Work analysis device

Abstract

To provide a work analysis device which visualizes work to be improved and a degree or a direction of the improvement, and can increase efficiency and speed of the improvement of the work.SOLUTION: A work analysis device that extracts a feature amount of motion of a worker from a work moving picture, and evaluates improvement of work on the basis of the extracted feature amount includes: an analysis part for partitioning motions of a worker for each work, and extracting feature amounts of the motions at each partitioned work, and determining unevenness, waste and impossibility of the motions in each work, on the basis of the feature amounts; and a display part for simultaneously displaying any one of the unevenness, the waste and the impossibility together with the feature amounts in each work, on a standard work combination form where contents and orders of each work, and a standard time required for completing each work are aligned in time series.SELECTED DRAWING: Figure 5

Description

[Technical Field]

[0001] The present invention relates to an apparatus for analyzing manual work such as the production or assembly of various articles. [Background technology]

[0002] In order to improve work, it is necessary to understand the actual work being performed on-site, and for this purpose, it is common to photograph the work being performed on-site with a camera. One example is described in Patent Document 1. The system described in Patent Document 1 is configured to obtain, after production work is completed, images showing the work content of each work process in succession in the order in which the work processes are performed. The system is configured to record information obtained by a work recognition means that recognizes the work to be worked on and images obtained by a photographing means that photographs the work content, and to edit the images of the specified work to be linked in the order in which the work processes are performed based on the recorded information and the flow of the production process. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2003-150230 Summary of the Invention [Problem to be solved by the invention]

[0004] The system described in Patent Document 1 allows the user to check the work content of multiple work processes for a completed product using a series of images, facilitating defect analysis, design changes, improvements to production equipment, and monitoring of process status. However, the system described in Patent Document 1 only acquires a series of images of work across multiple processes, and is unable to identify or identify work that needs improvement, the degree of improvement, or the direction of improvement from the images. In other words, although the system described in Patent Document 1 makes it possible to reproduce information that was previously obtained by visual inspection or memorization by fixing it as an image, and furthermore, converts information from multiple processes into a series of information, it may not necessarily function effectively for work analysis and improvement.

[0005] The present invention has been made with an eye on the above-mentioned technical problems, and aims to provide a work analysis device that can visualize the work that needs to be improved, the degree of improvement, or the direction of improvement, thereby making work improvements more efficient or faster. [Means for solving the problem]

[0006] In order to achieve the above-mentioned object, the present invention provides a work analysis device that extracts feature amounts of worker movements from work videos and evaluates work improvement based on the extracted feature amounts, and is characterized by comprising: an analysis unit that classifies the worker movements into tasks and extracts feature amounts of the movements for each classified task, and determines unevenness, waste, and unreasonableness in the movements for each task based on the feature amounts; and a display unit that displays either the unevenness, waste, or unreasonableness together with the feature amounts for each task on a standard work combination sheet that chronologically lists the content and order of each task and the standard time required to complete each task.

[0007] In the present invention, the feature amount may include the skeletal coordinates of the worker obtained from the work video, the unevenness may be the amount of deviation from a predetermined standard value of the maximum, minimum and average values ​​of the time required from the start to the end of each of multiple times the work is performed, the waste may be the difference between the movement of the worker during the work and a predetermined standard movement, and the strain may be the difference between the posture of the worker during the work and a predetermined standard posture. [Effects of the Invention]

[0008] According to the present invention, the inconsistencies, waste, and unreasonableness of a worker's work can be determined from work videos taken of the worker while working. Therefore, the tracking of the worker's movements can be obtained automatically without manual intervention. Furthermore, at least one of the inconsistencies, waste, and unreasonableness of the movements is simultaneously displayed on the standard work combination sheet for each work. This allows the worker to easily grasp at a glance the tasks that need improvement, the direction or content of the improvement, and the degree of improvement, thereby enabling efficient and appropriate work improvement. Furthermore, the use of work videos allows for easy and quick tracking of repeated (or repetitive) work.

