Work analysis apparatus and work analysis method

Abstract

To suppress the occurrence of malfunctions caused by human error. [Solution] The work analysis device comprises an analysis unit that analyzes the actual work content of workers based on the motion data of workers in a sewing factory detected by a detection device, a regular data acquisition unit that acquires regular work content, and a recognition unit that recognizes the difference between the actual work content and the regular work content.

Description

Technical Field

[0001] The technology disclosed in this specification relates to a work analysis device and a work analysis method.

Background Art

[0002] As disclosed in Patent Document 1, when producing sewn products in a sewing factory, industrial sewing machines are used. The industrial sewing machines are operated by workers.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In a sewing factory, the work to be performed by a worker may be predetermined. Due to human error, there is a possibility that the worker may not perform the required work. If the required work is not performed, problems may occur such as having to return the sewing target from the subsequent process to the previous process for re-sewing or defective sewn products flowing into the market.

[0005] The technology disclosed in this specification aims to suppress the occurrence of problems caused by human error.

Means for Solving the Problems

[0006] This specification discloses a work analysis device. The work analysis device includes an analysis unit that analyzes the actual work content of a worker based on the operation data of the worker in a sewing factory detected by a detection device, a normal data acquisition unit that acquires the normal work content that is the normal work content, and a recognition unit that recognizes the difference between the actual work content and the normal work content.

Effects of the Invention

[0007] The technology disclosed herein suppresses the occurrence of malfunctions caused by human error. [Brief explanation of the drawing]

[0008] [Figure 1] Figure 1 is a schematic diagram showing a sewing management system according to an embodiment. [Figure 2] Figure 2 is a schematic perspective view showing the relationship between the sewing machine, the operator, and the camera according to this embodiment. [Figure 3] Figure 3 is a hardware configuration diagram showing a controller according to an embodiment. [Figure 4] Figure 4 is a functional block diagram showing a sewing management system according to an embodiment. [Figure 5] Figure 5 is a diagram illustrating the method for generating a trained model according to the embodiment. [Figure 6] Figure 6 is a diagram illustrating how to use the trained model according to the embodiment. [Figure 7] Figure 7 is a diagram illustrating the method for obtaining normalized data according to the embodiment. [Figure 8] Figure 8 shows an example of the standard work procedures and actual work procedures according to the embodiment. [Figure 9] Figure 9 shows an example of output data output to the monitor according to this embodiment. [Figure 10] Figure 10 shows an example of output data output to the monitor according to the embodiment. [Figure 11] Figure 11 shows an example of output data output to the monitor according to the embodiment. [Figure 12] Figure 12 is a flowchart showing the work analysis method according to the embodiment. [Modes for carrying out the invention]

[0009] The embodiments will be described below with reference to the drawings. In the embodiments, the terms left, right, front, back, top, and bottom will be used as appropriate to describe the positional relationships of each part. These terms indicate relative positions or directions with respect to the center of the sewing machine 2.

[0010] [Sewing Management System] Figure 1 is a schematic diagram showing a sewing management system 1 according to an embodiment. The sewing management system 1 manages sewing equipment. Sewing equipment is a series of machines necessary for producing sewn products in a sewing factory. Examples of sewing equipment include sewing machines, fabric inspection machines, shrinking machines, fabric spreading machines, cutting machines, bonding machines, and pressing machines. In this embodiment, the sewing equipment includes a sewing machine 2. The sewing management system 1 manages the sewing machine 2. The sewing machine 2 is an industrial sewing machine that operates in a sewing factory. Multiple sewing machines 2 are installed in the sewing factory. For the sake of simplicity, only one sewing machine 2 is shown in Figure 1. Sewn products are produced when the material to be sewn is sewn by the sewing machine 2. Examples of sewing machines 2 installed in a sewing factory include a lockstitch sewing machine for straight stitching the material to be sewn, an overlock sewing machine for overcasting the edges of the material to be sewn, a flatlock sewing machine for flat stitching the material to be sewn, and a bar tack sewing machine for bar tack stitching the material to be sewn.

[0011] In a sewing factory, worker WR performs their duties. Worker WR operates sewing equipment. In this embodiment, worker WR operates sewing machine 2. Sewing machine 2 comprises a sewing machine frame 3, a needle bar 4, a balance wheel 5, a needle plate 6, a sewing machine motor 7, and a controller 9.

[0012] The needle bar 4 holds the sewing machine needle 4N. The needle bar 4 moves back and forth in the vertical direction. The needle bar 4 is supported by the sewing machine frame 3. The sewing machine needle 4N has a thread hole through which the upper thread passes. The sewing machine needle 4N holds the upper thread on the inner surface of the thread hole. As the needle bar 4 moves back and forth in the vertical direction, the sewing machine needle 4N moves back and forth in the vertical direction while holding the upper thread.

[0013] The thread guide 5 supplies the upper thread to the sewing needle 4N. The thread guide 5 is supported by the sewing machine frame 3. The thread guide 5 has a holding hole through which the upper thread passes. The thread guide 5 holds the upper thread on the inner surface of the holding hole. The thread guide 5 reciprocates in the vertical direction while holding the upper thread. By reciprocating in the vertical direction, the thread guide 5 pays out the upper thread used for sewing the sewing target or pulls up the upper thread. The tension of the upper thread supplied to the sewing needle 4N is adjusted by a thread conditioner (not shown) supported by the sewing machine frame 3. The thread conditioner applies tension to the upper thread supplied to the sewing needle 4N.

[0014] The needle plate 6 supports the sewing target from below. The needle bar 4 is arranged above the needle plate 6. The sewing needle 4N held by the needle bar 4 and the needle plate 6 face each other. The sewing target supported by the needle plate 6 is pressed from above by a presser (not shown). The needle plate 6 has a needle hole through which the sewing needle 4N can pass. The sewing needle 4N passing through the sewing target supported by the needle plate 6 passes through the needle hole. On the needle plate 6, feed teeth (not shown) for feeding the sewing target supported by the needle plate 6 in the feed direction are arranged.

