Cutting processing system, control method, program, and control system

Abstract

To enhance the work efficiency of the cutting work.SOLUTION: A cutting system 1 includes a cutting unit 10, an imaging unit 20 and a control device 60. The cutting unit 10 includes a cutting tool 11 that performs cutting to an object 3. The cutting unit 10 performs cutting to the object 3 by using the cutting tool 11 with control by the control device 60. The imaging unit 20 images the object 3. The control device 60 includes a region extraction unit 67 and a processing control unit 71. The region extraction unit 67 extracts a processing possible region R1 where the cutting work can be performed on a surface of the object 3 by inputting a pre-processing image obtained by imaging the object 3 by the imaging unit 20 to a first learned model LM1. The processing control unit 71 controls the cutting unit 10 so as to perform cutting to the processing possible region R1.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure generally relates to a cutting system, a control method, a program, and a control system, and more particularly to a cutting system, a control method, a program, and a control system that cuts an object. [Background technology]

[0002] Patent Document 1 discloses a cutting device. This cutting device is composed of an imaging device that captures images of the material to be cut placed in the cutting area on a pallet, a cutting torch, and a control device. The imaging device captures images of the cutting area on the pallet where the material to be cut is placed. The control device processes the images captured by the imaging device and assigns a pre-stored figure to be cut to the material to be cut. The control device then controls the cutting torch based on the assigned data to cut the material. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2003-251464 Summary of the Invention [Problem to be solved by the invention]

[0004] When there are areas on the surface of the material being cut that are unsuitable for cutting (such as knots in the wood), the work of determining and setting the workable areas other than the areas that are unsuitable for cutting on the material being cut is done by hand, which results in poor work efficiency.

[0005] An object of the present disclosure is to provide a cutting system, a cutting method, a program, and a control system that can improve the efficiency of cutting work. [Means for solving the problem]

[0006] A cutting system according to one aspect of the present disclosure includes a cutting unit, an imaging unit, and a control device. The cutting unit has a cutting tool for cutting an object. The cutting unit cuts the object using the cutting tool under the control of the control device. The imaging unit Before performing the first cutting process using the cutting tool the object Before processing image The control device includes a region extraction unit and a processing control unit. The region extraction unit extracts an image of the object obtained by the imaging unit. The aforementioned By inputting the unprocessed image into the first trained model, on the surface of the object, The first For cutting Removed unsuitable areas that cannot be processed The machining control unit extracts a machining area. The first cutting processing The cutting unit is controlled to perform the above.

[0007] A control method according to one aspect of the present disclosure is a control method for a cutting system including a cutting unit and an imaging unit, and includes an area extraction step and a processing control step. The cutting unit has a cutting tool for cutting an object. The imaging unit includes: Before performing the first cutting process using the cutting tool In the region extraction step, the imaging unit captures an image of the object. Before the first cutting process By inputting the before-machining image into the first trained model, the first cutting process is performed on the surface of the object. Areas that are unsuitable for processing have been removed. In the machining control step, a machining area is extracted. The first cutting process The cutting unit is controlled to perform the above.

[0008] A program according to one aspect of the present disclosure is a program for causing a computer system to execute the control method.

[0009] A control system according to an embodiment of the present disclosure includes an area extraction unit and a machining control unit. The area extraction unit inputs a pre-machined image of the object captured by an imaging unit before a first cutting process into a first trained model, thereby extracting a region on the surface of the object. , the firstCutting processing Areas that are unsuitable for processing have been removed. The machining control unit extracts a machining area. The first cutting processing To do , cut A cutting unit having a cutting tool is controlled. [Effects of the Invention]

[0010] According to the present disclosure, the work efficiency of cutting processing can be improved. [Brief explanation of the drawings]

[0011] [Figure 1] FIG. 1 is a block diagram showing the configuration of a cutting system according to an embodiment. [Figure 2] FIG. 2 is a schematic system configuration diagram of the cutting processing system. [Figure 3] FIG. 3 is a flowchart showing the operation of the cutting system. [Figure 4] FIG. 4 is an explanatory diagram illustrating an imaging operation by the imaging unit of the cutting processing system. [Figure 5] FIG. 5 is an explanatory diagram showing an example of a plurality of images captured by an imaging unit of the cutting processing system. [Figure 6] FIG. 6 is an explanatory diagram showing an example of an unprocessed image synthesized by the image synthesis unit of the cutting system. [Figure 7] FIG. 7 is an explanatory diagram showing an example of a pre-processing image of the upper surface extracted by the upper surface extraction unit of the cutting system. [Figure 8] FIG. 8 is an explanatory diagram showing the extraction result of the cutting system extracting the machineable area from the before-machining image. [Figure 9] FIG. 9 is an external perspective view showing an example of a workpiece cut out from a target object by the cutting system. [Figure 10] FIG. 10 is a diagram showing an example of an arrangement pattern created by the nesting processing unit of the cutting system. [Figure 11]FIG. 11 is a perspective view showing the appearance of an object after being cut by the cutting system. DETAILED DESCRIPTION OF THE INVENTION

[0012] The embodiments and modifications described below are merely examples of the present disclosure, and the present disclosure is not limited to the following embodiments and modifications. Various modifications other than the following embodiments and modifications are possible depending on the design, etc., as long as they do not deviate from the technical concept of the present disclosure.

[0013] (Embodiment) Hereinafter, a cutting system 1 according to an embodiment will be described with reference to FIGS.

[0014] (1) Overview As shown in FIGS. 1 and 2, the cutting system 1 according to this embodiment includes a cutting unit 10, an imaging unit 20, and a control device 60 serving as a control system 2. The cutting unit 10 has a cutting tool 11 that cuts an object 3 (see FIGS. 2 and 4) that is the target of a cutting operation. The cutting unit 10 performs cutting on the object 3 using the cutting tool 11 under the control of the control device 60. The imaging unit 20 captures an image of the object 3. As shown in FIG. 1, the control device 60 serving as the control system 2 includes an area extraction unit 67 and a machining control unit 71. The area extraction unit 67 inputs a before-machining image obtained by capturing an image of the object 3 with the imaging unit 20 into a first trained model LM1, thereby extracting a machineable area R1 (see FIG. 8) on the surface of the object 3 that can be machined. The machining control unit 71 controls the cutting unit 10 to perform cutting on the machineable area R1.

[0015] Here, the target object 3 is, for example, a member such as wood, metal, or synthetic resin. The following description will be given using an example in which the target object 3 is wood. The "machinable area" that can be cut is the area on the surface of the target object 3 excluding the unmachinable area. The "unmachinable area" is an area unsuitable for cutting, in other words, an area where cutting is best avoided. For example, if the target object 3 is wood, the unmachinable area may include an area where knots 301 exist or an area where the wood grain direction in the finished product cut from the target object 3 is aesthetically undesirable. The unmachinable area may include an area on the target object 3 where warping, unevenness, dirt, etc. exist. If the target object 3 is a scrap wood from a previous process such as cutting a product, it may include an area where processing marks 302 from previous processing such as cutting exist. The unmachinable area may also include an area where a fixing member (e.g., a screw, etc.) 303 exists for fixing the target object 3 to the table-like base 50 (see FIG. 2 ) provided in the cutting system 1. The unmachinable area is preferably set to be slightly wider than the area unsuitable for cutting, and preferably includes the area unsuitable for cutting and a buffer area of ​​a predetermined width provided around it.

