WO2021010181A1 - Inspection device, inspection method, positioning method, and program - Google Patents

Abstract

This inspection device 1 comprises: a robot arm 31 for gripping that grips an inspection target object 7; a camera 43; a measurement position calculation unit 142 that calculates a measurement position 10 for the camera 43 relative to the inspection target object 7, on the basis of shape data 8 for the inspection target object 7; a trajectory generation unit 143 that generates a trajectory for the robot arm 31 for gripping, on the basis of the measurement position 10; a robot control unit 13 that controls the action of the robot arm 31 for gripping, on the basis of the trajectory, and positions the inspection target object 7; a measurement unit 15 that, after positioning, uses the camera 43 and measures the inspection target object 7; and an inspection unit 16 that inspects the inspection target object 7 on the basis of the measured data.

Description

Inspection equipment, inspection methods, positioning methods, and programs

This disclosure relates to inspection equipment, inspection methods, positioning methods, and programs.

Conventionally, in product manufacturing lines, etc., automation of visual inspection such as scratches and chips on products has been promoted by an inspection device that combines a sensor such as a camera and a robot arm.

For example, Patent Document 1 discloses an inspection device including a robot arm provided with a plurality of cameras, and an inspection object imaged by a plurality of cameras is inspected based on the imaging data. By judging the quality, it is possible to inspect the product with high quality without variation.

Further, Patent Document 2 includes an inspection device composed of a robot arm equipped with a sensor unit, an inspection jig composed of a robot arm equipped with a holder for holding an inspection object, and a control unit. A visual inspection method for performing a visual inspection of an inspection object is disclosed by the visual inspection system. In particular, a visual inspection method has been proposed that combines the automatic extraction of appearance abnormalities of the inspection target by a computer and the quality judgment of the appearance abnormality portion by an operator.

International Publication No. 2015/011782 JP-A-2018-59830

However, in Patent Documents 1 and 2, since it is necessary to teach the operation of the robot arm for each shape of the inspection object and the inspection item by using a programming pendant or the like, it takes a lot of man-hours to prepare for the inspection. It was inefficient.

The present disclosure has been made in view of the above-mentioned problems, and an object of the present disclosure is to provide an inspection device or the like that does not require teaching work of a robot arm.

According to one embodiment of the present disclosure, the measurement position of the measurement sensor with respect to the inspection object is calculated based on the gripping robot arm for gripping the inspection object, the measurement sensor, and the shape data of the inspection object. The measurement position calculation unit, the trajectory generation unit that generates the trajectory of the gripping robot arm based on the measurement position, and the operation of the gripping robot arm based on the trajectory are controlled to position the inspection object. Provided is an inspection device including a robot control unit for performing the robot control unit, a measurement unit for measuring the inspection object by the measurement sensor after positioning, and an inspection unit for inspecting the inspection object based on the measured data. Will be done.

Further, according to one embodiment of the present disclosure, the computer calculates the measurement position of the measurement sensor with respect to the inspection object based on the shape data of the inspection object, and the inspection object is determined based on the measurement position. A step of generating a trajectory of the gripping robot arm to be gripped, a step of controlling the operation of the gripping robot arm based on the trajectory to position the inspection object, and after positioning, the inspection by the measurement sensor. An inspection method is provided for performing a step of measuring an object and a step of inspecting the inspection object based on the measured data.

Further, according to one embodiment of the present disclosure, the computer is a measurement position calculation unit that calculates the measurement position of the measurement sensor with respect to the inspection object based on the shape data of the inspection object, and the inspection object based on the measurement position. A trajectory generating unit that generates a trajectory of a gripping robot arm that grips an object, a robot control unit that controls the operation of the gripping robot arm based on the trajectory and positions the inspection object, and a measurement after positioning. A program is provided that functions as a measuring unit that measures the inspection object by a sensor and an inspection unit that inspects the inspection object based on the measured data.

Further, according to one embodiment of the present disclosure, a step of calculating a measurement position with respect to the inspection object based on the shape data of the inspection object and a grip for gripping the inspection object based on the measurement position. A positioning method for executing a step of generating a trajectory of the robot arm for inspection and a step of controlling the operation of the gripping robot arm based on the trajectory and positioning the object to be inspected are provided.

According to the present disclosure, an inspection device or the like that does not require teaching work of a robot arm is provided.

The figure which shows the example of the whole structure of the inspection apparatus 1. The figure which shows the example of the 2D optical system 42 The figure which shows the example of the 2D optical system 42 Block diagram showing an example of the hardware configuration of the control device 6 Block diagram showing an example of the functional configuration of the control device 6 Diagram showing the outline of the inspection process A flowchart showing the flow of processing for setting the inspection site 80 and the inspection specification 9. The figure which shows the example of the shape data 8. The figure which shows the example of setting the inspection site 80 The figure which shows the example of setting the inspection specification 9. Flow chart showing the flow of processing to calculate the measurement position 10 The figure which shows the example of the geometric shape 81 (polyhedron) The figure which shows the example of the geometric shape 81 (spherical grid) The figure which shows the example which sets the measurement candidate position 101 The figure which shows a mode that the measurable area P1 of each measurement candidate position 101 is integrated. The figure which shows the example of the measurement inspection table 100 Diagram showing the overall flow of inspection A flowchart showing the flow of the inspection operation executed by the inspection device 1. Flowchart showing the flow of measurement / inspection processing The figure which shows the extraction example of the defect candidate 71 Diagram showing a display example of defective parts The figure which shows the transition graph of the defective product occurrence rate The figure which shows the display example of the correlation analysis result which shows the correlation between the defect occurrence rate predicted from the manufacturing process parameter, and the actual defect occurrence rate. The figure which shows the example of the whole structure of the inspection apparatus 1A

Hereinafter, embodiments of the present disclosure (hereinafter referred to as “the present embodiment”) will be described in detail based on the drawings.

(1. Configuration of inspection device 1)
FIG. 1 is a diagram showing an overall configuration of the inspection device 1 of the present embodiment. The inspection device 1 is a device that inspects the appearance of the inspection object 7, and is mainly composed of a supply mechanism 2, a gripping mechanism 3, a measurement mechanism 4, a discharge mechanism 5, and a control device 6.

The supply mechanism 2 includes a conveyor 21 for supplying the inspection object 7 and a vision sensor 22 for recognizing the position and orientation of the inspection object 7. The vision sensor 22 is arranged above the conveyor 21 and recognizes the position and orientation of the inspection object 7 on the conveyor 21. The inspection object 7 is gripped by the gripping robot arm 31a or 31b based on the recognized position and posture of the inspection object 7.

Note that the supply mechanism 2 is not limited to the example shown in the figure. For example, the supply mechanism 2 may be composed of a container (not shown) in which inspection objects 7 are piled up and a vision sensor 22 arranged above the container. In this case, the position and orientation of the inspection object 7 are recognized by the vision sensor 22, and the inspection object 7 is gripped by the gripping robot arm 31a or 31b.