[0015] A bobbin (not shown) is arranged below the needle plate 6. The bobbin supplies the lower thread to the sewing target. The bobbin is rotatable. The sewing machine 2 sews the sewing target by the cooperation of the sewing needle 4N held by the needle bar 4 and the bobbin.

[0016] The sewing machine motor 7 generates power for operating the needle bar 4, the thread guide 5, and the bobbin. Due to the power generated by the sewing machine motor 7, the needle bar 4 and the thread guide 5 reciprocate in the vertical direction. The bobbin rotates so that the lower thread is supplied to the sewing target in synchronization with the reciprocating movement of the needle bar 4 in the vertical direction. The sewing machine 2 sews the sewing target by the cooperation of the sewing needle 4N held by the needle bar 4 and the bobbin.

[0017] The controller 9 includes a computer. The controller 9 outputs control commands to control the sewing machine 2. The controller 9 outputs control commands to drive at least the sewing machine motor 7. The controller 9 includes a sewing machine drive circuit for driving the sewing machine motor 7. The sewing machine drive circuit includes a switching element such as a field-effect transistor (FET). The sewing machine motor 7 is driven based on the drive current supplied from the sewing machine drive circuit.

[0018] The sewing management system 1 manages the sewing machines 2 and the workers WR in the sewing factory. The sewing management system 1 includes a server 10, a camera 12, a monitor 22, and a reader 23.

[0019] Server 10 includes a computer. Server 10 communicates with the controller 9 of the sewing machine 2 via a communication system 13. Server 10 may be located outside the sewing factory or located inside the sewing factory. Examples of the communication system 13 include a local area network (LAN), the internet, a mobile phone network, and a satellite network.

[0020] Camera 12 captures images of workers WR in a sewing factory. Camera 12 captures images of workers WR operating sewing equipment. Camera 12 captures images of workers WR operating sewing machine 2. Camera 12 captures images of sewing machine 2 operated by workers WR. Camera 12 is an example of a detection device that outputs motion data indicating the actions of workers WR. The motion data includes images captured by camera 12. The motion data includes the actions of workers WR captured by camera 12. The motion data includes the actions of sewing machine 2 captured by camera 12. The images captured by camera 12 include video. An RGB camera is given as an example of camera 12. Camera 12 may also be an RGB-D camera or a hyperspectral camera.

[0021] Controller 9 acquires the operation data of the worker WR and the operation data of the sewing machine 2 captured by camera 12. Controller 9 acquires the captured images taken by camera 12. The captured images taken by camera 12 include images of the worker WR and images of the sewing machine 2. Controller 9 transmits the captured images taken by camera 12 to server 10.

[0022] The monitor 22 includes an output device that provides output data to the operator WR. Examples of output devices include a display device and an audio output device. The display device provides display data to the operator WR as output data. The audio output device provides audio data to the operator WR as output data. The display device includes a flat panel display such as a liquid crystal display or an organic EL display. The audio output device includes a buzzer. The audio data includes a buzzer sound (warning sound). The monitor 22 may also include an input device such as a touch panel.

[0023] Reader 23 is a barcode reader that reads barcodes. The barcode may be a one-dimensional barcode or a two-dimensional barcode. As will be described later, worker WR is provided with a barcode containing regular data indicating the regular work content that worker WR should perform. Worker WR has reader 23 read the barcode.

[0024] [The relationship between sewing machines, workers, and cameras] Figure 2 is a schematic perspective view showing the relationship between the sewing machine 2, the operator WR, and the camera 12 according to this embodiment. As shown in Figure 2, the sewing machine 2 is installed on the upper surface of the sewing machine table 14. The sewing machine 2 is installed in the center of the upper surface of the sewing machine table 14 in the left-right direction. The sewing machine 2 is installed in the center of the upper surface of the sewing machine table 14 in the front-back direction.

[0025] The sewing machine table 14 is supported by table legs. A chair 15 is placed behind the sewing machine table 14. The operator WR operates the sewing machine 2 while seated in the chair 15. A pedal 16 is located below the sewing machine table 14. The pedal 16 is operated by the operator WR. The operator WR operates the pedal 16 with their foot. The sewing machine motor 7 is driven or stopped based on the operation state of the pedal 16.

[0026] Multiple cameras 12 are arranged. In one embodiment, cameras 12 are arranged on both the left and right sides of the sewing machine 2. Camera 12 includes a first camera 12L arranged on the left side of the sewing machine 2 and a second camera 12R arranged on the right side of the sewing machine 2.

[0027] The camera 12 is supported by a support member 17. The support member 17 is fixed to the sewing machine table 14. The support member 17 protrudes upward from the top surface of the sewing machine table 14. The camera 12 is supported by the sewing machine table 14 via the support member 17. The support member 17 includes a first support member 17L fixed to the left end of the sewing machine table 14 and a second support member 17R fixed to the right end of the sewing machine table 14.

[0028] In this embodiment, the camera 12 is covered by a cover 18. The cover 18 protects the camera 12. The cover 18 is fixed to a support member 17. The cover 18 includes a first cover 18L that covers the first camera 12L and a second cover 18R that covers the second camera 12R.

[0029] A support column 19 is fixed to the front right end of the top surface of the sewing machine table 14. A light fixture 20 is positioned at the top end of the support column 19. A thread stand 21 is positioned in the center of the support column 19. The light fixture 20 illuminates the sewing machine 2 and the hands of the operator WR. The light fixture 20 illuminates the top surface of the sewing machine table 14. The thread stand 21 supports the thread 24 wound on the spool.

[0030] The monitor 22 is positioned on the top surface of the sewing machine table 14. The monitor 22 may be integrated with the sewing machine 2. The monitor 22 may be attached to the sewing machine frame 3. The leader 23 is positioned on the top surface of the sewing machine table 14. The leader 23 may be fixed to the sewing machine table 14. The leader 23 may be a portable leader.