[0016] According to this configuration, the area extraction unit 67 extracts the machinable area R1 by inputting the before-machining image of the object 3 captured by the imaging unit 20 into the first trained model LM1, so there is no need for a person to manually set the machinable area by looking at the object 3. This has the advantage of improving the work efficiency of cutting processing by the cutting processing system 1.

[0017] In the following, the front-rear direction, left-right direction, and up-down direction are defined as shown in FIG. 2. The left-right direction is the direction in which a pair of first members 31, which will be described later, are aligned. The up-down direction is the thickness direction of the object 3. The front-rear direction is a direction perpendicular to both the left-right direction and the up-down direction. Although arrows representing these directions (up, down, left, right, front, rear) are shown in FIG. 2 for the purpose of assisting the explanation, these arrows have no substance. Furthermore, the definition of the above directions is not intended to limit the manner in which the cutting processing system 1 of this embodiment is used.

[0018] (2) Composition The detailed configuration of the cutting system 1 according to this embodiment will be described below.

[0019] As described above, the cutting system 1 includes the cutting unit 10, the imaging unit 20, and the control device 60 serving as the control system 2. The cutting system 1 also includes a cutting chip removal unit 40, a robot unit 30, and a base 50. The cutting system 1 also displays information such as the processable area extracted by the control device 60 on a display terminal 80 used by a user of the cutting system 1.

[0020] (2.1) Robot Unit As shown in FIG. 2, the robot unit 30 has a pair of first members 31, a pair of second members 32, and a third member 33.

[0021] The pair of first members 31 are arranged to sandwich a base 50 on which the target object 3 is placed. Each first member 31 extends in the front-to-rear direction. The pair of first members 31 correspond one-to-one to the pair of second members 32, and support the corresponding second members 32 so that the corresponding second members 32 are movable in the front-to-rear direction.

[0022] The pair of second members 32 are supported by the corresponding first members 31 so as to be movable in the front-rear direction. The pair of second members 32 extend in the up-down direction. That is, the pair of second members 32 are supported by the corresponding first members 31 so as to have one end in the up-down direction movable in the front-rear direction.

[0023] The third member 33 extends in the left-right direction. The third member 33 is supported by the pair of second members 32 so as to be movable in the up-down direction. Specifically, one end of both ends of the third member 33 in the left-right direction is supported by one of the pair of second members 32. The other end of both ends of the third member 33 is supported by the other of the pair of second members 32.

[0024] The movement of the pair of second members 32 in the front-rear direction and the movement of the third member 33 in the left-right direction are controlled by a control device 60.

[0025] (2.2) Cutting unit As shown in FIGS. 1 and 2, the cutting unit 10 has a cutting tool 11 that cuts an object 3 that is the target of the cutting work. The cutting unit 10 has a main body 12. The cutting tool 11 is attached to the main body 12. The cutting unit 10 cuts the object 3 using the cutting tool 11 under the control of a control device 60. Here, the object 3 is, for example, wood.

[0026] The cutting unit 10 is configured to allow the attachment and detachment of cutting tools 11. An appropriate cutting tool 11 is attached to the cutting unit 10 depending on the material of the object 3, the shape of the portion to be cut, etc. In other words, the cutting unit 10 is equipped with multiple types of cutting tools 11 that can be used depending on the material of the object 3, the shape of the portion to be cut, etc., and a holder to which the cutting tools 11 can be replaced. The cutting unit 10 attaches a cutting tool 11 selected from the multiple types of cutting tools 11 depending on the material of the object 3, the shape to be cut, etc., to the holder, and performs cutting processing on the object 3 using the cutting tool 11 attached to the holder.

[0027] The cutting unit 10 is attached to the third member 33 of the robot unit 30 so as to be movable in the left-right direction. The cutting unit 10 moves in the left-right direction on the third member 33 under the control of the control device 60.

[0028] (2.3) Cutting debris removal unit The cutting chip removal unit 40 is controlled by the control device 60 to remove cutting chips generated when the cutting unit 10 cuts the target object 3. The cutting chip removal unit 40 is attached to the cutting unit 10. Specifically, it is attached between the cutting tool 11 and the main body 12.

[0029] As shown in FIG. 1, the cutting chip removal unit 40 includes a suction unit 41, a rotary brush unit 42, a polishing unit 43, and a spraying unit 44.

[0030] The suction unit 41 sucks the cutting chips. Specifically, the suction unit 41 sucks the cutting chips under the control of the control device 60.

[0031] The rotating brush unit 42 removes chips adhering to the cutting surface after cutting the object 3. The rotating brush unit 42 is formed in a cylindrical shape with a central axis along the vertical direction, and brushes are provided on the side surface. The rotating brush unit 42 rotates around the central axis as a rotation axis under the control of the control device 60, thereby removing chips adhering to the cutting surface.

[0032] The polishing unit 43 polishes the cutting surface. The polishing unit 43 has a cylindrical rotating drum with the central axis as its axis, and an abrasive material such as sandpaper held on the circumferential surface of the rotating drum. The rotating drum of the polishing unit 43 is arranged so as to be sandwiched between the rotating brush unit 42. When the rotating brush unit 42 rotates under the control of the control device 60, the rotating drum of the polishing unit 43 rotates, and the abrasive material held on the circumferential surface of the rotating drum polishes the cutting surface. Note that the polishing unit 43 is not an essential component of the cutting chip removal unit 40.

[0033] The cutting chip removal unit 40 removes not only cutting chips generated by cutting by the cutting unit 10 but also polishing powder generated by polishing by the polishing section 43. Specifically, in the cutting chip removal unit 40, the rotating brush section 42 removes polishing powder generated by polishing by the polishing section 43 from the cutting surface, and the suction section 41 sucks up the removed polishing powder together with the cutting chips.

[0034] The blowing unit 44 blows air onto the machined portion (machined portion) of the target object 3. This causes cutting chips (including cutting chips generated by cutting and polishing powder generated by polishing by the polishing unit 43) accumulated at the machined portion to fly up into the air, and the cutting chips can be sucked up by the suction unit 41.

[0035] (2.4) Imaging unit The imaging unit 20, under the control of the control device 60, captures a pre-processing image of the object 3 before cutting by the cutting unit 10. The imaging unit 20, under the control of the control device 60, also captures a post-processing image of the object 3 after cutting by the cutting unit 10. The post-processing image is an image of the object 3 in a state where the workpiece processed by cutting has not been completely separated from the object 3, i.e., where a part of the workpiece is connected to the object.

[0036] The imaging unit 20 is attached to the third member 33 of the robot unit 30 so as to be movable in the left-right direction. Furthermore, the imaging unit 20 is movable in the front-back direction together with the third member 33 of the robot unit 30 as the third member 33 moves in the front-back direction. In other words, the imaging unit 20 is indirectly attached to the cutting unit 10 via the robot unit 30, and captures images of a predetermined range including the processing position where the cutting tool 11 of the cutting unit 10 performs cutting. In other words, the imaging unit 20 is attached to the movably provided cutting unit 10, and captures images of the target object 3 at each of a plurality of imaging positions as shown in FIG. 4.