Further, the supply mechanism 2 may be composed of a tray (not shown) in which the inspection objects 7 are arranged in a fixed posture. In this case, since the position and posture of the inspection object 7 are known in advance, it is not necessary to provide the vision sensor 22. That is, the inspection object 7 is gripped by the gripping robot arm 31a or 31b without the vision sensor 22.

Further, the supply mechanism 2 may be composed of a parts feeder (not shown). Also in this case, since the inspection object 7 is aligned in a constant posture, the inspection object 7 is gripped by the gripping robot arm 31a or 31b without the vision sensor 22.

The gripping mechanism 3 is composed of articulated (for example, 6-axis) robot arms (grasping robot arms 31a and 31b) arranged in a double-armed manner on both sides of the support column 18. The gripping robot arms 31a and 31b include arm portions 311a and 311b and hand portions 312a and 312b, respectively. The hand portions 312a and 312b are provided at the tips of the arm portions 311a and 311b.

The hand portions 312a and 312b may have any form as long as they can grip the inspection object 7, but the structure in which more than half of the inspection object 7 is exposed while being gripped by the hand portions 312a and 312b. Is desirable. If the hand portions 312a and 312b have such a structure, the entire surface of the inspection object 7 can be measured by one holding operation between the gripping robot arms 31a and 31b. As the hand portions 312a and 312b, a holding type, a suction type, or the like can be adopted.

In the present disclosure, the gripping robot arm 31a or 31b may be referred to as a gripping robot arm 31. In this case, the arm portion of the gripping robot arm 31 is referred to as the arm portion 311 and the hand portion is referred to as the hand portion 312.

The measurement mechanism 4 includes an articulated (for example, 6-axis) robot arm (measurement robot arm 41) arranged above the support column 18, a two-dimensional optical system (hereinafter referred to as “2D optical system 42”), and It is composed of a three-dimensional sensor (hereinafter referred to as "3D sensor 45"). The 2D optical system 42 and the 3D sensor 45 are fixed to, for example, the tip of the measuring robot arm 41 directly or via another member.

2 and 3 are diagrams showing the 2D optical system 42. FIG. 2 is a perspective view of the 2D optical system 42, and FIG. 3 is an XZ cross-sectional view at the center of the 2D optical system 42. As shown in FIGS. 2 and 3, the 2D optical system 42 includes a camera 43 including a main body 43a and a lens 43b, and illuminations 44 arranged in a plurality of hemispheres. The camera 43 is arranged above the apex of the hemisphere and images the inspection object 7.

The camera 43 (main body 43a) has, for example, a CCD (Charge Coupled).
Device) Image sensor and CMOS (Complementary Metal Oxide)
Semiconductor) An image sensor such as an image sensor is installed. The illumination 44 is, for example, a white LED. When the inspection object 7 is imaged, the lighting of the illumination 44 is switched, and two-dimensional images having different illumination angles can be acquired. A plurality of cameras 43 may be arranged.

The 3D sensor 45 is a sensor that measures the three-dimensional shape of the inspection object 7. The 3D sensor 45 can be used in any area such as a twin-lens or multi-lens stereo camera, an active stereo camera equipped with a light projecting unit such as a laser or a projector, and a device using the time-of-flight method. Anything that can measure three-dimensional data will do.
The camera 43 and the 3D sensor 45 are examples of the measurement sensors in the present disclosure.

The discharge mechanism 5 is composed of a conveyor 51 for discharging the inspection object 7. A plurality of conveyors 51 may be provided so that the inspection object 7 can be classified according to the pass / fail of the inspection and the type of defect.

The control device 6 is a computer that controls the operations of the supply mechanism 2, the gripping mechanism 3, and the measuring mechanism 4.
Hereinafter, the hardware configuration (FIG. 4) and the functional configuration (FIG. 5) of the control device 6 will be described.

(2. Hardware configuration of control device 6)
FIG. 4 is a block diagram showing a hardware configuration of the control device 6. As shown in the figure, in the control device 6, the control unit 61, the storage unit 62, the communication unit 63, the input unit 64, the monitor 65, the peripheral device I / F unit 66, the UPS 67, and the like are connected via the bus 69. It is realized by a general-purpose computer. However, the present invention is not limited to this, and various configurations can be adopted depending on the application and purpose.

The control unit 61 is composed of a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), and the like. The CPU calls a program stored in the storage unit 62, ROM, recording medium, etc. into a work memory area on the RAM and executes the program, drives and controls each device connected via the bus 69, and the control device 6 performs the program. Realize the processing described later.

ROM is a non-volatile memory and permanently holds computer boot programs, programs such as BIOS, and data. The RAM is a volatile memory, and includes a work area used by the control unit 61 to perform various processes while temporarily holding programs, data, and the like loaded from the storage unit 62, the ROM, the recording medium, and the like.

The storage unit 62 is an HDD (Hard Disk Drive) or the like, and stores a program executed by the control unit 61, data necessary for program execution, an OS (Operating System), and the like. As for the program, a control program corresponding to the OS and an application program for causing a computer to execute a process described later are stored. Each of these program codes is read by the control unit 61 as necessary, transferred to the RAM, and read by the CPU to execute various processes.

In the present embodiment, the shape data 8 (for example, CAD data) of the inspection object 7 is stored in the storage unit 62. Further, a ROS (Robot Operation System) is constructed in the storage unit 62, and it is possible to automatically generate the trajectories of the gripping robot arm 31 and the measuring robot arm 41 by using the ROS simulator function. It has become.

Further, an optical system simulator is constructed in the storage unit 62, and the measurable area P (imaging range) in the shape data 8 is calculated by inputting the shape data 8 of the inspection object 7 and the conditions of the optical system. It is possible to do. Further, the storage unit 62 stores in advance the learned deep learning device 72 used at the time of inspection.

The communication unit 63 has a communication control device, a communication port, etc., and is a communication interface that mediates communication between a computer and a network, and controls communication between other computers via the network. The network can be wired or wireless.

The input unit 64 inputs data and has, for example, a pointing device such as a keyboard and a mouse, and an input device such as a numeric keypad. Operation instructions, operation instructions, data input, and the like can be given to the computer via the input unit 64. The monitor 65 has a display device such as a liquid crystal panel, a logic circuit (video adapter or the like) for realizing a video function of a computer in cooperation with the display device. The input unit 64 and the monitor 65 may be integrated as in the touch panel display.

The peripheral device I / F (Interface) unit 66 is a port for connecting the peripheral device to the computer, and the computer transmits / receives data to / from the peripheral device via the peripheral device I / F unit 66. The peripheral device I / F unit 66 is composed of USB (Universal Serial Bus), LAN, IEEE1394, RS-232C, etc., and usually has a plurality of peripheral device I / Fs. The connection form with peripheral devices may be wired or wireless. The control device 6 includes a conveyor 21, a conveyor 51, a vision sensor 22, a gripping robot arm 31, a measuring robot arm 41, a 2D optical system 42, a 3D sensor 45, etc. via the peripheral device I / F (Interface) unit 66. Is connected with.