[0031] Camera 12 is positioned above the top surface of the sewing machine table 14. Camera 12 captures images of the operator WR and the sewing machine 2. Camera 12 captures images of the hands of operator WR operating the sewing machine 2. Camera 12 captures images of objects placed on the top surface of the sewing machine table 14. Camera 12 captures images of the thread stand 21 and the thread 24 attached to the thread stand 21. Camera 12 captures images of the object to be sewn placed on the needle plate 6. The first camera 12L captures images of the sewing machine 2 and operator WR from the left side. The second camera 12R captures images of the sewing machine 2 and operator WR from the right side.

[0032] Note that the number and position of the cameras 12 described with reference to Figure 2 are just examples. There may be one camera 12, or any number of three or more cameras 12. The cameras 12 may be fixed to, for example, the sewing machine frame 3. The cameras 12 may also image the sewing machine 2 and the worker WR from above.

[0033] [controller] Figure 3 is a hardware configuration diagram showing a controller 9 according to an embodiment. The controller 9 includes a computer. The controller 9 has a processor 31 such as a CPU (Central Processing Unit), a main memory 32 including non-volatile memory such as ROM (Read Only Memory) and volatile memory such as RAM (Random Access Memory), a storage 33, an input / output interface 34 including input / output circuits, and a communication interface 35 including communication circuits. The functions of the controller 9 are stored in the storage 33 as a computer program 36. The processor 31 reads the computer program 36 from the storage 33, loads it into the main memory 32, and executes processing according to the computer program 36. The computer program 36 may be distributed to the server 10 via a network.

[0034] Server 10 includes a computer. Like controller 9, server 10 also has a processor 31, main memory 32, storage 33 for storing computer programs 36, an input / output interface 34, and a communication interface 35.

[0035] Figure 4 is a functional block diagram showing a sewing management system 1 according to an embodiment. The sewing management system 1 comprises a controller 9, a server 10, a camera 12, a monitor 22, and a reader 23. In this embodiment, the server 10 functions as a work analysis device.

[0036] The sewing machine motor 7, camera 12, monitor 22, and leader 23 are each connected to the controller 9. The controller 9 and the server 10 communicate via the communication system 13.

[0037] Server 10 has multiple functional units. The functions of the functional units of Server 10 are performed by the processor 31 and the storage 33. The functional units of Server 10 include an operation data acquisition unit 41, an analysis unit 42, a regular data acquisition unit 43, a recognition unit 44, an output unit 45, a control unit 46, and a storage unit 47.

[0038] The motion data acquisition unit 41 acquires motion data of worker WR in the sewing factory, as captured by the camera 12. The motion data acquisition unit 41 acquires motion data of worker WR operating the sewing machine 2, as captured by the camera 12. The motion data acquisition unit 41 acquires motion data of the sewing machine 2 operated by worker WR, as captured by the camera 12. The motion data includes the movements of worker WR and the movements of the sewing machine 2. The motion data includes images captured by the camera 12. The motion data acquisition unit 41 acquires images captured by the camera 12. The images captured by the camera 12 include video. The motion data acquisition unit 41 acquires images of worker WR and images of the sewing machine 2 via the communication system 13.

[0039] The analysis unit 42 analyzes the work content of worker WR based on the motion data of worker WR acquired by the motion data acquisition unit 41. The motion data of worker WR includes images of worker WR captured by the camera 12. The analysis unit 42 analyzes the work content of worker WR based on the images of worker WR operating the sewing machine 2. In this embodiment, analyzing the work content of worker WR includes analyzing the actual work content of worker WR. The analysis unit 42 analyzes the actual work content of worker WR based on the images of worker WR operating the sewing machine 2. The work content of worker WR includes multiple work items. The work content of worker WR includes identifying the work items performed by worker WR.

[0040] Examples of work items include adjustment work, which involves adjusting the sewing machine 2's adjustment items, including adjusting the thread tension using the thread tension control; picking up the object to be sewn before sewing; sewing the object using the sewing machine 2; checking the object after sewing by visually inspecting it; placing the object after sewing; changing the thread 24 on the thread stand 21; and transferring the object to the next process.

[0041] The analysis unit 42 analyzes the work content of worker WR based on an artificial intelligence (AI) algorithm. A pre-trained model is stored in the memory unit 47, which takes captured images of worker WR operating the sewing machine 2 as input and outputs the work content of worker WR. The analysis unit 42 analyzes the work content of worker WR using the pre-trained model stored in the memory unit 47.

[0042] Figure 5 is a diagram illustrating a method for generating a trained model according to an embodiment. The trained model is trained to identify the work performed by worker WR based on training images of worker WR operating the sewing machine 2. In the embodiment, the trained model is trained to identify at least the work items performed by worker WR based on training images of worker WR operating the sewing machine 2. In the embodiment, the trained model is generated by deep learning. As shown in Figure 5, during the training phase of deep learning, multiple training images generated by capturing the training work of worker WR operating the sewing machine 2 are input to the learner 40.

[0043] The training image may be generated by camera 12 capturing the training work of operator WR, or by a camera other than camera 12 capturing the training work of operator WR. The learning device 40 may be a functional unit of server 10, or it may be implemented by a computer different from server 10.

[0044] The learning device 40 learns the feature quantities of the instruction tasks contained in the training images. The learning device 40 learns from multiple training images and extracts the feature quantities of the instruction tasks contained in the training images. Based on the feature quantities of the instruction tasks, the learning device 40 takes an image of the operator WR operating the sewing machine 2 as input and generates a trained model that outputs the operator WR's work content. The trained model generated by the learning device 40 is stored in the memory unit 47.