[0037] As shown in FIGS. 1 and 2 , the imaging unit 20 includes an imaging device 21. The imaging device 21 captures images of at least a portion of the object 3. In this embodiment, because the imaging device 21 cannot capture an image of the entire object 3 at once, the control device 60 moves the imaging unit 20 to cause the imaging device 21 to capture images of the object 3 at multiple imaging positions. That is, the imaging unit 20 is attached to the movably mounted cutting unit 10 and captures images of the object 3 multiple times at different imaging positions. The image composition unit 65 of the control device 60 then combines the images captured at the multiple imaging positions to generate a single image representing the entire object 3. In other words, the image composition unit 65 generates a pre-machined image by combining multiple images obtained by the imaging unit 20 capturing images of the object 3 at each of the multiple imaging positions before cutting by the cutting unit 10. Similarly, the image composition unit 65 generates a pre-machined image by combining multiple images obtained by the imaging unit 20 capturing images of the object 3 at each of the multiple imaging positions after cutting by the cutting unit 10. In addition, among the multiple images taken to combine one pre-processing image or one post-processing image, two adjacent images may partially overlap at the boundary, or may be in contact at the boundary without overlapping.

[0038] As described above, in this embodiment, even if the entire object 3 does not fit within the angle of view of the imaging unit 20, a pre-processing image representing the entire object 3 can be generated by combining multiple images of the object 3 captured by changing the imaging position of one imaging unit 20. Therefore, there is no need to provide multiple imaging units 20 to capture the entire object 3, and by reducing the number of imaging units 20, the system configuration of the cutting processing system 1 can be simplified.

[0039] The imaging device 21 has a wide-angle lens to widen the imaging range, and by capturing images over a wider range than when it does not have a wide-angle lens, the number of times it is necessary to capture an image of the entire object 3 can be reduced.

[0040] (2.5) Pedestal The base 50 is disposed so as to be sandwiched between the pair of first members 31. As shown in FIG. 2, the base 50 has a plate portion 51, a plurality of (four in the illustrated example) positioning units 52, and a tool holding unit 53.

[0041] The target object 3 is placed on the plate portion 51. Two of the four positioning units 52 are aligned in the left-right direction, and the remaining two positioning units 52 are aligned in the front-to-rear direction. The placement position of the target object 3 is determined by bringing one side of the target object 3 into contact with the two positioning units 52 aligned in the left-to-right direction and bringing another side intersecting the first side into contact with the two positioning units 52 aligned in the front-to-rear direction. The target object 3 may be fixed to the plate portion 51 by threading fixing members 303 passed through holes provided in the four corners of the target object 3 into screw holes in the plate portion 51.

[0042] The tool holding unit 53 holds one or more cutting tools 11 (cutting tools 11a and 11b in the illustrated example) to be attached to the main body 12 of the cutting unit 10. The cutting unit 10 selects an appropriate cutting tool 11 from the multiple types of cutting tools 11 held in the tool holding unit 53 depending on the material of the object 3 or the shape of the portion to be cut out by cutting, and attaches the selected cutting tool 11 to a holder provided in the main body 12. The cutting unit 10 then cuts the object 3 using the cutting tool 11 attached to the holder.

[0043] (2.6) Display terminal The display terminal 80 is a computer terminal used by a user of the cutting system 1. The display terminal 80 may be a personal computer, a notebook computer, a tablet computer, a smartphone, or the like.

[0044] The display terminal 80 includes a communication unit 81 , a display unit 82 , and a control unit 83 .

[0045] The communication unit 81 is a communication interface for communicating with the control device 60. The communication unit 81 communicates with the control device 60 via, for example, a cable 90. Note that the communication unit 81 may communicate with the control device 60 via wireless communication.

[0046] The display unit 82 is a display device such as a liquid crystal display.

[0047] The control unit 83 displays the received various pieces of information on the display unit 82 based on the various pieces of information received by the communication unit 81 from the control device 60 .

[0048] (2.7) Control device As shown in FIG. 1, the control device 60 includes a communication unit 61, a storage unit 62, and a control unit 63.

[0049] The control device 60 includes, for example, a computer system having a processor and a memory. The processor executes a program stored in the memory, causing the computer system to function as the control unit 63. The program executed by the processor is pre-recorded in the memory of the computer system here, but may also be provided by being recorded on a non-transitory recording medium such as a memory card, or may be provided via a telecommunications line such as the Internet.

[0050] The communication unit 61 is a communication interface for communicating with each of the cutting unit 10, the imaging unit 20, the robot unit 30, the cutting debris removal unit 40, and the display terminal 80. The communication unit 61 communicates with each of the cutting unit 10, the imaging unit 20, the robot unit 30, the cutting debris removal unit 40, and the display terminal 80 via a cable 90. The communication unit 61 may also communicate with each of the cutting unit 10, the imaging unit 20, the robot unit 30, the cutting debris removal unit 40, and the display terminal 80 via wireless communication.

[0051] The storage unit 62 is configured by a device selected from a ROM (Read Only Memory), a RAM (Random Access Memory), an EEPROM (Electrically Erasable Programmable Read Only Memory), or the like.

[0052] The storage unit 62 stores first to fourth trained models LM1 to LM4 that are used when the control unit 63 performs the inference phase.

[0053] The storage unit 62 also stores programs necessary for cutting. For example, the storage unit 62 stores a tool control program, a cutting chip removal program, and an imaging control program. The tool control program is a program for controlling the cutting unit 10 (cutting tool 11) to move along a target shape on the object 3. The cutting chip removal program is a program set so that the cutting chip removal unit 40 removes cutting chips. The imaging control program is a program set so that the operation of the imaging unit 20 is controlled. The imaging control program includes, for example, an imaging path (imaging order) for the object 3.

[0054] The memory unit 62 stores processing drawing data and workpiece information of the workpiece to be processed from the target object 3. The processing drawing data may be two-dimensional CAD drawing data or three-dimensional CAD drawing data. The processing drawing data includes the dimensions of the workpiece. The workpiece information includes information related to the workpiece. The information related to the workpiece includes, for example, information on whether or not three-dimensional processing is to be performed, the maximum depth to be processed, the minimum value of the gap to be provided between the workpieces when cutting multiple workpieces, the presence or absence of holes in the workpieces, etc.

[0055] The storage unit 62 also stores attribute information of the target object 3 in association with the processing drawing data. The attribute information includes, for example, information such as the type of material, size, and thickness of the target object 3. The attribute information may also include information such as process information indicating the processing process for the target object 3, entry position information indicating the entry position of the tool (cutting tool 11), and retract position information indicating the retract position (position where the tool (cutting tool 11) leaves the target object 3). The process information is, for example, information indicating the path along which the cutting unit 10 is moved in order to process the contour of the workpiece to be processed from the target object 3.

[0056] As shown in FIG. 1 , the control unit 63 includes functions such as a correction unit 64, an image synthesis unit 65, an upper surface extraction unit 66, a region extraction unit 67, a nesting processing unit 68, a pass / fail determination unit 69, an output unit 70, a processing control unit 71, and a learning unit 72. The learning unit 72 also includes first, third, and fourth learning units 73, 75, and 76 that generate first, third, and fourth trained models LM1, LM3, and LM4, respectively, by performing machine learning. Note that in FIG. 1 , the correction unit 64, the image synthesis unit 65, the upper surface extraction unit 66, the region extraction unit 67, the nesting processing unit 68, the pass / fail determination unit 69, the output unit 70, the processing control unit 71, and the learning unit 72 do not represent actual components, but rather represent functions realized by the control unit 63. Similarly, the first, third, and fourth learning units 73, 75, and 76 do not represent actual components, but rather represent functions realized by the control unit 63.