The UPS 67 is an uninterruptible power supply that continues to supply power even when the power is cut off due to a power failure or the like.
The bus 69 is a route that mediates the transfer of control signals, data signals, and the like between the devices.

The control device 6 may be configured by one computer, or may be configured so that a plurality of computers cooperate to execute the operation of the inspection device 1. In the following description, as a simple configuration example, an example in which the control device 6 is configured by one computer will be described.

(3. Functional configuration of control device 6)
FIG. 5 is a block diagram showing a functional configuration of the control device 6. As shown in the figure, the control device 6 is composed of the functions of the conveyor control unit 11, the work recognition unit 12, the robot control unit 13, the operation setting unit 14, the measurement unit 15, the inspection unit 16, and the data display unit 17. To.

The conveyor control unit 11 sends a control command to the drive units of the conveyors 21 and 51 to control the operation of the conveyors 21 and 51. Specifically, the conveyor control unit 11 operates the conveyor 21 on which the uninspected object 7 is placed in the supply direction (X direction in FIG. 1), and moves the inspection object 7 to the lower part of the vision sensor 22. Supply. After the inspection is completed, the conveyor control unit 11 operates the conveyor 51 on which the inspected object 7 is placed in the discharge direction (X direction in FIG. 1) to discharge the inspection object 7.

The work recognition unit 12 recognizes the position and orientation of the inspection object 7 supplied to the lower part of the vision sensor 22 by the vision sensor 22. The operation of the gripping robot arm 31 is controlled based on the recognized position and posture of the inspection object 7, and the inspection object 7 is gripped by the hand unit 312.

The robot control unit 13 sends a control command to the drive unit of each joint of the gripping robot arm 31 and the measuring robot arm 41, and controls the operation of the gripping robot arm 31 and the measuring robot arm 41.

Specifically, the robot control unit 13 moves the hand unit 312 of the gripping robot arm 31 to the picking position on the conveyor 21 based on the orbit generated by the orbit generation unit 143, which will be described later, and the inspection object. It is controlled so that the hand portion 312 grips the predetermined position of 7. As a result, the inspection object 7 on the conveyor 21 is picked.

Further, the robot control unit 13 controls the trajectories of the gripping robot arm 31 and the measurement robot arm 41 based on the trajectories generated by the orbit generation unit 143, which will be described later, after gripping the inspection object 7 (after picking). Then, the inspection object 7 and the camera 43 are positioned. After positioning the inspection target 7 and the camera 43, the inspection target 7 is measured.

Further, the robot control unit 13 controls the inspection object 7 so as to be switched between the hand unit 312a of the gripping robot arm 31a and the hand unit 312b of the gripping robot arm 31b. For example, a rule for gripping the inspection object 7 by the gripping robot arm 31a or 31b is set in advance for each measurement position 10, and the control unit 61 performs a holding operation based on this rule.

Further, after the inspection of the inspection object 7 is completed, the robot control unit 13 moves the hand unit 312 of the gripping robot arm 31 to the release position on the conveyor 51 based on the trajectory generated by the trajectory generation unit 143 described later. At the same time, the gripping state of the inspection object 7 by the hand portion 312 is released. As a result, the inspected object 7 that has been inspected is released on the conveyor 51.

The motion setting unit 14 makes settings necessary for motion control of the gripping robot arm 31 and the measurement robot arm 41. As shown in FIG. 5, the operation setting unit 14 includes an inspection site setting unit 140, an inspection specification setting unit 141, and a measurement position calculation unit 142 (geometry generation unit 142a, measurement candidate position setting unit 142b, measurable area calculation unit 142c, It is composed of a measurement position selection unit 142d) and an orbit generation unit 143.

The inspection site setting unit 140 reads the shape data 8 (for example, CAD data) of the inspection object 7 from the storage unit 62, and sets the inspection site 80 to be inspected for the shape data 8.

The inspection specification setting unit 141 sets the inspection specification 9 for each inspection site 80 set by the inspection site setting unit 140. The inspection specification 9 includes information on the defect type 91 and the defect specification 92. The defect type 91 indicates the type of defect to be inspected, such as a "convex" defect, a "concave" defect, and a burr. The defect specification 92 is a limit standard for each defect, and is used as a standard for determining the harmfulness / harmlessness of each defect.

Specific processing of the inspection site setting unit 140 and the inspection specification setting unit 141 will be described later (see FIGS. 7 to 10).

The measurement position calculation unit 142 calculates the measurement position 10 of the inspection object 7 based on the shape data 8. The measurement position 10 is the relative position of the camera 43 with respect to the shape data 8 (inspection object 7), and is, for example, the position coordinates of the camera 43 with the center of the shape data 8 (inspection object 7) as the origin. .. The measurement position calculation unit 142 calculates a plurality of measurement positions 10 around the shape data 8 so that all the inspection sites 80 set in the shape data 8 are measured (imaged) by the camera 43. The trajectories of the gripping robot arm 31 and the measuring robot arm 41 are automatically generated based on the measurement position 10.

As shown in FIG. 5, the measurement position calculation unit 142 includes a geometry generation unit 142a, a measurement candidate position setting unit 142b, a measurable area calculation unit 142c, and a measurement position selection unit 142d.

The geometry generation unit 142a generates a geometry 81 that includes the shape data 8 and can define each direction of the shape data 8.

The measurement candidate position setting unit 142b sets one or more measurement candidate positions 101 in each direction defined by the geometric shape 81.

The measurable area calculation unit 142c calculates the measurable area P in the shape data 8 by the optical system simulator for each measurement candidate position 101 set by the measurement candidate position setting unit 142b.

The measurement position selection unit 142d selects the measurement position 10 from the measurement candidate positions 101 based on the measurable area P of each measurement candidate position 101. For example, the measurement position selection unit 142d selects the measurement position 10 so that all the inspection sites 80 (inspection target areas) are measured and the number of measurements is minimized.

The specific processing of the measurement position calculation unit 142 (geometry generation unit 142a, measurement candidate position setting unit 142b, measurable area calculation unit 142c, measurement position selection unit 142d) will be described later (see FIGS. 11 to 15).

The orbit generation unit 143 builds a CAD model of the inspection device 1 and peripheral equipment on the ROS, and then uses the PRM (Probabilistic Roadmap Planner) method and RRT (Rapidly).
A path planning method (Motion Planning) such as the exploring Random Tree method is used to generate the trajectories of the gripping robot arm 31 and the measuring robot arm 41 avoiding interference.

Specifically, the trajectory generation unit 143 calculates the picking position of the inspection object 7 (target position and target posture of the hand unit 312) based on the position and orientation of the inspection object 7 recognized by the work recognition unit 12. Then, the trajectory of the gripping robot arm 31 is generated based on the picking position. By controlling the gripping robot arm 31 by this trajectory, the inspection object 7 on the conveyor 21 is gripped (picked) by the hand portion 312 of the gripping robot arm 31.