[0045] The learning device 40 may generate a trained model using supervised machine learning. In supervised machine learning, each of the multiple training images is assigned a correct label indicating the content of the training work of the operator WR. The learning device 40 may use supervised machine learning on the training images to generate a trained model that takes images of operator WR operating the sewing machine 2 as input and outputs the content of operator WR's work.

[0046] Figure 6 is a diagram illustrating how to use the trained model according to the embodiment. The memory unit 47 stores a trained model that has been trained to identify the work content of worker WR based on training images of worker WR operating the sewing machine 2. The analysis unit 42 uses the trained model stored in the memory unit 47 to determine the work content of worker WR. The motion data acquisition unit 41 acquires images of worker WR operating the sewing machine 2. The images are generated when the camera 12 captures the actual work of worker WR operating the sewing machine 2. The analysis unit 42 inputs the images generated by capturing the actual work of worker WR into the trained model to determine the actual work content of worker WR. As the images generated by capturing the actual work of worker WR are input into the trained model, the actual work content, which is the actual work content of worker WR, is output from the trained model.

[0047] Note that the training image does not have to be an image of worker WR operating sewing machine 2. The trained model may be trained to identify the work performed by worker WR based on an artificially created training image of the worker. Examples of artificially created training images of the worker include images generated by image generation AI, images generated by computer graphics (CG), and illustrations.

[0048] Furthermore, the trained model may be trained to identify the operator WR's tasks based on, for example, an image of sewing machine 2, an image of the object being sewn, and text explaining what it means to operate sewing machine 2. After training the model with the concepts of sewing machine 2 and the object being sewn using images of sewing machine 2 and the object being sewn, a trained model that can identify the operator WR's tasks can be generated by adding text explanations that describe what it means to operate sewing machine 2 and sew, thereby enabling the model to understand its operation and process.

[0049] Furthermore, the trained model may be generated without using training images of worker WR operating sewing machine 2. The trained model may be trained to identify worker WR's work content based on training images of worker WR when not operating sewing machine 2. In sewing factories, worker WR often operates sewing machine 2 for 20% to 30% of the day. Therefore, to identify worker WR's work content, a trained model can be generated using training images of worker WR when not operating sewing machine 2. Examples of worker W's work content other than operating sewing machine 2 include setting aside items to be sewn, transporting items after sewing, meetings, and breaks. In other words, by using a model trained on images of worker WR when not operating sewing machine 2, the work content of worker WR other than operating sewing machine 2 can be identified with high accuracy, and thus work other than operating sewing machine 2 becomes the operation of sewing machine 2. Therefore, training images of worker WR operating sewing machine 2 are not necessarily used in the generation of the trained model.

[0050] The regular data acquisition unit 43 acquires regular data indicating the regular work content that the worker WR should perform. The regular data includes regular work content, which is the regular work content. The regular data acquisition unit 43 acquires regular work content.

[0051] Figure 7 is a diagram illustrating the method for acquiring regular data according to the embodiment. A specification sheet 26 is provided to the operator WR who operates the sewing machine 2. The specification sheet 26 is provided to the operator WR together with the object to be sewn 25. The specification sheet 26 is a document that describes the regular work to be performed by the operator WR.

[0052] Specification 26 specifies the quantity of sewing targets 25 that worker WR is to sew as part of the regular work procedures.

[0053] Specification 26 lists the standard materials to be used for sewing with sewing machine 2 as part of the standard work procedures. The standard materials include the thread to be used for sewing and the item to be sewn 25. In the example shown in Figure 7, the color and thickness of the thread to be used are listed as matters concerning the standard materials. The color and type of the fabric to be sewn are also listed as matters concerning the standard materials. Operator WR operates sewing machine 2 so that the standard item to be sewn is sewn with the standard thread.

[0054] Specification 26 lists the standard work items that worker WR should perform. Multiple standard work items are listed. In the example shown in Figure 7, there are four standard work items. The standard work items include work item A, work item B, work item C, and work item D. Worker WR performs the work according to the standard work items.

[0055] The standard work procedure includes the sequence of multiple work items that worker WR should perform. In the example shown in Figure 7, the standard work sequence specifies that the work should be performed in the order of work item A, work item B, work item C, and work item D. That is, work item A is performed first, followed by work item B, then work item C, and then work item D. Worker WR performs the work according to the standard work sequence.

[0056] A barcode 27 is included in specification 26. Barcode 27 contains regular data indicating the regular work content that worker WR should perform. In the embodiment, barcode 27 includes regular material data indicating regular materials, regular work item data indicating regular work items, and regular work sequence data indicating regular work sequence.

[0057] Operator WR has the reader 23 read barcode 27. By reading barcode 27, the reader 23 obtains regular data including regular work details. The regular data read by the reader 23 is transmitted to the server 10 via the communication system 13. The regular data acquisition unit 43 obtains regular work details, which are the regular work details, via the communication system 13.

[0058] The worker WR performs the work after having the barcode 27 read by the reader 23. The worker WR is captured by the camera 12 while performing the work. The analysis unit 42 analyzes the actual work content of the worker WR based on the image of the worker WR captured by the camera 12. The analysis unit 42 analyzes the work items actually performed by the worker WR based on the image of the worker WR. The analysis unit 42 analyzes the sequence of work actually performed by the worker WR based on the image of the worker WR. The analysis unit 42 analyzes the actual materials actually used for sewing by the sewing machine 2 based on the image of the worker WR, the sewing machine 2, and the area around the sewing machine 2. The actual materials include the thread 24 attached to the thread stand 21. The actual materials include the sewing target 25 placed on the needle plate 6. The analysis unit 42 analyzes the color and thickness of the thread 24 attached to the thread stand 21. The analysis unit 42 analyzes the color and type of fabric that is the sewing target 25 placed on the needle plate 6. The analysis unit 42 analyzes the combination of the fabric 25, which is the object to be sewn, placed on the needle plate 6, and the thread 24 attached to the thread stand 21.