[0057] The correction unit 64 performs distortion correction on the image captured by the imaging unit 20. The imaging device 21 included in the imaging unit 20 is equipped with a wide-angle lens so that it can capture a wide range, and therefore, distortion such as a square becoming barrel-shaped may occur in the image captured by the imaging device 21. The correction unit 64 performs correction processing on the image captured by the imaging device 21 to correct distortion caused by the wide-angle lens, and outputs a corrected image in which distortion caused by the wide-angle lens has been reduced. Note that, since the imaging unit 20 captures images of the object 3 multiple times at different imaging positions to capture one unprocessed image, the correction unit 64 performs distortion correction on each of the multiple images obtained by the imaging unit 20 capturing images of the object 3 at each of the multiple imaging positions.

[0058] The image composition unit 65 generates a composite image representing the entire object 3 by combining multiple images captured by the imaging device 21 of the imaging unit 20 at multiple imaging positions. Here, the image composition unit 65 can generate a pre-processing image P10 (see FIG. 6) of the object 3 before processing by combining multiple images (e.g., six images P11 to P16 in FIG. 5) captured by the imaging unit 20 before the cutting unit 10 performs cutting on the object 3. In other words, the image composition unit 65 generates a pre-processing image by combining multiple images obtained by the imaging unit 20 capturing images of the object 3 multiple times. Note that the before-processing image P10 shown in FIG. 6 does not show the surface condition of the object 3. The image composition unit 65 can also generate a post-processing image of the object 3 after processing by combining multiple images captured by the imaging unit 20 after the cutting unit 10 performs cutting on the object 3.

[0059] 5, the multiple images P11 to P16 are images that show only the top surface of the object 3, but because the imaging device 21 is equipped with a wide-angle lens, each of the images P11 to P16 may also show the side surface of the object 3. Therefore, the unprocessed image P10 (see FIG. 6) obtained by combining the multiple images P11 to P16 may be an image that shows not only the top surface 310 of the object 3 but also the side surface 320.

[0060] Furthermore, because the imaging device 21 is equipped with a wide-angle lens, distortion occurs in the image captured by the imaging unit 20, such as distorting a square into a barrel shape. In this embodiment, when the image synthesis unit 65 synthesizes the before-processing images, it generates the before-processing image by synthesizing multiple corrected images after the correction unit 64 has performed distortion correction on the multiple images, thereby reducing distortion in the before-processing image. Note that when the image synthesis unit 65 synthesizes the after-processing image, it generates the after-processing image by synthesizing multiple corrected images after the correction unit 64 has performed distortion correction on the multiple images, thereby reducing distortion in the after-processing image.

[0061] The upper surface extraction unit 66 extracts a partial image P20 (see FIG. 7) corresponding to the upper surface 310 of the object 3 from the pre-processed image (pre-processed image P10 in FIG. 6) generated by the image synthesis unit 65. For example, the upper surface extraction unit 66 extracts the partial image P20 corresponding to the upper surface 310 of the object 3 by inputting the pre-processed image P10 to the fourth trained model LM4. The fourth trained model LM4 is a learning model generated by, for example, the fourth learning unit 76 performing machine learning using multiple pre-processed images P10 showing the upper surface 310 and the side surface 320 as training data. Note that in general image processing, calculation processing such as edge detection is required to extract the upper surface 310 from an image showing the upper surface 310 and the side surface 320. In this embodiment, the pre-processed image P10 is input to the fourth trained model LM4 to extract the partial image P20 of the upper surface 310. Therefore, processing time can be reduced compared to when the partial image P20 of the upper surface is extracted by image processing such as edge detection.

[0062] The area extraction unit 67 extracts a cuttable area on the surface (top surface) of the object 3 where cutting is possible by inputting the before-machining image into the first trained model LM1. Here, the area extraction unit 67 extracts a cuttable area R1 from the partial image P20 extracted by the top surface extraction unit 66 in the before-machining image. Because the cutting unit 10 performs cutting on the top surface of the object 3, by extracting the cuttable area R1 from the partial image P20 of the top surface excluding the side surfaces, etc. of the object 3, the amount of calculation required for the extraction process of the cuttable area R1 can be reduced, and the time required for the extraction process can be shortened.

[0063] The nesting processing unit 68 creates an arrangement pattern PT1 (see FIG. 10 ) that assigns machining positions of one or more workpieces 100 (see FIG. 9 ) cut out from the target object 3 to the before-machining image. The nesting processing unit 68 creates an arrangement pattern PT1 that maximizes the number of workpieces 100 that can be arranged in the machined area R1, for example, based on the shape and size of the machined area R1, machining drawing data of the workpieces 100 stored in the memory unit 62, and the shape of the workpieces 100 when completed. The nesting processing unit 68 performs arithmetic processing using a genetic algorithm to create an arrangement pattern PT1 that optimally arranges multiple workpieces 100 in the machined area R1. The arrangement pattern PT1 is a diagram in which figures showing the shapes of the workpieces 100 in a planar view are arranged at the machining positions of the workpieces 100. In FIG. 10 , the arrangement positions of the workpieces 100 are indicated by two-dot chain lines. Note that the arrangement rule by which the nesting processing unit 68 arranges one or more workpieces 100 in the machined area R1 can be changed as appropriate. The nesting processing unit 68 may arrange one or more workpieces 100 in the machineable area R1 so that the area occupied by the machineable area R1 when a predetermined number of workpieces 100 are arranged therein is minimized, for example. The number of workpieces 100 arranged in the machineable area R1 by the nesting processing unit 68 is not limited to one type. The nesting processing unit 68 may arrange multiple types of workpieces 100 with different shapes or sizes in the machineable area R1, and may create an arrangement pattern PT1 that allows multiple types of workpieces 100 to be arranged in the machineable area R1 in their respective numbers.

[0064] The machining control unit 71 controls the cutting unit 10 and the robot unit 30 to perform cutting on the machineable region R1. More specifically, the machining control unit 71 controls the cutting unit 10 and the robot unit 30 to perform cutting at the machining positions of one or more workpieces 100 based on the placement pattern PT1 created by the nesting processing unit 68. Since the cutting unit 10 is moved by the robot unit 30, the machining control unit 71 controls the cutting unit 10 and the robot unit 30 to perform cutting on one or more workpieces 100 at the machining positions set in the placement pattern PT1.

[0065] The acceptability determination unit 69 determines whether the processing of the object 3 is acceptable based on a processed image obtained by capturing an image of the object 3 with the imaging unit 20 after the object 3 is processed by the cutting unit 10. Specifically, the acceptability determination unit 69 determines whether the processing of the object 3 is acceptable by inputting the processed image to the third trained model LM3. The third trained model LM3 is a trained model generated by machine learning using, as training data, good product data related to the post-processing state of the object 3 determined to be a good product and the post-processing image of an object determined to be a defective product. The good product data includes at least one of drawing data related to the workpiece 100 (e.g., two-dimensional CAD drawing data or three-dimensional CAD drawing data), image data of a virtual image created based on the drawing data related to the workpiece 100, and the post-processing image of the object determined to be a good product. In addition, the post-processing image is an image captured in a state in which one or more workpieces 100 are partially connected to the object 3, and one or more workpieces 100 are detached from the object 3 by cutting the connection portion with the object 3.