[0009] In the present invention, the worker's skeletal coordinates can be used to understand the worker's movements. Skeletal coordinates can be calculated, for example, as the position of each worker's joint and the angle of the lines connecting the joints. The time required for a task is calculated based on the skeletal coordinates, and the time is compared with a standard value to determine the deviation. In other words, the length of the task time for multiple tasks can be determined by comparing it with a standard value, allowing for quantification and understanding of inconsistencies. Similarly, waste and strain can be identified by comparing the worker's movements obtained from the skeletal coordinates with standard movements. Since the worker's movements and standard movements are image data obtained from a video of the task, waste and strain can be quantified and understood. In other words, inconsistencies, waste, and strain can be quantitatively and non-personally identified, allowing for efficient, rapid, and objective understanding of the task, thereby enabling improvements to be made. Furthermore, by performing a similar analysis after improvements are made, it becomes possible to easily and accurately determine the appropriateness of the improvements. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is a schematic diagram showing how a work video is obtained by filming the work state of a worker at a work site. [Figure 2] FIG. 2 is a block diagram showing the functional configuration of a controller. [Figure 3] FIG. 10 is a schematic diagram illustrating an example of a skeletal model of a worker in a bent position. [Figure 4] 1 is a flowchart illustrating an example of control of analysis of work and output of the analysis results in an embodiment of the present invention. [Figure 5] FIG. 10 is a diagram illustrating an example of a standard work combination chart. DETAILED DESCRIPTION OF THE INVENTION

[0011] Next, an embodiment of the present invention will be described with reference to the accompanying drawings. Note that the embodiment described below is merely an example of how the present invention can be implemented, and is not intended to limit the present invention.

[0012] The work analysis device according to an embodiment of the present invention is a device for analyzing the presence, content, and extent of unevenness, waste (hereinafter sometimes referred to as "muda"), and overreaching (hereinafter sometimes referred to as "muri") in multiple tasks performed consecutively by a worker. It is configured to capture video of the worker performing the task at a work site, acquire video of the task, and analyze the worker's movements based on the video image. FIG. 1 schematically illustrates an example of a work situation involving a worker 1, in which the worker 1 performs a task such as assembling parts onto a workpiece 3 transported by a conveyor or other transport equipment 2. Therefore, as indicated by the arrows in FIG. 1, the worker 1 faces the workpiece 3 on the transport equipment 2, faces a table 4 on which parts and tools are placed, or moves (walks) between the position on the transport equipment 2 and the position on the table 4. Naturally, the worker also changes the orientation of his or her body, including his or her head, from side to side, and even bends or stands up in relation to the workpiece 3 or parts on the table 4. The worker also moves his or her hands and arms up, down, and crosses them. Such movements of the worker 1 together with the work environment are photographed by a two-dimensional (2D) or three-dimensional (3D) camera 5 over a predetermined period of time.

[0013] A controller 6 is provided to process the work video captured by the camera 5. The controller 6 is mainly configured as a microcomputer consisting of a processor (CPU), memory elements (RAM, ROM), and an interface, and is configured to perform various processes using image data and data input from an input device 7, etc., and to output and display the results of the processing to an output device 8. Examples of the input device 7 include a keyboard and a mouse. Examples of the output device 8 include a display and a printer.

[0014] FIG. 2 is a block diagram showing the functional configuration of the controller 6. First, the controller 6 includes an input unit 6a that captures work video captured by the camera 5. The captured work video is footage of the work site captured over a predetermined period of time, and the movements of the worker 1 contained therein are continuous across a series of tasks. Therefore, an annotation unit 6b is provided to segment the movements of the worker 1 into individual tasks. One example of annotation is a process of segmenting the series of videos into segments corresponding to the start and end of each task performed by the worker 1, and also assigning task names to the segmented sections. This is performed by manually operating the input device 7, such as a keyboard or mouse.

[0015] The controller 6 includes a skeleton estimation unit 6c that processes the movements of the worker 1 captured as an image as skeletal coordinates. Specifically, this process creates a skeletal model by connecting the joints of the worker 1 with straight lines. For example, the positions of the shoulders, arms, and legs are expressed as coordinate values ​​in an orthogonal three-dimensional coordinate system with the waist joint as the origin. Alternatively, the positions and angles of the arms or hands are expressed as coordinate values ​​in the orthogonal three-dimensional coordinate system with the shoulder as the origin. The movements of the worker 1 are analyzed as movements of the skeletal model using these coordinate values. In FIG. 1, joints are indicated by "○" marks, and shoulders, elbows, wrists, waists or hip joints, knees, ankles, etc. are used as indicators. Skeletal coordinates are created by connecting these points. The positions and orientations of the joints and the directions of the lines connecting the joints allow the positions and orientations of the worker 1's body, arms, legs, etc., as well as the duration of a given posture, to be quantitatively determined. This allows the posture, movement, hand movements, and task of the worker 1 to be extracted as feature quantities.