[0059] The recognition unit 44 compares the actual work content analyzed by the analysis unit 42 with the regular work content acquired by the regular data acquisition unit 43. The recognition unit 44 recognizes the difference between the actual work content and the regular work content. The recognition unit 44 recognizes the difference between the actual work items and the regular work items. The recognition unit 44 recognizes the difference between the actual work sequence and the regular work sequence. The recognition unit 44 recognizes the difference between the actual material and the regular material. Actual work items refer to the work items actually performed by worker WR as analyzed by the analysis unit 42. Actual work sequence refers to the work sequence actually performed by worker WR as analyzed by the analysis unit 42. Actual material refers to the material actually used for sewing by sewing machine 2 as analyzed by the analysis unit 42.

[0060] Figure 8 shows an example of the standard work content and actual work content according to the embodiment. In the example shown in Figure 8, the standard work content to be performed by worker WR includes work item A, work item B, work item C, and work item D. In other words, there are four standard work items. The standard work content also includes a standard work sequence that specifies that the work should be performed in the order of work item A, work item B, work item C, and work item D.

[0061] If the analysis unit 42 determines that all four work items were actually performed according to the correct work sequence, as shown in the actual work content of Pattern 1 in Figure 8, the recognition unit 44 recognizes that there is no difference between the actual work content and the correct work content. The recognition unit 44 recognizes that there is no difference between the actual work items and the correct work items. The recognition unit 44 recognizes that there is no difference between the actual work sequence and the correct work sequence. The recognition unit 44 determines that the actual work content performed by worker WR is acceptable (OK).

[0062] As shown in the actual work content of Pattern 2 in Figure 8, due to human error, it is possible that work item C, for example, may not be performed among the four work items. The analysis unit 42 can analyze whether work item C has not been performed based on the image of worker WR captured by the camera 12. If the analysis unit 42 determines that work item C has not been performed, the recognition unit 44 recognizes that there is a difference between the actual work item and the regular work item. The recognition unit 44 recognizes that there is a omission in the actual work item compared to the regular work item. The recognition unit 44 recognizes that there was a work omission for work item C. The recognition unit 44 determines that the actual work content by worker WR is unacceptable (NG).

[0063] As shown in the actual work content of Pattern 3 in Figure 8, there is a possibility that the four work items may not be performed in the correct work order due to human error. The analysis unit 42 can analyze the actual work order based on the image of worker WR captured by the camera 12. If all four work items are performed, but the analysis unit 42 determines that they were not performed in the correct work order, the recognition unit 44 recognizes that there is a difference between the actual work order and the correct work order. The recognition unit 44 recognizes that although there were no omissions in the work items, the work was not performed in the correct work order. The recognition unit 44 determines that the actual work content by worker WR is unacceptable (NG).

[0064] Furthermore, while the regular work content considers the regular work items, the regular work order does not necessarily need to be considered. As shown in the actual work content of Pattern 4 in Figure 8, even if the four work items are not performed according to the regular work order, if the analysis unit 42 determines that all four work items have been performed, the recognition unit 44 recognizes that there is no difference between the actual work items and the regular work items. The recognition unit 44 recognizes that there were no omissions in the work items. The recognition unit 44 determines that the actual work content by worker WR is acceptable (OK).

[0065] The output unit 45 generates output data to be output to the monitor 22. The output data generated in the output unit 45 is transmitted to the controller 9 via the communication system 13. The controller 9 outputs the output data to the monitor 22.

[0066] Figures 9, 10, and 11 are diagrams showing examples of output data output to the monitor 22 according to the embodiment. The monitor 22 includes a display device and an audio output device. Figures 9, 10, and 11 each show examples of display data being shown on the monitor 22 as output data.

[0067] Figure 9 shows the monitor 22 before the worker WR begins work. When the barcode 27 is read by the reader 23, the regular data acquisition unit 43 acquires the regular work details. After the regular work details are acquired from the regular data acquisition unit 43, the output unit 45 displays the data shown in Figure 9 on the monitor 22 before the analysis unit 42 begins analyzing the actual work details of the worker WR.

[0068] As shown in Figure 9, the output unit 45 generates correlation data 50 as output data, which shows the relationship between regular work items and work items that were actually performed. The controller 9 displays the correlation data 50 on the monitor 22. In the example shown in Figure 9, the correlation data 50 includes symbols 51 that are displayed to correspond to each of the multiple regular work items. The display form of the symbols 51 is changed based on whether or not the worker WR has performed the work items. The symbols 51 include a first symbol 51A whose display form is changed based on whether or not work item A has been performed, a second symbol 51B whose display form is changed based on whether or not work item B has been performed, a third symbol 51C whose display form is changed based on whether or not work item C has been performed, and a fourth symbol 51D whose display form is changed based on whether or not work item D has been performed. Before the work is started (before the analysis unit 42 starts analyzing the actual work content of worker WR), work items A, B, C, and D have not been performed. The symbols 51 are displayed on the monitor 22 to indicate that the work items have not been performed.

[0069] After the operator WR makes the barcode 27 read by the reader 23, the operation is started. The operator WR performing the operation is imaged by the camera 12. The analysis unit 42 analyzes the actual operation content, which is the actual operation content of the operator WR, based on the captured image of the operator WR captured by the camera 12. The analysis unit 42 analyzes whether the operator WR is working in accordance with the regular operation content based on the captured image of the operator WR. The analysis unit 42 analyzes whether the operator WR has executed each of the operation item A, operation item B, operation item C, and operation item D based on the captured image of the operator WR. The analysis unit 42 recognizes the operation items actually executed by the operator WR among the regular operation items based on the captured image of the operator WR. The correlation data 50 displayed on the monitor 22 shows the relationship between the regular operation items and the operation items actually executed by the operator WR. The output unit 45 changes the display form of the symbol 51 corresponding to the operation items actually executed by the operator WR among the plurality of regular operation items.