Further, the trajectory generation unit 143 generates the orbits of the gripping robot arm 31 and the measurement robot arm 41 based on the measurement position 10 calculated by the measurement position calculation unit 142. Specifically, the trajectory generation unit 143 has a gripping robot arm 31 and a measurement so that the relative position of the camera 43 with respect to the inspection object 7 is the measurement position 10 calculated by the measurement position calculation unit 142. Generates the trajectory of the robot arm 41. By controlling the gripping robot arm 31 and the measuring robot arm 41 by this trajectory, the inspection object 7 and the camera 43 are positioned at the measurement points. As a result, the inspection object 7 can be measured and inspected without teaching by a programming pendant or the like, and a highly efficient inspection is realized.

Further, after the inspection is completed, the trajectory generating unit 143 generates the trajectory of the gripping robot arm 31 based on the release position of the inspection object 7 (the target position and the target posture of the hand unit 312). By controlling the gripping robot arm 31 by this trajectory, the inspected object 7 that has been inspected is released on the conveyor 51.

After positioning the inspection target 7 and the camera 43, the measurement unit 15 measures the inspection target 7.
As shown in FIG. 5, the measuring unit 15 includes a first measuring unit 151 and a second measuring unit 152.

The first measurement unit 151 sends a measurement instruction to the 2D optical system 42 and measures the two-dimensional image data D1 of the inspection object 7.

The second measurement unit 152 sends a measurement instruction to the 3D sensor 45, and measures (shape measurement) the three-dimensional shape data D2 of the inspection object 7.

The inspection unit 16 inspects the inspection object 7 based on the measurement data measured by the measurement unit 15 (151, 152).
As shown in FIG. 5, the inspection unit 16 is composed of a first inspection unit 161 and a second inspection unit 162.

The first inspection unit 161 extracts the defect candidate 71 based on the two-dimensional image data D1 of the inspection object 7 measured by the first measurement unit 151.

In the present embodiment, the first inspection unit 161 is a type of machine learning, deep learning (Deep).
The defect candidate 71 is extracted by using the deep learner 72 trained by Learning). The deep learning device 72 is a discriminator that has been trained based on the quality image data (learning data) with / without defects prepared for each type of defect, and is stored in the storage unit 62 in advance. As the deep learning method, SegNet, ResNet, or a method in which SegNet and ResNet are used in combination can be used, but the method is not limited to these methods.

The second inspection unit 162 performs a detailed shape inspection of the defect candidate 71 extracted by the first inspection unit 161 based on the three-dimensional shape data D2 of the inspection object 7 measured by the second measurement unit 152. Dimensional inspection) to determine the harmful / harmlessness of the final defect.

FIG. 6 is a diagram showing an outline of the inspection process. As shown in the figure, first, the defect candidate 71 is extracted from the two-dimensional image data D1 using the deep learning device 72 (first inspection), and when the defect candidate 71 is not extracted, the inspection target is ". Judged as "good" (no defects). On the other hand, when the defect candidate 71 is extracted, the shape inspection (dimension inspection) of the defect candidate 71 is further performed by the three-dimensional shape data D2 (second inspection), and the defect specification 92 of the defect candidate 71 is obtained. By collating, the final pass / fail judgment is made.

(4. Setting of inspection site and inspection specifications)
With reference to FIGS. 7 to 10, a process of setting an inspection site 80 to be inspected and an inspection specification 9 (defect type 91, defect specification 92) for the shape data 8 of the inspection object 7 prior to the inspection. Will be explained.

FIG. 7 is a flowchart showing a flow of processing for setting the inspection site 80 and the inspection specification 9.
First, the control unit 61 of the control device 6 reads the shape data 8 of the inspection object 7 from the storage unit 62 and displays it on the monitor 65 (step S1).
FIG. 8 shows an example of shape data 8 of the inspection object 7 (joint).

Subsequently, the control unit 61 (inspection site setting unit 140) sets the inspection site 80 to be inspected with respect to the shape data 8 displayed on the monitor 65 (step S2). This setting is performed by a user operation via the input unit 64. For example, as in the example of FIG. 9, the user can use the shape data 8 (CAD data of the pipe joint) as the inspection part 80 as the inspection part 80a (outer surface of the joint body) and the inspection part 80b (joint end). The outer surface of) is set.

Subsequently, the control unit 61 (inspection specification setting unit 141) accepts the setting of the inspection specification 9 (defect type 91, defect specification 92) for each inspection site 80 set in step S2 (step S3). For example, as shown in FIG. 10, as the inspection specification 9a (defect type 91, defect specification 92) of the inspection part 80a (outer surface of the joint main body), the inspection specification 9a-1 (“convex”, “Φ3.0 mm, height” 2.0 mm ") and inspection specifications 9a-2 ("concave "," Φ3.0 mm, depth 1.0 mm ") are set.

In addition, as inspection specifications 9b (defect type 91, defect specifications 92) of the inspection site 80b (outer surface of the joint end), inspection specifications 9b-1 (“convex”, “Φ3.0 mm, height 2.0 mm”), Inspection specifications 9b-2 (“concave”, “Φ3.0 mm, depth 1.0 mm”) and inspection specifications 9b-3 (“burrs”, “0.5H × 2L (mm)”) are set.

As will be described later, the control device 6 selects (calculates) the measurement position 10 so that all the inspection sites 80 set in step S2 are measured. Further, the control device 6 determines the quality of the inspection object 7 based on the inspection specifications 9 (defect type 91, defect specifications 92) of each inspection site 80 set in step S3.

(5. Calculation of measurement position)
The process of calculating the measurement position 10 of the inspection object 7 will be described with reference to FIGS. 11 to 15. Based on the measurement position 10 calculated here, the trajectories of the gripping robot arm 31 and the measurement robot arm 41 are generated.

FIG. 11 is a flowchart showing a flow of processing for calculating the measurement position 10.
First, the control unit 61 (geometry generation unit 142a) of the control device 6 generates the geometry 81 that includes the shape data 8 of the inspection object 7 and can define each direction of the shape data 8 (step S11). ). For example, a polyhedron 81 including the shape data 8 as shown in FIG. 12 and a spherical grid 81 including the shape data 8 as shown in FIG. 13 are generated.

In the case of a polyhedron (FIG. 12), each direction of the shape data 8 is defined by the direction of each surface of the polyhedron (normal direction of the surface). In the case of the spherical grid (FIG. 13), each direction of the shape data 8 is defined by the direction of the intersection of the grid lines. In the present embodiment, the control unit 61 (geometric generation unit 142a) will be described as generating the polyhedron 81 shown in FIG. The polyhedron 81 is, for example, a regular dodecahedron or a regular icosahedron, but is not limited thereto.