[0066] The quality determining unit 69 determines whether the processing is good or bad, and when a defect B1 or the like exceeding a predetermined size occurs in the contour portion of the workpiece 100 produced by cutting the target object 3 as shown in FIG. 11 , the quality determining unit 69 can determine whether the processing is bad. The quality determining unit 69 may also determine whether the pattern (e.g., wood grain) on the surface of the workpiece 100 matches the desired pattern. The quality determining unit 69 may also determine whether the processing is good or bad by determining the presence or absence of unevenness, dirt, etc. on the surface of the workpiece 100. In this way, the quality determining unit 69 determines whether the processing is good or bad, thereby improving the quality of the workpiece 100.

[0067] The output unit 70 outputs the judgment result of the pass / fail judgment unit 69. For example, the output unit 70 outputs the judgment result of the pass / fail judgment unit 69 to the display terminal 80 via the communication unit 61, and the judgment result of the pass / fail judgment unit 69 is displayed on the display unit 82 of the display terminal 80. This allows a user of the cutting system 1 to understand the pass / fail of the machining of the workpiece 100. The output unit 70 may further output at least one of the before-machining image P10 synthesized by the image synthesis unit 65, the partial upper surface image P20 extracted by the upper surface extraction unit 66, the machineable region R1 extracted by the region extraction unit 67, and the arrangement pattern PT1 created by the nesting processing unit 68. This allows a user of the cutting system 1 to understand the before-machining image P10, the partial upper surface image P20, the machineable region R1, the arrangement pattern PT1, etc. of the target object 3.

[0068] The learning unit 72 includes a first learning unit 73, a third learning unit 75, and a fourth learning unit 76 that generate the above-mentioned first, third, and fourth trained models LM1, LM3, and LM4, respectively, through machine learning. Note that the second learning unit 74 shown in FIG. 1 will be described in a modified example.

[0069] The first learning unit 73 generates the first trained model LM1 by performing machine learning using, for example, a data set of images of a plurality of objects 3, each having an unmachinable area unsuitable for cutting, as training data. The first learning unit 73 generates the first trained model LM1 by performing machine learning using, as training data, a data set of images of a plurality of objects 3, each having an unmachinable area unsuitable for cutting, and information on the unmachinable area set by the user in each of the images of the plurality of objects 3. When an image of the object 3 (before-processing image P10) is input, the first trained model LM1 determines the unmachinable area present in the before-processing image P10 and extracts a machineable area R1 excluding the unmachinable area.

[0070] The third learning unit 75 generates the third trained model LM3 by performing machine learning using, as training data, a data set of, for example, multiple processed images after cutting the object 3 and information on the quality of the processing judged by a user for each of the multiple processed images. When a processed image capturing the state after cutting the object 3 is input, the third trained model LM3 outputs the result of judging the quality of the processing. If the third trained model LM3 judges that the processing is poor, it may further output information on the part where the processing is judged to be poor.

[0071] The fourth learning unit 76 generates a fourth trained model LM4 by performing machine learning using as training data a data set of a plurality of images each showing the top surface and side surface of the object 3 and an image of the top surface portion detected by image processing for each of the plurality of images, or information on the image of the top surface portion set by the user. When an image showing the top surface and side surface of the object 3 is input, the fourth trained model LM4 outputs a partial image of the top surface.

[0072] The first, third, and fourth learning units 73, 75, and 76 train an artificial intelligence program (algorithm) using training data. The artificial intelligence program is a machine learning model, and for example, a neural network, which is a type of hierarchical model, is used. The first, third, and fourth learning units 73, 75, and 76 generate trained models by causing the neural network to perform machine learning (e.g., deep learning) using a training dataset. The first, third, and fourth learning units 73, 75, and 76 may also improve the performance of the trained models by re-training using a newly acquired training dataset.

[0073] It is not essential that the control device 60 includes the first, third, and fourth learning units 73, 75, and 76. The control device 60 may perform the inference phase using the first, third, and fourth trained models LM1, LM3, and LM4 created in an external system. It is also not essential that the first, third, and fourth trained models LM1, LM3, and LM4 are stored in the memory unit 62 of the control device 60; for example, the inference phase may be performed using the first, third, and fourth trained models LM1, LM3, and LM4 that exist on the cloud.

[0074] (3) Operation Hereinafter, the operation of the cutting system 1 according to this embodiment and the operation of the control device 60 as the control system 2 will be described with reference to Fig. 3. Note that the flowchart shown in Fig. 3 is merely an example of a control method for the cutting system 1 and the control device 60 according to this embodiment, and the order of processes may be changed as appropriate, and processes may be added or omitted as appropriate.

[0075] A user of the cutting system 1 or a robot that handles the object 3 to be cut or otherwise processed places the object 3 on the plate portion 51 in a state where the object 3 has been positioned by the positioning unit 52 of the base 50 .

[0076] When the object 3 is placed on the pedestal 50, the control device 60 transmits an imaging instruction to the imaging unit 20 and the robot unit 30 to have the imaging unit 20 capture an image of the object 3 before processing (step S1). In this embodiment, the imaging unit 20 captures images of the object 3 at each of a plurality of imaging positions, and the images captured at the plurality of imaging positions are combined to generate a single before-processing image. To this end, the control device 60 controls the robot unit 30 to move the imaging unit 20 to the plurality of imaging positions in a predetermined order (see FIG. 4). When the robot unit 30 moves the imaging unit 20 and the imaging unit 20 arrives at each of the plurality of imaging positions, the control device 60 controls the imaging unit 20 to capture an image of the object 3.

[0077] Once the imaging unit 20 has completed capturing multiple images (e.g., images P11 to P16 in FIG. 6) required to generate one unprocessed image, the correction unit 64 performs distortion correction on each of the multiple images to generate multiple corrected images (step S2).

[0078] Then, the image synthesis unit 65 synthesizes the plurality of corrected images to generate an unprocessed image P10 (see FIG. 6) in which the plurality of corrected images are joined at their boundaries (step S3). At this time, the output unit 70 may output the unprocessed image P10 synthesized by the image synthesis unit 65 to the display terminal 80, and the user of the cutting system 1 can check the unprocessed image P10 displayed on the display unit 82.

[0079] There is a possibility that not only the top surface 310 of the object 3 but also the side surface 320 and the like are reflected in the unprocessed image P10. Therefore, the top surface extraction unit 66 inputs the unprocessed image P10 into the fourth trained model LM4 to extract a partial image P20 (see FIG. 7) corresponding to the top surface 310 of the object 3 (step S4).

[0080] When the partial image P20 of the upper surface is extracted, the region extraction unit 67 extracts the machinable region R1 by inputting the partial image P20 of the upper surface into the first trained model LM1 (step S5). At this time, the output unit 70 may output the partial image P20 (see FIG. 8) on which a marker such as a dashed line indicating the machinable region R1 is displayed to the display terminal 80, and the user of the cutting system 1 can confirm the machinable region R1 based on the partial image P20 displayed on the display unit 82.

[0081] After extracting the machinable area R1, the nesting processing unit 68 performs calculations using a genetic algorithm based on the information on the machinable area R1 and the information on the workpiece 100 to create an arrangement pattern PT1 (see Figure 10) in which one or more workpieces 100 are arranged in the machinable area R1 (step S6).