[0070] In the example shown in FIG. 9, each of the operation item A, operation item B, operation item C, and operation item D has not been executed yet. Therefore, the four symbols 51 (51A, 51B, 51C, 51D) are displayed in the first display form. The display form includes colors. For example, the symbol 51 shown in FIG. 9 is displayed in the first color. The display form may include characters. The symbol 51 shown in FIG. 9 includes the character "not yet".

[0071] FIG. 10 shows the monitor 22 when all of the four regular operation items (operation item A, operation item B, operation item C, and operation item D) are executed and it is recognized by the recognition unit 44 that there is no difference between the actual operation content and the regular actual operation content. In the example shown in FIG. 10, since all of the operation item A, operation item B, operation item C, and operation item D have been executed, the four symbols 51 (51A, 51B, 51C, 51D) are displayed in the second display form. The symbol 51 shown in FIG. 10 is displayed in the second color. The symbol 51 shown in FIG. 10 includes the character "completed".

[0072] If the recognition unit 44 recognizes that all four regular work items have been executed, the output unit 45 generates completion data 52 indicating that all regular work items have been executed. The controller 9 displays the completion data 52 on the monitor 22. In the example shown in Figure 10, the completion data 52 includes character data indicating that all regular work items have been completed.

[0073] Figure 11 shows the monitor 22 when the recognition unit 44 recognizes a difference between the actual work content and the regular work content. The analysis unit 42 can identify that worker WR is performing a task different from the regular work content based on the image of worker WR captured by camera 12. Worker WM may start another task due to human error even though the regular work content has not been completed. If worker WM starts another task without performing, for example, task item C, one of the four regular work items, the analysis unit 42 can analyze that task item C has not been performed based on the image of worker WR captured by camera 12. If the analysis unit 42 determines that task item C has not been performed, the recognition unit 44 recognizes a difference between the actual work item and the regular work item.

[0074] If the recognition unit 44 detects a discrepancy between the actual work content and the standard work content, the output unit 45 generates warning data 53 indicating that there is a discrepancy between the actual work content and the standard work content. The controller 9 displays the warning data 53 on the monitor 22. The warning data 53 includes character data indicating that the standard work content has not been completed.

[0075] When the recognition unit 44 recognizes that the execution of work item C is missing for a regular work item, the output unit 45 causes the third symbol 51C corresponding to work item C to be displayed in the third display form. The third symbol 51C is displayed on the monitor 22 so as to indicate that work item C has not been executed. The third symbol 51C is displayed in, for example, the third color. The third symbol 51C shown in FIG. 11 includes the character "not yet". In the example shown in FIG. 11, since work items A, B, and D have been executed by the worker WR, each of the first symbol 51A, the second symbol 51B, and the fourth symbol 51D is displayed in the first display form.

[0076] When the recognition unit 44 recognizes that there is a difference between the actual work content and the regular work content, the output unit 45 may generate voice data as warning data 53. As the voice data, a buzzer sound indicating that there is a difference between the actual work content and the regular work content may be output from the monitor 22.

[0077] When the recognition unit 44 recognizes that there is a difference between the actual work content and the regular work content, the control unit 46 outputs a stop signal to stop the sewing machine 2. The stop signal output from the control unit 46 is transmitted to the controller 9 via the communication system 13. The controller 9 stops the sewing machine 2 based on the stop signal output from the control unit 46. Stopping the sewing machine 2 includes stopping the sewing machine motor 7.

[0078] The stop signal includes a prohibition signal for prohibiting the driving of the sewing machine 2. Even if the worker WR attempts to drive the sewing machine 2 for another work in spite of the difference between the actual work content and the regular work content, the driving of the sewing machine 2 is prohibited by the stop signal output from the control unit 46.

[0079] [Work analysis method] Figure 12 is a flowchart illustrating the work analysis method according to the embodiment. Below, an example is described in which the difference between the actual work content and the regular work content is analyzed in a conversion operation in which the sewing targets of the sewing machine 2 are changed from a first lot (first type) to a second lot (second type). A lot refers to the smallest unit of sewing targets 25 that are sewn under the same conditions. Thread 24 corresponding to the sewing targets 25 is attached to the thread stand 21. When changing from the first lot to the second lot, the thread 24 attached to the thread stand 21 is changed. For example, if the sewing targets 25 of the first lot are made of black fabric, black thread is attached to the thread stand 21. If the sewing targets 25 of the second lot are made of blue fabric, the thread 24 attached to the thread stand 21 is changed from black thread to blue thread. The recognition unit 44 analyzes the difference between the actual work content and the regular work content of the worker WR in the conversion operation in which the sewing targets 25 of the sewing machine 2 are changed from a first lot to a second lot.

[0080] After the work on the first lot is completed, the specification sheet 26 for the second lot, along with the sewing targets 25 for the second lot, is provided to worker WR. Worker WR has the reader 23 read the barcode 27 written on the specification sheet 26.

[0081] The reader 23 reads the barcode 27 to obtain regular data including the regular work content. The regular data read by the reader 23 is transmitted to the server 10 via the communication system 13. The regular data acquisition unit 43 obtains the regular work content via the communication system 13 (step S1).

[0082] The standard work procedures include multiple standard work items. The standard work items for the transition from the first lot to the second lot include the thread replacement work, which involves replacing the thread 24 mounted on the thread stand 21. When a first thread stand 21 for mounting the upper thread and a second thread stand 21 for mounting the lower thread are provided, the thread replacement work includes the upper thread replacement work and the lower thread replacement work. The standard work items include the upper thread replacement work item and the lower thread replacement work item.

[0083] The worker WR starts the conversion operation after having the barcode 27 read by the reader 23. The worker WR performing the conversion operation is imaged by the camera 12. The thread stand 21 is positioned within the field of view of the camera 12. The camera 12 can image the thread stand 21. The image of the worker WR captured by the camera 12 is transmitted to the server 10 via the communication system 13. The operation data acquisition unit 41 acquires the image of the worker WR via the communication system 13 (step S2).