Subsequently, the control unit 61 (measurement candidate position setting unit 142b) sets one or more measurement candidate positions 101 in each direction defined by the geometric shape 81 (step S12). In the present embodiment, the control unit 61 sets one or more measurement candidate positions 101 (position coordinates of the camera 43) on the surface normal 84 passing through the center 83 of each surface 82 of the polyhedron 81 (geometric shape 81). The imaging direction of each camera 43 installed at the measurement candidate position 101 is a direction perpendicular to the surface 82. That is, the image pickup center of each camera 43 is the center 83 of each surface 82.

FIG. 14 shows an example in which the measurement candidate position 101 is set on the surface normal 84 passing through the center 83 of the surface 82 having the polyhedron 81 (regular icosahedron). In the example of the figure, three measurement candidate positions 101 are set on the surface normal 84. The measurement candidate position 101 is appropriately set based on optical conditions such as the field of view range of the camera 43, the depth of the subject, and the working distance.

The control unit 61 (measurement candidate position setting unit 142b) sets the measurement candidate position 101 for all the surfaces 82 of the polyhedron 81. For example, when three measurement candidate positions 101 are set for each surface 82 of the regular icosahedron of FIG. 14, a total of 60 measurement candidate positions 101 are set.

Subsequently, the control unit 61 (measurable area calculation unit 142c) calculates the measurable area P in the shape data 8 for each measurement candidate position 101 set in step S12 by using the optical system simulator (step S13). ).

Then, the control unit 61 (measurement position selection unit 142d) measures all the inspection sites 80 (inspection target areas) based on the measurable region P of each measurement candidate position 101 calculated in step S13, and The measurement position 10 that minimizes the number of measurements is selected from the measurement candidate positions 101 (step S14).

For example, the control unit 61 (measurement position selection unit 142d) selects measurement candidate positions 101 one by one in a random order from all the measurement candidate positions 101, and measures the selected measurement candidate positions 101. The possible area P will be integrated. The region in which the measurable region P is integrated is called the integrated region R. Then, the control unit 61 discontinues the selection of the measurement candidate position 101 at the stage where the integrated region R covers (includes) all the inspection sites (inspection target regions), and measures the measurement candidate positions 101 selected so far. Select as 10.

A specific description will be given with reference to FIG. For example, as shown in FIG. 15, when the measurement candidate positions 101a, 101b, and 101c are randomly selected in this order, the integrated region R is sequentially generated as R1, R2, and R3. Here, the integrated region R1 is equivalent to the measurable region P1 of the measurement candidate position 101a, the integrated region R2 is the region in which the measurable region P2 of the measurement candidate position 101b is integrated with the integrated region R1, and R3 is the integrated region R2. This is a region in which the measurable region P3 of the measurement candidate position 101c is integrated.

Subsequently, similarly, the selection of the measurement candidate position 101 and the integration of the measurable area P (expansion of the integrated area R) are sequentially repeated, and the integrated area R covers (includes) all the inspection sites 80 (inspection target areas). The selection of the measurement candidate position 101 is terminated at the stage, and the measurement candidate position 101 selected so far is selected as the measurement position 10. If the entire area of the measurable area P of the selected measurement candidate position 101 is included in the integrated area R generated up to that point, another measurement candidate position 101 is reselected. That is, the measurement position 10 is selected so that the already measured regions are not measured in duplicate.

If the integrated region R, which integrates the measurable regions P of all the measurement candidate positions 101, cannot cover all the inspection sites (inspection target regions), the control unit 61 increases the number of surfaces. The polyhedron 81 (polyhedron 81 with increased directional resolution) is regenerated, and the processes of steps S11 to S14 are executed again.

FIG. 16 shows a measurement inspection table 100 that holds various inspection information and the like in association with the measurement position 10 selected in step S14. In the measurement inspection table 100, the inspection site 80, the defect type 91, and the defect specification 92 at each measurement position 10 are linked to each measurement position 10 (measurement positions 10-1, 10-2, ...) Arranged in the measurement order. , Lighting condition 93, first measurement data 94, second measurement data 95, and inspection result 96 information are retained.

The inspection site 80 is an inspection site 80 (inspection site that can be measured / inspected at each measurement position 10) included in the measurable region P of each measurement position 10 among the inspection sites 80 set in step S1 of FIG. .. The defect type 91 and the defect specification 92 are the defect type 91 and the defect specification 92 associated with the inspection site 80 set in step S2 of FIG. 7.

The lighting condition 93 is a condition for which lighting 44 is to be turned on at the time of measurement, and is predetermined according to, for example, the type of defect type 91 (“convex”, “concave”, “burr” ...).

In the first measurement data 94 and the second measurement data 95, the storage destination information of the measurement data is recorded after the measurement is executed. Information on the inspection result is recorded in the inspection result 96 after the inspection is executed.

(6. Operation of inspection device 1)
The operation of the inspection device 1 will be described with reference to FIGS. 17 to 23.

FIG. 17 shows an outline of the operation of the inspection device 1. First, as shown in FIG. 17A, the inspection device 1 controls the trajectory of the gripping robot arm 31 to grip (pick) the inspection object 7 on the conveyor 21. Subsequently, as shown in FIG. 17B, the inspection device 1 controls the trajectories of the gripping robot arm 31 and the measurement robot arm 41, and moves the inspection object 7 to the measurement point. Here, in order to measure all the inspection sites of the inspection object 7, the inspection device 1 performs the measurement a plurality of times while changing the positional relationship between the inspection object 7 and the camera 43. Further, as shown in FIG. 17C, the inspection device 1 switches the inspection target 7 between the gripping robot arms 31a and 31b as necessary. After the inspection is completed, as shown in FIG. 17D, the inspection device 1 controls the trajectory of the gripping robot arm 31 and releases the inspection object 7 on the conveyor 51.

The details of the operation of the inspection device 1 (control device 6) will be described with reference to the flowchart of FIG.

First, the control unit 61 (conveyor control unit 11) of the control device 6 controls the operation of the conveyor 21 on which the uninspected object 7 is placed, and supplies the inspection object 7 to the lower part of the vision sensor 22 ( Step S21).

Subsequently, the control unit 61 (work recognition unit 12) recognizes the position and orientation of the inspection object 7 supplied in step S21 by the vision sensor 22 (step S22).

Subsequently, the control unit 61 (robot control unit 13) moves the gripping robot arm 31 (31a or 31b) based on the position and posture of the inspection object 7 recognized in step S12, and the hand unit 312 inspects it. The object 7 is controlled to be gripped (step S23).

Specifically, first, the control unit 61 (orbit generation unit 143) has the target position and the target position of the hand unit 312 capable of gripping the inspection object 7 based on the position and posture of the inspection object 7 recognized in step S22. The target posture is calculated, and the trajectory of the gripping robot arm 31 is generated based on the calculated target position and target posture of the hand unit 312.