[0082] After creating the arrangement pattern PT1, the machining control unit 71 creates control parameters to be used by a tool control program to move the cutting unit 10 to the machining positions of one or more workpieces 100 and perform cutting on the one or more workpieces 100, based on the arrangement pattern PT1. The control parameters used by the tool control program include, for example, information on the movement path along which the cutting unit 10 moves to trace the machining positions, and information on work parameters (such as the shape and depth of the cut) related to the work of cutting the object 3 at the machining positions. The machining control unit 71 controls the cutting unit 10 and the robot unit 30 by executing the tool control program using the created control parameters, and causes the cutting unit 10 to perform cutting on the object 3 at the machining positions of one or more workpieces 100 (step S7).

[0083] When the cutting process is completed, the control device 60 transmits an imaging instruction to the imaging unit 20 and the robot unit 30 to have the imaging unit 20 capture an image of the processed state of the object 3. At this time, the control device 60 controls the robot unit 30 to move the imaging unit 20 to a plurality of imaging positions in a predetermined order. When the robot unit 30 moves the imaging unit 20 and the imaging unit 20 arrives at each of the plurality of imaging positions, the control device 60 controls the imaging unit 20 to capture an image of the object 3.

[0084] When the imaging unit 20 has completed capturing a plurality of images required to generate one processed image, the correction unit 64 performs distortion correction on each of the plurality of images to generate a plurality of corrected images.

[0085] Then, the image synthesis unit 65 synthesizes the multiple corrected images to synthesize a processed image in which the multiple corrected images are joined at their boundaries (step S8). At this time, the output unit 70 may output the processed image synthesized by the image synthesis unit 65 to the display terminal 80, and the user of the cutting processing system 1 can check the processed image displayed on the display unit 82.

[0086] When the processed image is captured, the quality determination unit 69 inputs the processed image into the third trained model LM3 to determine whether the processing of the object 3 is good or bad (step S9). At this time, the output unit 70 outputs the determination result of the quality determination unit 69 to the display terminal 80 (step S10), and the user of the cutting processing system 1 can check the result of the quality determination of the processing displayed on the display unit 82.

[0087] (4) Variations The following are examples of modifications: The modifications described below can be applied in appropriate combination with the above-described embodiment.

[0088] The above embodiment is merely one of various embodiments of the present disclosure, and various modifications can be made to the above embodiment depending on the design and the like as long as the object of the present disclosure can be achieved.

[0089] Furthermore, functions similar to those of the cutting system 1 or the control system 2 may be embodied in a control method for the cutting system 1, a computer program, or a non-transitory recording medium on which a program is recorded. A control method for the cutting system 1 according to one embodiment is a control method for the cutting system 1 including a cutting unit 10 and an imaging unit 20, and includes an area extraction step and a machining control step. The cutting unit 10 has a cutting tool 11 that performs machining on the object 3. The imaging unit 20 captures an image of the object 3. In the area extraction step, a machineable area R1 on the surface of the object 3 where machining is possible is extracted by inputting the before-machining image obtained by imaging the object 3 with the imaging unit 20 into a first trained model LM1. In the machining control step, the cutting unit 10 is controlled to perform machining on the machineable area R1. A program according to one embodiment is a program for causing a computer system to execute the above control method.

[0090] The cutting system 1 and the control system 2 in the present disclosure each include a computer system. The computer system primarily comprises a processor and a memory as hardware. The functions of the cutting system 1 and the control system 2 in the present disclosure are realized by the processor executing a program stored in the memory of the computer system. The program may be pre-stored in the memory of the computer system, provided via a telecommunications line, or provided in a non-transitory recording medium readable by the computer system, such as a memory card, an optical disk, or a hard disk drive. The processor of the computer system is composed of one or more electronic circuits including a semiconductor integrated circuit (IC) or a large-scale integrated circuit (LSI). The integrated circuits, such as ICs and LSIs, are referred to by different names depending on the degree of integration, and include integrated circuits called system LSIs, very large-scale integrations (VLSIs), or ultra-large-scale integrations (ULSIs). Furthermore, field-programmable gate arrays (FPGAs), which are programmed after the LSI is manufactured, or logic devices capable of reconfiguring the connections within the LSI or the circuit partitions within the LSI, can also be used as processors. The electronic circuits may be integrated into one chip or distributed across multiple chips. The chips may be integrated into one device or distributed across multiple devices. The computer system referred to here includes a microcontroller having one or more processors and one or more memories. Therefore, the microcontroller is also composed of one or more electronic circuits including a semiconductor integrated circuit or a large-scale integrated circuit.

[0091] In the above embodiment, the nesting processing unit 68 creates the arrangement pattern PT1 by calculation processing using a genetic algorithm, but the arrangement pattern PT1 may also be created using a trained model. That is, the nesting processing unit 68 may create the arrangement pattern PT1 (see FIG. 10) that assigns processing positions for one or more workpieces 100 (see FIG. 9) cut out from the target object 3 by inputting an image of the processable region R1 in the before-processing image into the second trained model LM2.

[0092] The learning unit 72 also preferably includes a second learning unit 74 that generates the second trained model LM2. The second learning unit 74 generates the second trained model LM2 by performing machine learning using, for example, a data set of multiple images each including the machinable region R1, machining drawing data of the workpiece 100, and information on an arrangement pattern PT1 of the workpiece 100 set by the user for each of the multiple images as training data. When an image including the machinable region R1 (e.g., a partial image P20 of the upper surface) and information on the workpiece 100 are input, the second trained model LM2 outputs an arrangement pattern PT1 for arranging one or more workpieces 100 in the machinable region R1. The second learning unit 74 may retrain the second trained model LM2 using, as training data, a data set of the arrangement pattern PT1 created by the nesting processing unit 68 and the judgment result of the pass / fail judgment unit 69. Specifically, the second learning unit 74 may re-learn the placement pattern PT1 so as to reduce defective products by performing machine learning using the placement pattern PT1 and the results of the pass / fail judgment unit 69's judgment on the pass / fail of the processing of the object 3 that has been cut based on the placement pattern PT1 as training data.

[0093] The second learning unit 74 causes an artificial intelligence program (algorithm) to learn using teacher data. The artificial intelligence program is a machine learning model, and for example, a neural network, which is a type of hierarchical model, is used. The second learning unit 74 generates a trained model by causing the neural network to perform machine learning (for example, deep learning) using a training dataset. The second learning unit 74 may also improve the performance of the trained model by re-training using a newly acquired training dataset.

[0094] It is not essential that the control device 60 includes the second learning unit 74. The control device 60 may perform the inference phase using a second trained model LM2 created in an external system. It is also not essential that the second trained model LM2 be stored in the memory unit 62 of the control device 60, and the inference phase may be performed using a second trained model LM2 that exists on the cloud, for example.

[0095] Furthermore, it is not essential for the cutting system 1 that multiple functions are integrated into a single housing; the components of the cutting system 1 may be distributed across multiple housings. Furthermore, at least some of the functions of the cutting system 1, for example, some of the functions of the cutting system 1, may be implemented using the cloud (cloud computing) or the like. Similarly, it is not essential for the control system 2 that multiple functions are integrated into a single housing; the components of the control system 2 may be distributed across multiple housings. Furthermore, at least some of the functions of the cutting system 1 or the control system 2, for example, some of the functions of the control unit 63 of the control system 2, may be implemented using the cloud (cloud computing) or the like. For example, it is not essential for the control system 2 to include the first to fourth learning units 73 to 76; some or all of the first to fourth learning units 73 to 76 may be provided in an external device different from the control device 60. That is, the control system 2 may perform the inference phase using the first to fourth trained models LM1 to LM4 created in an external device. Furthermore, it is not essential for the control system 2 to hold the first to fourth trained models LM1 to LM4, and the inference phase may be performed using the first to fourth trained models LM1 to LM4 on the cloud.