[0084] A pre-trained model is generated in advance and stored in the memory unit 47, taking the captured images of the worker WR as input and the work content of the worker WR as output. The analysis unit 42 inputs the captured images of the worker WR into the pre-trained model and determines the actual work content of the worker WR (step S3).

[0085] The recognition unit 44 recognizes the difference between the regular work content acquired in step S1 and the actual work content determined in step S3 (step S4). The recognition unit 44 determines whether or not there is a difference between the regular work content and the actual work content. The recognition unit 44 determines whether or not the operator WM performed the upper thread replacement work item and the lower thread replacement work item (step S5).

[0086] In step S5, if it is determined that there is no difference between the actual work content and the regular work content (step S5: No), that is, if it is determined that the upper thread replacement work item and the lower thread replacement work item have been executed, the output unit 45 generates completion data 52 indicating that all of the regular work content has been executed. The completion data 52 generated by the output unit 45 is output from the monitor 22 (step S6). The operator WR can then start sewing the second lot of sewing targets 25 using the sewing machine 2.

[0087] In step S5, if a discrepancy is detected between the actual work content and the standard work content (step S5: Yes), that is, if it is determined that one or both of the upper thread replacement work items and the lower thread replacement work items are not performed, the output unit 45 generates warning data 53 indicating that the standard work content has not been completed. The warning data 53 generated by the output unit 45 is output from the monitor 22. The control unit 46 also outputs a stop signal to stop the sewing machine 2. The stop signal prevents the sewing machine 2 from being driven (step S7).

[0088] In step S3, the analysis unit 42 may analyze the actual material based on the image of the object placed on the upper surface of the sewing machine table 14 captured by the camera 12. The analysis unit 42 may also analyze the thread 24 attached to the thread stand 21. The recognition unit 44 may recognize whether the actual thread 24 attached to the thread stand 21 is the genuine thread (blue thread) from the second lot. If a difference is recognized between the actual thread 24 and the genuine thread, warning data 53 may be output from the monitor 22, or a stop signal to stop the sewing machine 2 may be output.

[0089] In step S3, the analysis unit 42 may analyze the combination of the fabric, which is the sewing target 25, placed on the needle plate 6, and the thread 24 attached to the thread stand 21. The recognition unit 44 may recognize whether the actual combination of the fabric, which is the sewing target 25, placed on the needle plate 6, and the thread 24 attached to the thread stand 21 is the correct combination of the second lot (the combination of blue fabric and blue thread). If a difference is recognized between the actual combination and the correct combination, warning data 53 may be output from the monitor 22, or a stop signal to stop the sewing machine 2 may be output.

[0090] The positions of the needle plate 6 and the thread stand 21 are fixed. The position of the camera 12 is also fixed. That is, the relative positions of the camera 12, the needle plate 6, and the thread stand 21 do not change. Therefore, if a camera coordinate system is defined for the image captured by the camera 12, the coordinates of the needle plate 6 and the thread stand 21 in the image do not change. The analysis unit 42 may set a predetermined window that includes the coordinates of the needle plate 6 and identify the actual color of the fabric, which is the sewing target 25, placed in the window. Similarly, the analysis unit 42 may set a predetermined window that includes the coordinates of the thread stand 21 and identify the actual color of the thread 24 placed in the window.

[0091] [effect] As described above, in this embodiment, the server 10 includes an analysis unit 42 that analyzes the actual work content of worker WR based on the captured image, which is the motion data of worker WR operating the sewing machine 2 captured by the camera 12; a regular data acquisition unit 43 that acquires regular work content, which is the regular work content; and a recognition unit 44 that recognizes the difference between the actual work content and the regular work content.

[0092] According to this embodiment, the recognition unit 44 recognizes the difference between the actual work content and the regular work content, thereby suppressing the occurrence of work omissions by the worker WR. Because the recognition unit 44 recognizes the difference between the actual work content and the regular work content, if some work items in the regular work content are not performed, measures can be taken to perform the omitted work items before the next work content is performed. As a result, the return of the sewing object 25 from the subsequent process to the preceding process for re-sewing and the outflow of defective sewn products to the market are suppressed, thereby suppressing the occurrence of defects caused by human error.

[0093] In one embodiment, the storage unit 47 stores a trained model that has been trained to identify the work performed by the worker WR. The analysis unit 42 inputs the captured image of the worker WR operating the sewing machine 2 into the trained model and can determine the actual work performed by the worker WR.

[0094] In this embodiment, correlation data 50 showing the relationship between regular work items and work items actually performed is displayed on the monitor 22. By checking the correlation data 50 displayed on the monitor 22, the worker WR can confirm the regular work items and the work items actually performed. When all regular work items have been completed, completion data 52 indicating that all regular work items have been completed is displayed on the monitor 22. The display of completion data 52 on the monitor 22 allows the worker WR to recognize that the necessary work has been completed.

[0095] If worker WR starts another task before all regular work items are completed, warning data 53 indicating that all regular work items have not been completed is output from monitor 22. Based on warning data 53, worker WR can recognize the existence of missing work items. Worker WR can then complete the missing work items. If worker WR starts another task before all regular work items are completed, the operation of sewing machine 2 is prohibited or stopped. This allows worker WR to recognize the existence of missing work items.

[0096] [Other embodiments] In the embodiment described above, the regular data acquisition unit 43 acquires regular data, including the regular work content, from the reader 23 that reads the barcode 27. For example, if the monitor 22 includes an input device, the worker WR may input the regular data by operating the input device of the monitor 22. The regular data acquisition unit 43 may acquire the regular data, including the regular work content, from the input device.

[0097] In the above-described embodiment, some or all of the functional parts of the server 10 may be provided in the controller 9. The controller 9 may analyze the work content of the worker WR based on the captured images of the worker WR. The server 10 may be omitted.