Then, the control unit 61 (robot control unit 13) controls the trajectory of the gripping robot arm 31 based on the generated trajectory, and causes the hand unit 312 to grip the predetermined position of the inspection object 7. To control. As a result, the inspection object 7 is picked by the gripping robot arm 31.

Subsequently, the control unit 61 (robot control unit 13) controls the trajectories of the gripping robot arm 31 and the measurement robot arm 41, and positions the inspection object 7 and the camera 43 at the measurement point (step S24).

Specifically, first, in the control unit 61 (orbit generation unit 143), the relative position of the camera 43 with respect to the inspection object 7 is the measurement position 10 (in the first measurement) of the measurement inspection table 100 (see FIG. 16). The trajectory of the gripping robot arm 31 and the measurement robot arm 41 is calculated so as to be at the measurement position 10-1), and a trajectory is generated.

Here, since the measurement position 10 is only the position relative to the inspection object 7 of the camera 43, the trajectories of the gripping robot arm 31 and the measurement robot arm 41 cannot be uniquely determined. Therefore, the control unit 61 (orbit generation unit 143) selects the orbits of the gripping robot arm 31 and the measurement robot arm 41 such that the relative position of the camera 43 with respect to the inspection object 7 is the measurement position 10. A trajectory that satisfies a predetermined condition is uniquely determined. The orbits satisfying the predetermined conditions are, for example, the orbits in which the average movement amount of the gripping robot arm 31 and the measurement robot arm 41 is minimized, and the average movement time of the gripping robot arm 31 and the measurement robot arm 41 is minimum. Orbit, etc.

Then, the control unit 61 (robot control unit 13) controls the trajectories of the gripping robot arm 31 and the measurement robot arm 41 based on the generated trajectories, and positions the inspection object 7 and the camera 43.

Subsequently, the control unit 61 (measurement unit 15, inspection unit 16) executes measurement / inspection of the inspection object 7 (step S25).

Here, the measurement / inspection process will be described with reference to the flowchart of FIG.

First, the control unit 61 (first measurement unit 151) sends a measurement instruction to the 2D optical system 42 and measures the two-dimensional image data D1 of the inspection object 7 (step S41). Specifically, the control unit 61 refers to the measurement inspection table 100 and measures the inspection site 80 that can be inspected at the corresponding measurement position 10 for each defect type 91. At this time, the control unit 61 may switch the illumination condition 93 according to each defect type 91.

For example, referring to FIG. 16, the inspection site 80 at the measurement position 10-1 is “site A”, and the defect type 91 to be inspected is a “convex” defect and a “concave” defect. The lighting condition 93 suitable for inspecting "convex" defects is "condition 1", and the lighting condition 93 suitable for inspecting "concave" defects is "condition 2". Therefore, when the control unit 61 (first measurement unit 151) measures the “convex” defect in the “site A”, the control unit 61 turns on the illumination 44 based on the illumination condition 93 “condition 1” and then measures with the camera 43. I do. Further, when measuring the "concave" defect in the "part A", the illumination 44 is turned on based on the illumination condition 93 "condition 2", and then the measurement is performed by the camera 43.

Regardless of the defect type 91, the control unit 61 (first measurement unit) may turn on the determined lighting 44 (for example, turn on all the lighting 44) to perform the measurement, or the lighting 44 may be turned on. The measurement may be performed while switching the above one by one. In the latter case, two-dimensional images for the number of illuminations having different illumination angles can be obtained.

The control unit 61 (first measurement unit 151) stores the two-dimensional image data D1 (measurement data) of the measured inspection object 7 in the storage unit 62, and stores it in the first measurement data 94 of the measurement inspection table 100. Record information.

Subsequently, the control unit 61 (first inspection unit 161) inputs the two-dimensional image data D1 of the inspection object 7 measured in step S41 into the deep learning device 72, and inputs the defect candidate 71 from the two-dimensional image data D1. Extract (step S42).

FIG. 20 shows an extraction example of defect candidate 71. FIG. 20A is an image taken by the camera 43 of the inspection object 7, and FIG. 20B is an image in which the degree of defects obtained by deep learning is drawn with a contour map (contour map) (FIG. 20A). It is an image in which a contour map is superimposed on the photographed image of. As shown in FIG. 20B, for example, a portion where the degree of defect exceeds a predetermined threshold value is extracted as defect candidate 71 (71a, 71b, 71c).

If the defect candidate 71 is not extracted in step S42 (step S43; No), the control unit 61 (first inspection unit 161) determines that the inspection object 7 is “good” (no defect) and inspects it. The result is recorded in the inspection result column 96 of the measurement inspection table 100 (step S46).

When the defect candidate 71 is extracted in step S42 (step S43; Yes), the control unit 61 (second measurement unit 152) sends a measurement instruction to the 3D sensor 45 to provide three-dimensional shape data of the inspection object 7. D2 is measured (step S44), the measured three-dimensional shape data D2 is stored in the storage unit 62, and the storage destination information is recorded in the second measurement data 95 of the measurement inspection table 100. When performing measurement by the 3D sensor 45, the control unit 61 may control the gripping robot arm 31 or the measurement robot arm 41 to adjust the measurement position of the 3D sensor 45 with respect to the inspection object 7.

Subsequently, the control unit 61 (second inspection unit 162) inspects the shape of the defect candidate 71 extracted in step S42 based on the three-dimensional shape data D2 of the inspection object 7 measured in step S44. Dimension inspection) is performed to judge the quality.

For example, it is assumed that the defect candidate 71 of the "convex" defect is extracted from the two-dimensional image data D1 at the measurement position 10-1 in FIG. In this case, the control unit 61 (second inspection unit 162) recognizes the shape of the defect candidate 71 (“convex” defect) from the three-dimensional shape data D2, and the defect specification 92 “φ3.0 mm, height” of the measurement inspection table 100. Shape inspection (dimension inspection) is performed by collating with "2.0 mm". For example, when the control unit 61 (second inspection unit 162) recognizes that the φ (diameter) of the defect candidate 71 is 3.0 mm or more and the height is 2.0 mm or more, it is “No” (“convex”). "There is a defect"), and in other cases, it is judged as "Good" (no "convex" defect).

Then, the control unit 61 (second inspection unit 162) records the inspection result in the inspection result column 96 of the measurement inspection table 100 (step S46).

Returning to the description of the flowchart of FIG.
When all the inspections have not been completed (step S26; No), the control unit 61 needs to switch between the gripping robot arms 31a and 31b of the inspection object 7 in order to perform the next measurement / inspection. (Step S27).

For example, for each measurement position 10, a rule for gripping the inspection object 7 by the gripping robot arm 31a or 31b is set in advance, and the control unit 61 determines whether or not it is necessary to change hands based on this rule. To judge.

When it is not necessary to change hands, the control unit 61 returns to step S24 and moves the gripping robot arm 31 and the measurement robot arm 41 to the next measurement point.