[0096] (summary) The cutting system (1) of the first aspect includes a cutting unit (10), an imaging unit (20), and a control device (60). The cutting unit (10) has a cutting tool (11) that cuts an object (3). The cutting unit (10) cuts the object (3) using the cutting tool (11) under the control of the control device (60). The imaging unit (20) captures an image of the object (3). The control device (60) has an area extraction unit (67) and a processing control unit (71). The area extraction unit (67) inputs a before-processing image (P10) obtained by capturing an image of the object (3) with the imaging unit (20) into a first trained model (LM1), thereby extracting a machinable area (R1) on the surface of the object (3) where cutting processing is possible. The processing control unit (71) controls the cutting unit (10) to perform cutting in the machinable area (R1).

[0097] According to this aspect, the area extraction unit (67) extracts the machinable area (R1) by inputting the before-machining image (P10) of the object (3) captured by the imaging unit (20) into the first trained model (LM1), eliminating the need for a person to manually set the machinable area (R1) by looking at the object (3). This can improve the efficiency of cutting work by the cutting system (1).

[0098] In the cutting processing system (1) of the second aspect, the control device (60) in the first aspect further includes a nesting processing unit (68). The nesting processing unit (68) creates an arrangement pattern (PT1) that assigns processing positions of one or more workpieces (100) to be cut out from the target object (3) in the processable region (R1). The processing control unit (71) controls the cutting unit (10) to perform cutting at the processing positions based on the arrangement pattern (PT1) created by the nesting processing unit (68).

[0099] According to this embodiment, since there is no need for a person to set the arrangement pattern (PT1), the efficiency of the cutting work by the cutting system (1) can be further improved.

[0100] In the cutting processing system (1) of the third aspect, in the first or second aspect, the control device (60) further includes a first learning unit (73). The first learning unit (73) generates a first trained model (LM1) by performing machine learning using images of a plurality of objects (3) each having unmachinable areas unsuitable for cutting processing as training data.

[0101] According to this aspect, the first learning unit (73) performs machine learning to generate the first trained model (LM1), so that a more appropriate region can be extracted as the machineable region (R1).

[0102] In the cutting processing system (1) of the fourth aspect, in any one of the first to third aspects, the control device (60) further includes a quality determination section (69) and an output section (70). The quality determination section (69) determines the quality of processing of the object (3) based on a processed image obtained by the imaging unit (20) capturing an image of the object (3) after processing. The output section (70) outputs the determination result of the quality determination section (69).

[0103] According to this aspect, the user of the cutting system (1) can know whether the cutting is good or bad based on the determination result of the output unit (70).

[0104] In the cutting processing system (1) of the fifth aspect, in the second aspect, the control device (60) further includes a quality determination unit (69) and a second learning unit (74). The nesting processing unit (68) creates an arrangement pattern (PT1) by inputting an image of the machinable region (R1) in the before-machining image (P10) into the second trained model. The quality determination unit (69) determines the quality of the machining of the object (3) based on a processed image obtained by the imaging unit (20) capturing an image of the object (3) after machining. The second learning unit (74) re-trains the second trained model (LM2) using the arrangement pattern (PT1) created by the nesting processing unit (68) and the judgment result of the quality determination unit (69) as training data.

[0105] According to this aspect, the result of the quality determination unit (69) determining whether the processing of the object (3) is good or bad can be reflected in the creation of the placement pattern (PT1).

[0106] In the cutting processing system (1) of the sixth aspect, in the fourth aspect, the quality determination unit (69) determines the quality of the processing of the object (3) by inputting the processed image to the third trained model (LM3). The third trained model (LM3) is a trained model generated by machine learning using, as training data, good product data relating to the state after processing of the object (3) determined to be good and the processed image of the object (3) determined to be bad.

[0107] According to this aspect, the quality determining unit (69) can determine the quality of the machining using the third trained model (LM3).

[0108] The cutting processing system (1) of a seventh aspect is the same as any one of the first to sixth aspects, and further includes an image synthesis unit (65). The imaging unit (20) is attached to the movably provided cutting unit (10) and captures images of the object (3) at each of a plurality of imaging positions. The image synthesis unit (65) generates an unprocessed image (P10) by synthesizing a plurality of images obtained by the imaging unit (20) capturing images of the object (3) at each of the plurality of imaging positions.

[0109] According to this embodiment, even if the entire object (3) cannot be captured within the angle of view of one imaging unit (20), a pre-processing image (P10) of the entire object (3) can be obtained using one imaging unit (20), thereby reducing the number of imaging units (20) and simplifying the configuration of the cutting processing system (1).

[0110] The cutting processing system (1) of the eighth aspect is the seventh aspect, further including a correction unit (64) that performs distortion correction on each of a plurality of images obtained by the imaging unit (20) capturing images of the object (3) at each of a plurality of imaging positions. The image synthesis unit (65) generates an unprocessed image (P10) by synthesizing a plurality of corrected images after the distortion correction of the plurality of images has been performed by the correction unit (64).

[0111] According to this aspect, the correction unit (64) performs distortion correction, and the image synthesis unit (65) can generate an unprocessed image (P10) with less distortion.

[0112] In the cutting processing system (1) of the ninth aspect, in any one of the first to eighth aspects, the control device (60) further includes an upper surface extraction unit (66) that extracts a partial image (P20) corresponding to the upper surface of the object (3) from the before-processing image (P10). The area extraction unit (67) extracts a workable area (R1) from the partial image (P20) extracted by the upper surface extraction unit (66) in the before-processing image (P10).

[0113] According to this aspect, the area extraction unit (67) extracts the workable area (R1) from the partial image (P20) corresponding to the top surface of the object (3), and therefore the amount of processing can be reduced compared to when the workable area (R1) is extracted from an image that also includes the side surfaces other than the top surface.

[0114] In the cutting processing system (1) of the tenth aspect, in the ninth aspect, the upper surface extraction unit (66) extracts a partial image (P20) corresponding to the upper surface of the object (3) by inputting the before-processing image (P10) into the fourth trained model (LM4). The fourth trained model (LM4) is a trained model generated by performing machine learning using multiple images showing the upper and side surfaces of the object (3) as training data.

[0115] According to this aspect, the time required for the process of extracting the partial image (P20) corresponding to the upper surface can be reduced compared to when the partial image (P20) corresponding to the upper surface is extracted by image processing such as edge detection.

[0116] The control method of the eleventh aspect is a control method for a cutting processing system (1) including a cutting unit (10) and an imaging unit (20), and includes an area extraction step and a processing control step. The cutting unit (10) has a cutting tool (11) that performs cutting on an object (3). The imaging unit (20) captures an image of the object (3). In the area extraction step, a before-processing image (P10) obtained by capturing an image of the object (3) with the imaging unit (20) is input into a first trained model (LM1), thereby extracting a machinable area (R1) on the surface of the object (3) where cutting processing is possible. In the processing control step, the cutting unit (10) is controlled to perform cutting on the machinable area (R1).

[0117] According to this aspect, in the area extraction step, the before-machining image (P10) of the object (3) captured by the imaging unit (20) is input to the first trained model (LM1) to extract the machinable area (R1), so there is no need for a person to manually set the machinable area (R1) by looking at the object (3). Therefore, the work efficiency of the cutting process by the cutting process system (1) can be improved.