[0098] In the embodiments described above, the output data explained with reference to Figures 9, 10, and 11 may be displayed on a monitor connected to the server 10, for example. That is, the output data may be output from a monitor located remotely from the sewing machine 2.

[0099] In the above embodiment, the detection device for detecting the worker WR's movements is the camera 12. The worker WR's movement data is the image of the worker WR captured by the camera 12. The detection device for detecting the worker WR's movements may also be a thermograph that detects the worker WR's temperature (body temperature), for example.

[0100] In the above-described embodiment, the work analysis device analyzes the work content of worker WR operating sewing machine 2. The work analysis device may also analyze the work content of worker WR operating sewing equipment other than sewing machine 2. Examples of sewing equipment other than sewing machine 2 include fabric inspection machines, shrinking machines, fabric spreading machines, cutting machines, bonding machines, and pressing machines. Furthermore, the work analysis device may not only analyze the work content of worker WR in the sewing process using a sewing machine, but also analyze the work content of worker WR in all work processes performed in a sewing factory, such as ironing and bonding processes.

[0101] A trained model for identifying the work performed by worker WR may be generated based on training images of workers such as those cutting fabric, printing on fabric (e.g., transferring logos), marking on fabric (e.g., marking sewing lines), ironing, inspecting, and folding and packaging finished products.

[0102] Furthermore, by training the model to identify the work performed by worker WR based on a training image of a worker operating sewing machine 2 and at least one training image of a worker cutting fabric, a worker printing on fabric (such as transferring a logo), a worker marking fabric (such as marking sewing lines), a worker ironing, a worker inspecting, and a worker folding and packaging the finished product, the work performed by worker WR can be identified with high accuracy.

[0103] Furthermore, an image of a robot, such as that disclosed in Japanese Patent Publication No. 2021-147721, may be used as a training image for the worker performing the ironing or the worker folding the finished product. [Explanation of Symbols]

[0104] 1...Sewing management system, 2...Sewing machine, 3...Sewing machine frame, 4...Needle bar, 4N...Sewing machine needle, 5...Balance, 6...Needle plate, 7...Sewing machine motor, 9...Controller, 10...Server, 12...Camera, 12L...First camera, 12R...Second camera, 13...Communication system, 14...Sewing machine table, 15...Chair, 16...Pedal, 17...Support member, 17L...First support member, 17R...Second support member, 18...Cover, 18L...First cover, 18R...Second cover, 19...Support column, 20...Lighting, 21...Thread stand, 22...Monitor, 23...Leader, 24...Thread, 25...Sewing object 26...Specifications, 27...Barcode, 31...Processor, 32...Main memory, 33...Storage, 34...Input / Output interface, 35...Communication interface, 36...Computer program, 40...Learner, 41...Operation data acquisition unit, 42...Analysis unit, 43...Regular data acquisition unit, 44...Recognition unit, 45...Output unit, 46...Control unit, 47...Storage unit, 50...Correlation data, 51...Symbol, 51A...First symbol, 51B...Second symbol, 51C...Third symbol, 51D...Fourth symbol, 52...Completion data, 53...Warning data, WR...Operator.

Claims

1. An analysis unit analyzes the actual work content of the workers in a sewing factory based on the motion data of the workers detected by the detection device, A regular data acquisition unit that acquires regular work content, which is the regular work content, The system includes a recognition unit that recognizes the difference between the actual work content and the regular work content. Work analysis device.

2. The detection device includes a camera, The aforementioned operation data includes an image of the worker captured by the camera, The work analysis apparatus according to claim 1.

3. The system includes a storage unit that stores a trained model that has been trained to identify the work content of the aforementioned worker, The analysis unit inputs the captured images of the worker into the trained model to determine the actual work performed. The work analysis apparatus according to claim 2.

4. The aforementioned work includes multiple work items, The recognition unit recognizes the difference between the actual work item and the regular work item. The work analysis apparatus according to claim 1.

5. The aforementioned work content includes the sequence of work for multiple work items, The recognition unit recognizes the difference between the actual work sequence and the standard work sequence. The work analysis apparatus according to claim 1.

6. The aforementioned worker operates a sewing machine in the sewing factory, The analysis unit analyzes the actual work content based on the motion data of the operator operating the sewing machine. The work analysis apparatus according to claim 1.

7. The aforementioned work includes materials used in the sewing operation of the sewing machine, The recognition unit recognizes the difference between the actual material and the standard material. The work analysis apparatus according to claim 6.

8. The aforementioned sewing machine has a thread stand, The material includes the thread attached to the thread stand. The work analysis apparatus according to claim 7.

9. The aforementioned material includes a combination of the object to be sewn and the thread. The work analysis apparatus according to claim 7.

10. The recognition unit analyzes the difference between the actual work content and the regular work content in the conversion operation of changing the sewing target of the sewing machine from the first lot to the second lot. The work analysis apparatus according to claim 6.

11. The aforementioned work includes multiple work items, The analysis unit recognizes the work items that have been performed from among the regular work items, It includes an output unit that generates correlation data showing the relationship between regular work items and completed work items. The work analysis apparatus according to claim 1.

12. The output unit, when it recognizes that all regular work items have been executed, generates completion data indicating that all regular work items have been executed. The work analysis apparatus according to claim 11.

13. If the aforementioned discrepancy is detected, warning data is generated. Equipped with an output section, The work analysis apparatus according to claim 1.

14. The aforementioned worker operates sewing equipment in the sewing factory, If the aforementioned difference is determined to exist, the system includes a control unit that outputs a stop signal to stop the sewing machine. The work analysis apparatus according to claim 1.

15. Based on the motion data of workers in the sewing factory detected by the detection device, the actual work content of the workers is analyzed, To obtain the official work details, This includes recognizing the difference between the actual work content and the standard work content, Work analysis method.

Images