On the other hand, when it is necessary to change hands (step S27; Yes), the control unit 61 (robot control unit 13) inspects the inspection object 7 with the hand unit 312a of the gripping robot arm 31a and the hand unit 312b of the gripping robot arm 31b. It is controlled so as to switch between and (step S28).

For example, suppose that the inspection object 7 is changed from the hand portion 312a of the gripping robot arm 31a to the hand portion 312b of the gripping robot arm 31b. In this case, the control unit 61 first sets the gripping robot arm 31a so that the hand portion 312a of the gripping robot arm 31a and the hand portion 312b of the gripping robot arm 31b have predetermined positions and postures at the time of changing hands. The trajectory of the gripping robot arm 31b is controlled, and the hand portion 312a and the hand portion 312b are positioned.

Subsequently, the control unit 61 causes the hand unit 312b to grip the predetermined position of the inspection object 7, and after the hand unit 312b grips the object 7, releases the grip of the hand unit 312a. As a result, the inspection target 7 is switched between the hand portions 312a and 312b.

Then, after the inspection target 7 is changed, the process returns to step S24, and the control unit 61 (robot control unit 13) moves the gripping robot arm 31 and the measurement robot arm 41 after the change to the next measurement point, and the inspection target. Position the object 7 and the camera 43.

The processes of steps 24 to S28 described above are repeatedly executed until all the inspections are completed (step S26; Yes).

After the inspection of the inspection object 7 is completed, the control unit 61 (robot control unit 13) moves the gripping robot arm 31 holding the inspection object 7 to the grip release position on the conveyor 51, and the hand unit 312. Is controlled to release the gripping state of the inspection object 7 (step S29). As a result, the inspected object 7 that has been inspected is released on the conveyor 51. Then, the control unit 61 (conveyor control unit 11) controls the operation of the conveyor 51 on which the inspected object 7 is placed, and discharges the inspection object 7.

Finally, the control unit 61 (data display unit 17) displays information on the inspection result of the inspection object 7 on the monitor 67. For example, as shown in FIG. 21, the control unit 61 displays a captured image in which the defective portion is clearly shown. Further, as shown in FIG. 22, the control unit 61 displays a daily transition graph of the defective product occurrence rate, and as shown in FIG. 23, the defective occurrence rate predicted from the manufacturing process parameters and the actual defective occurrence rate. Correlation analysis results showing the correlation with the rate may be displayed.

The present embodiment has been described above with reference to the drawings. The inspection device 1 of the present embodiment includes a gripping robot arm 31 for gripping the inspection object 7, a measurement robot arm 41 provided with a camera 43, and a control device 6. In particular, the control device 6 calculates the measurement position 10 of the camera 43 with respect to the inspection object 7 based on the shape data 8 of the inspection object 7, and the gripping robot arm 31 and the measurement robot arm based on the measurement position 10. Generates 41 orbits. Then, the control device 6 controls the operations of the gripping robot arm 31 and the measuring robot arm 41 based on the generated trajectory, and positions the inspection object 7 and the camera 43 at the measurement points.

In this way, by automatically generating the trajectories of the gripping robot arm 31 and the measuring robot arm based on the measurement position 10 calculated from the shape data 8 (CAD data or the like) of the inspection object 7, as in the conventional case. In addition, it is possible to measure and inspect the inspection object 7 without teaching by a programming pendant or the like, and it is possible to realize a highly efficient inspection.

Further, according to the present embodiment, the measurement position 10 is calculated so that all the inspection sites 80 are measured. Specifically, a polyhedron 81 including the shape data 8 is generated, one or more measurement candidate positions 101 are set on the surface normal line 84 passing through the center 83 of each surface 82 of the polyhedron 81, and each measurement candidate position 101 is set. Based on the measurable region P of the above, the measurement position 10 in which all the inspection sites 80 (inspection target regions) are measured and the number of measurements is the minimum is selected. As a result, the optimum measurement position 10 for measuring all the inspection sites 80 is automatically determined.

Further, according to the present embodiment, each inspection site 80 of the shape data 8 is associated with an inspection specification 9 such as a defect type 91 and a defect specification 92 to be inspected. As a result, the quality judgment of the inspection object 7 can be efficiently performed based on the inspection specification 9.

Further, according to the present embodiment, the measurement / inspection of the inspection object 7 is executed in two stages. That is, the defect candidate 71 is extracted from the two-dimensional image data D1 using the deep learning device 72 (first inspection), and if the defect candidate 71 is not extracted, the inspection target is “good” (no defect). Is determined. On the other hand, when the defect candidate 71 is extracted, the shape inspection (dimension inspection) of the defect candidate 71 is further performed based on the three-dimensional shape data D2 (second inspection), and the final quality judgment is performed.

Conventionally, it has been common practice to create a rule for recognizing and classifying defects based on the image feature amount of the defect extracted in advance, and to extract defect candidates based on this rule. However, in such a rule-based algorithm, it is necessary to adjust parameters and threshold values for each individual condition such as defect form, item, and work shape. Further, it takes a lot of man-hours such as repeating the parameter adjustment work in order to reduce oversight and over-detection even at the time of startup adjustment of the inspection device or after operation. In addition, the judgment accuracy depends on the skill of the inspector, and there is a limit to the algorithm, so that the judgment accuracy is not sufficient for a complicated inspection object.

In this respect, in the present embodiment, by applying deep learning to the defect candidate extraction, complicated parameter adjustment and threshold adjustment become unnecessary, and the development man-hours can be significantly reduced. However, it is difficult to obtain numerical data such as defect dimensions, depth, and height only by deep learning, and since the internal processing is a black box, the results obtained by deep learning are technical. There is a problem that it is difficult to show the rationale.

Therefore, in the present embodiment, when defect candidates are extracted by deep learning, shape inspection (dimension inspection) is further performed on the defect candidates, and final quality judgment is performed. Since the shape inspection (dimension inspection) is performed based on numerical data such as the dimensions, depth, and height of defect candidates, the technical basis for the inspection results becomes clear. It is possible to perform the inspection only by the shape inspection, but since the three-dimensional shape measurement by the 3D sensor 45 takes a long measurement time, in the present embodiment, the first measurement / inspection is comprehensive based on the two-dimensional image. We adopted a two-step measurement / inspection method in which a detailed shape inspection (dimension inspection) based on a three-dimensional shape is performed by the second measurement / inspection.

Although the preferred embodiment of the inspection device 1 and the like according to the present disclosure has been described above with reference to the attached drawings, the present disclosure is not limited to such an example.

For example, as shown in FIG. 24, the 2D optical system 42 and the 3D sensor 45 may be fixed to the inspection table 19 or the like of the inspection device 1A. In this case, the trajectory control of the measurement robot arm 41 becomes unnecessary. That is, in step S23 of FIG. 18, the control unit 61 (orbit generation unit 143) of the control device 6 has a gripping robot arm so that the position of the camera 43 relative to the inspection object 7 is the measurement position 10. Generates 31 orbits. Then, the control unit 61 (robot control unit 13) controls the trajectory of the gripping robot arm 31 based on the generated trajectory, and positions the inspection object 7.

