illumination unit and the first imaging unit.

[0183] 10 Robot 22 Signal output unit (control unit) 31 Illumination unit (first illumination unit) 31a Polarizing filter (first polarizing filter) 32 Imaging unit (first imaging unit) 32a Polarizing filter (second polarizing filter) 33 Imaging unit (second imaging unit) 34 Illumination unit (second illumination unit) 41 Inspection control unit (control unit) 43 Machine learning model 61 Silver streak (silver defect) 62 Shine 63 Scratch 64 Color unevenness 71 Inspection image (first inspection image) 72 Inspection image (second inspection image) 73 Heat map 100 Inspection system 200 Resin molded product

Claims

1. An inspection system comprising: a first illumination unit which has a first polarizing filter and irradiates illumination light onto a resin molded product; a first imaging unit which has a second polarizing filter which has a polarization axis angle shifted from that of the first polarizing filter and images the resin molded product; and a control unit which performs control to detect silver defects that occur as silver defects on the resin molded product due to abnormalities during resin molding, based on a first inspection image of the resin molded product captured using the first illumination unit and the first imaging unit.

2. The inspection system according to claim 1, wherein the control unit performs the following: control to detect the silver defect of the resin molded product based on the first inspection image; and control to detect a defect in the resin molded product other than the silver defect based on a second inspection image of the resin molded product captured in a polarization state different from that of the first inspection image.

3. The inspection system according to claim 2, wherein the control unit performs control to detect a defect in the resin molded product that is different from the silver defect, based on the second inspection image captured through one of the first polarizing filter and the second polarizing filter without passing through the other.

4. The inspection system according to claim 3, further comprising a second imaging unit for imaging the resin molded product, wherein the control unit performs control to detect defects in the resin molded product that are different from the silver defects, based on the second inspection image captured using the first illumination unit and the second imaging unit.

5. The inspection system according to claim 4, wherein the control unit controls the first imaging unit and the second imaging unit to simultaneously image the resin molded product.

6. The inspection system according to claim 3, further comprising a second illumination unit that irradiates illumination light onto the resin molded product, wherein the control unit performs control to detect defects in the resin molded product that are different from the silver defects, based on the second inspection image captured using the second illumination unit and the first imaging unit.

7. The inspection system according to claim 6, wherein the control unit controls the first imaging unit to image the resin molded product while alternately lighting the first illumination unit and the second illumination unit.

8. The inspection system according to claim 2, wherein the control unit performs control to detect at least one of gloss, scratches, and color unevenness as defects of the resin molded product that are different from the silver defects, based on the second inspection image.

9. The inspection system according to claim 1, wherein the control unit performs control to detect the silver defect in the resin molded product using a machine learning model that has been trained to detect the silver defect in the resin molded product with the first inspection image as input.

10. The inspection system according to claim 9, wherein the control unit uses the machine learning model to perform control to acquire a heat map that is colored to change according to the accuracy of the silver defect in the resin molded product.

11. The inspection system according to claim 1, wherein the first polarizing filter and the second polarizing filter have polarization axes offset by 90 degrees.

12. The inspection system according to claim 1, further comprising a robot for moving the first illumination unit and the first imaging unit relative to the resin molded product.

13. The inspection system according to claim 12, wherein the control unit generates a pulse signal based on the relative movement amount of the first illumination unit and the first imaging unit with respect to the resin molded product for each relative movement amount of the first illumination unit and the first imaging unit, and controls the imaging operation of the resin molded product by the first illumination unit and the first imaging unit based on the generated pulse signal.

14. An inspection method comprising: irradiating a resin molded product with illumination light using a first illumination unit on which a first polarizing filter is arranged; imaging the resin molded product using a first imaging unit on which a second polarizing filter, whose polarization axis angle is shifted from that of the first polarizing filter, is arranged; and detecting silver defects that occur as silver defects in the resin molded product due to abnormalities during resin molding, based on a first inspection image of the resin molded product captured using the first illumination unit and the first imaging unit.

Images