[0118] A program according to a twelfth aspect is a program for causing a computer system to execute the control method according to the eleventh aspect.

[0119] According to this aspect, the efficiency of the cutting work performed by the cutting system (1) can be improved.

[0120] The control system of the thirteenth aspect includes an area extraction unit (67) and a machining control unit (71). The area extraction unit (67) extracts a machineable area (R1) on the surface of the object (3) where cutting work is possible by inputting a before-machining image (P10) obtained by an imaging unit (20) capturing an image of the object (3) into a first trained model (LM1). The machining control unit (71) controls a cutting unit (10) having a cutting tool (11) that cuts the object (3) so as to cut the machineable area (R1).

[0121] According to this aspect, the area extraction unit (67) extracts the machinable area (R1) by inputting the before-machining image (P10) of the object (3) captured by the imaging unit (20) into the first trained model (LM1), eliminating the need for a person to manually set the machinable area (R1) by looking at the object (3). This can improve the efficiency of cutting work by the cutting unit (10).

[0122] Not limited to the above aspects, various configurations (including modified examples) of the cutting processing system (1) according to the embodiment can be embodied as a control method for the cutting processing system (1), a (computer) program, or a non-temporary recording medium on which a program is recorded, etc.

[0123] The configurations according to the second to tenth aspects are not essential for the cutting system (1) and can be omitted as appropriate. [Explanation of symbols]

[0124] 1. Cutting processing system 3. Object 10 Cutting unit 11 Cutting tools 20 Imaging unit 60 Control device 64 Correction unit 65 Image synthesis unit 66 Top surface extraction part 67 Region extraction part 68 Nesting processing section 69 Good / bad judgement section 70 Output section 71 Processing control section 73 First Study Section 74 Second Study Section LM1 First trained model LM2 Second trained model LM3 Third trained model LM4 Fourth trained model P10 Unprocessed image P20 Partial Image PT1 placement pattern R1 Machinable area

Claims

1. The cutting unit, the imaging unit, and the control device are provided. the cutting unit has a cutting tool for cutting an object, and cuts the object using the cutting tool under the control of the control device; the imaging unit captures a pre-machining image of the object before a first cutting process is performed using the cutting tool; The control device an area extraction unit that extracts a machineable area on the surface of the object, excluding an unmachinable area that is unsuitable for the first cutting process, by inputting the before-machining image obtained by the imaging unit capturing an image of the object into a first trained model; a machining control unit that controls the cutting unit to perform the first cutting process on the machineable area, Cutting processing system.

2. The unmachinable area includes at least one of an area where knots exist in the wood when the object is wood, an area where warping exists in the object, an area where unevenness exists in the object, an area where dirt exists in the object, an area where processing marks from a second cutting process performed before the first cutting process exist, and an area where a fixing member for fixing the object exists. The cutting system according to claim 1 .

3. The control device further has a nesting processing unit that creates an arrangement pattern that assigns processing positions of one or more workpieces to be cut out from the target object to the processable area, the machining control unit controls the cutting unit to perform the first cutting process at the machining position based on the arrangement pattern created by the nesting processing unit. The cutting system according to claim 1 or 2.

4. The control device further includes a first learning unit, The first learning unit generates the first trained model by performing machine learning using images of a plurality of the objects each having the unmachinable region as training data. The cutting system according to any one of claims 1 to 3.

5. The control device further includes a quality determination unit and an output unit, the quality determination unit determines quality of the processing of the object based on a processed image obtained by the imaging unit capturing an image of the object after the first cutting process is performed, and the output unit outputs the judgment result of the quality judgment unit. The cutting system according to any one of claims 1 to 4.

6. The control device further has a good / bad judgment unit and a second learning unit, the nesting processing unit creates the arrangement pattern by inputting an image of the machineable area in the before-machining image obtained by capturing an image of the object before the first cutting process is performed by the imaging unit into a second trained model; the quality determination unit determines quality of the processing of the object based on a processed image obtained by the imaging unit capturing an image of the object after the first cutting process is performed, and The second learning unit re-learns the second trained model using the arrangement pattern created by the nesting processing unit and the judgment result of the quality judgment unit as training data. The cutting system according to claim 3 .

7. The pass / fail judgment unit judges whether the processing of the object is good or bad by inputting the processed image into a third trained model generated by performing machine learning using good product data regarding the state of the object after processing that is judged to be good and the processed image of the object that is judged to be bad as training data. The cutting system according to claim 5 .

8. Further comprising an image synthesis unit, the imaging unit is attached to the movably provided cutting unit, and images the object at each of a plurality of imaging positions; the image synthesis unit generates the pre-processing image before the first cutting process is performed by synthesizing a plurality of images obtained by the imaging unit capturing images of the object before the first cutting process is performed at each of the plurality of imaging positions. The cutting system according to any one of claims 1 to 7.

9. The imaging unit further comprises a correction unit that performs distortion correction on each of the plurality of images obtained by imaging the object at each of the plurality of imaging positions, the image synthesis unit synthesizes a plurality of corrected images obtained by performing the distortion correction on the plurality of images by the correction unit, thereby generating the pre-processing image before performing the first cutting process. The cutting system according to claim 8.

10. The control device further includes an upper surface extraction unit that extracts a partial image corresponding to the upper surface of the object from the pre-processing image before the first cutting process is performed; the area extraction unit extracts the machineable area from the partial image extracted by the upper surface extraction unit in the before-machining image before the first cutting process is performed; The cutting system according to any one of claims 1 to 9.

11. The top surface extraction unit extracts the partial image corresponding to the top surface of the object by inputting the pre-processing image before the first cutting process into a fourth trained model generated by performing machine learning using multiple images showing the top and side surfaces of the object as training data. The cutting system according to claim 10.

12. A control method for a cutting processing system including a cutting unit and an imaging unit, comprising: the cutting unit has a cutting tool for cutting an object; the imaging unit images the object before a first cutting process is performed using the cutting tool; The control method includes: an area extraction step of extracting a machineable area on the surface of the object, excluding an unmachinable area that is unsuitable for the first cutting process, by inputting a pre-machining image obtained by imaging the object with the imaging unit before performing the first cutting process into a first trained model; a machining control step of controlling the cutting unit to perform the first cutting process on the machineable area, Control method.

13. The unmachinable area includes at least one of an area where knots exist in the wood when the object is wood, an area where warping exists in the object, an area where unevenness exists in the object, an area where dirt exists in the object, an area where processing marks from a second cutting process performed before the first cutting process exist, and an area where a fixing member for fixing the object exists. The control method according to claim 12.

14. A method for causing a computer system to execute the control method according to claim 12 or 13. program.

15. An area extraction unit that extracts a machineable area on the surface of the object, excluding unmachinable areas that are unsuitable for the first cutting process, by inputting a pre-processing image obtained by an imaging unit capturing an image of the object before the first cutting process into a first trained model; a machining control unit that controls a cutting unit having a cutting tool so as to perform the first cutting process on the machineable area, Control system.

16. The unmachinable area includes at least one of an area where knots exist in the wood when the object is wood, an area where warping exists in the object, an area where unevenness exists in the object, an area where dirt exists in the object, an area where processing marks from a second cutting process performed before the first cutting process exist, and an area where a fixing member for fixing the object exists.

16. The control system of claim 15.

Images