Further, a 3D scanner and a tracking marker are mounted on the tip of the measurement robot arm 41, and the control unit 61 tracks the operation of the tracking marker with a tracking system to measure the three-dimensional shape of the inspection object 7. It may be configured as.

It is clear that a person skilled in the art can come up with various modifications or modifications within the scope of the technical idea disclosed in the present application, and these also naturally belong to the technical scope of the present disclosure. It is understood that.

1: Inspection device 2: Supply mechanism 2D: Measuring robot arm 3: Gripping mechanism 4: Measuring mechanism 5: Discharge mechanism 6: Control device 7: Inspection object 8: Shape data 9: Inspection specifications 10: Measurement position 11: Conveyor Control unit 12: Work recognition unit 13: Robot control unit 14: Operation setting unit 15: Measurement unit 16: Inspection unit 17: Data display unit 21: Conveyor 31: Gripping robot arm 31a: Gripping robot arm 31b: Gripping robot Arm 41: Measurement robot arm 42: 2D optical system 43: Camera 44: Lighting 45: 3D sensor 51: Conveyor 71: Defect candidate 72: Deep learner 80: Inspection site 80a: Inspection site 80b: Inspection site 81: Geometric shape 81: Spherical grid 81: Polyhedron 82: Surface 83: Center 84: Surface normal line 91: Defect type 92: Defect specification 93: Illumination condition 94: First measurement data 95: Second measurement data 96: Inspection result 100: Measurement inspection Table 101: Measurement candidate position 101a: Measurement candidate position 101b: Measurement candidate position 101c: Measurement candidate position 140: Inspection site setting unit 141: Inspection specification setting unit 142: Measurement position calculation unit 142a: Geometric generation unit 142b: Measurement candidate position setting Unit 142c: Measurable area calculation unit 142d: Measurement position selection unit 143: Orbit generation unit 151: First measurement unit 152: Second measurement unit 161: First inspection unit 162: Second inspection unit 311: Arm unit 311a: Arm Part 311b: Arm part 312: Hand part 312a: Hand part 312b: Hand part D1: Two-dimensional image data D2: Three-dimensional shape data P: Measurable area P1: Measurable area P2: Measurable area P3: Measurable area R : Integrated area R1: Integrated area R2: Integrated area

Claims (16)

1. A gripping robot arm that grips the object to be inspected, a measurement sensor,
   A measurement position calculation unit that calculates the measurement position of the measurement sensor with respect to the inspection object based on the shape data of the inspection object.
   A trajectory generating unit that generates a trajectory of the gripping robot arm based on the measurement position,
   A robot control unit that controls the operation of the gripping robot arm based on the trajectory and positions the inspection object.
   After positioning, the measuring unit that measures the inspection object with the measuring sensor and
   An inspection unit that inspects the inspection object based on the measured data,
   Inspection device equipped with.
2. The measurement sensor is provided on the measurement robot arm.
   The trajectory generating unit generates the trajectories of the gripping robot arm and the measuring robot arm.
   The inspection device according to claim 1, wherein the robot control unit controls the operations of the gripping robot arm and the measurement robot arm based on the trajectory, and positions the inspection object and the measurement sensor.
3. Further provided with an inspection site setting unit for setting an inspection site to be inspected in the shape data.
   The inspection device according to claim 1 or 2, wherein the measurement position calculation unit calculates the measurement position so that all inspection sites are measured by the measurement sensor.
4. The measurement position calculation unit
   A geometry generating unit that includes the shape data and generates a geometry that can define each direction of the shape data.
   A measurement candidate position setting unit that sets one or more measurement candidate positions in each direction defined by the geometric shape, and a measurement candidate position setting unit.
   The inspection device according to claim 3, further comprising a measurement position selection unit that selects a measurement position from the measurement candidate positions.
5. The geometry generator generates a polyhedron that includes the shape data,
   The inspection device according to claim 4, wherein the measurement candidate position setting unit sets the measurement candidate position on the normal direction of each surface of the polyhedron.
6. The inspection device according to claim 4 or 5, wherein the measurement position selection unit selects a measurement position so that all inspection sites are measured and the number of measurements is minimized.
7. The inspection device according to any one of claims 3 to 6, further comprising an inspection specification setting unit for setting a defect type to be inspected for an inspection site set by the inspection site setting unit.
8. The inspection specification setting unit further sets a limit standard which is a standard for determining harmful / harmlessness of the defect of the defect type.
   The inspection device according to claim 7, wherein the inspection unit inspects the inspection object by collating the measured data with the limit standard of the defect type.
9. The inspection device according to claim 7 or 8, wherein the measuring unit measures the inspection object under lighting conditions according to the defect type.
10. The measuring unit
    A first measuring unit that measures a two-dimensional image of the inspection object,
    The inspection device according to any one of claims 1 to 9, further comprising a second measuring unit for measuring the three-dimensional shape of the inspection object.
11. The inspection unit
    A first inspection unit that extracts defect candidates of the inspection object based on a two-dimensional image measured by the first measurement unit, and a first inspection unit.
    The inspection device according to claim 10, further comprising a second inspection unit that inspects the shape of the defect candidate extracted based on the three-dimensional shape measured by the second measurement unit.
12. The inspection device according to any one of claims 1 to 11, further comprising a display unit for displaying the inspection result of the inspection unit.
13. The inspection device according to any one of claims 1 to 12, wherein the robot control unit controls the inspection object gripped by the gripping robot arm so as to be gripped by another gripping robot arm.
14. The computer
    A step of calculating the measurement position of the measurement sensor with respect to the inspection object based on the shape data of the inspection object, and
    A step of generating a trajectory of a gripping robot arm that grips the inspection object based on the measurement position, and
    A step of controlling the operation of the gripping robot arm based on the trajectory and positioning the inspection object, and
    After positioning, the step of measuring the inspection object with the measurement sensor and
    Steps to inspect the inspection object based on the measured data,
    Inspection method to perform.
15. Computer,
    A measurement position calculation unit that calculates the measurement position of the measurement sensor with respect to the inspection object based on the shape data of the inspection object.
    A trajectory generator that generates a trajectory of a gripping robot arm that grips the inspection object based on the measurement position.
    A robot control unit that controls the operation of the gripping robot arm based on the trajectory and positions the inspection object.
    After positioning, the measuring unit that measures the inspection object with the measuring sensor,
    Inspection unit that inspects the inspection object based on the measured data,
    A program that functions as.
16. The computer
    A step of calculating the measurement position with respect to the inspection object based on the shape data of the inspection object, and
    A step of generating a trajectory of a gripping robot arm that grips the inspection object based on the measurement position, and
    A step of controlling the operation of the gripping robot arm based on the trajectory and positioning the inspection object, and
    Positioning method to perform.
    

Images