[0016] The controller 6 is provided with a strain analysis unit 6d that detects or determines whether the extracted movements of the worker 1 are strained. An example of strained movements is the angle of the upper body or arms. For example, as shown in Figure 3, the angle θ between the vertical line and the line representing the upper body from the waist corresponds to the bending angle of the worker 1. The larger this angle θ, the greater the strain on the worker 1. Furthermore, depending on the type or content of the work, postures such as raising the arms up to near the shoulders and stretching them horizontally require a great deal of muscle strength, which can be said to increase the strain on the worker. The posture of the worker 1, which is one of the features obtained from such skeletal coordinates or skeletal data, is sequentially detected for each work and each repeated cycle, and compared with a predetermined standard posture to detect or determine whether "strain" exists. The standard posture is determined appropriately in advance depending on the content of each work, input via the input device 7, and stored.

[0017] To reduce the burden of repetitive tasks, it is preferable for worker 1 to perform tasks without inconsistency. Therefore, the controller 6 is provided with an inconsistency analysis unit 6e for detecting or assessing inconsistencies in work. Inconsistency is a state in which the time required to complete a specific task deviates from a predetermined standard time, or the amount of deviation becomes larger than expected, with each task being performed. The inconsistency analysis unit 6e extracts the timing of the start and end of worker 1's movements, which are classified by annotations in the work video, to calculate the task time, and then calculates the longest, shortest, and average task time for multiple tasks performed. If these values ​​deviate significantly from the standard time (standard value), it is judged to be "inconsistent." The amount of deviation required for this judgment can be determined in advance as appropriate depending on the content of each task, input via the input device 7, and stored.

[0018] The controller 6 also includes a waste analysis unit 6f. Here, "waste" refers to a motion or task performed by the worker 1 that does not create added value. For example, a time-consuming situation occurs when parts or tools to be attached to the workpiece 3 are located far away, as shown in Figure 1 as the worker facing the table 4. Also, situations that require the worker 1 to change his / her grip on parts or tools, or situations that require the worker 1 to significantly change his / her posture, such as turning around, are examples of "waste." The worker 1's movements (features) can be calculated for each task based on the skeletal coordinates. These movements are compared with pre-determined standard movements (standard actions) to detect or determine "waste." The standard movements can be input as data from the input device 7 to the controller 6 and stored. Alternatively, an image of the task can be played back and displayed on the output device 8. A designated evaluator (not shown) can visually evaluate the image and input the evaluation results via the input device 7.

[0019] The overload analysis section 6d, the unevenness analysis section 6e, and the waste analysis section 6f correspond to the analysis section in the embodiment of the present invention.

[0020] The controller 6 is provided with an analysis result output section 6g that outputs the so-called analysis results obtained by detection or determination by the overload analysis section 6d, unevenness analysis section 6e, and waste analysis section 6f.

[0021] Furthermore, the controller 6 is provided with an analysis result confirmation / correction unit 6h. The camera 5 continuously captures images of the work site for a preset period of time. However, disturbances, such as a line stop or slowdown for some reason, or a temporary delay or abnormal movement of the worker 1 due to poor health, may occur. Such disturbances are also captured in the work video. Since the analysis results output from the analysis result output unit 6g may contain results due to such disturbances, it is preferable to confirm the results and correct them, such as by removing their influence. The analysis result confirmation / correction unit 6h compares the output analysis results with disturbance information during the work that was the subject of the analysis, confirms the analysis results, and corrects them as necessary. For example, the confirmation and correction may be performed by inputting disturbance information into the controller 6 and automatically determining whether disturbance data exists for each work period. Alternatively, the analysis results displayed on the output unit 8 may be corrected by operating the input unit 7.

[0022] The controller 6 is provided with a display output unit 6i that outputs and displays the confirmed and corrected analysis results. The output form may be a display as an image on the output device 8 described above, or a color print. The analysis results of overload, unevenness, and waste are output so as to be displayed simultaneously with the standard work.

[0023] An example of control for analyzing work and outputting the analysis results by the controller 6 is shown in the flowchart of Figure 4. First, a work video is read (step S1). The work video is taken by the camera 5 described above, focusing on the worker 1 on-site. The work video may be read directly from the camera 5, or may be read from a work video that has been temporarily saved on a recording medium (not shown).

[0024] Next, annotation is performed (step S2). This is a manual operation, in which the movements of the worker 1 who is continuously performing predetermined tasks are divided into tasks while watching the video. Therefore, the time required for each divided task (task time) is determined. In addition, a skeletal model (or skeletal coordinates) of the worker is created by attaching indicators (marks) to parts of the video that correspond to the worker's joints.

[0025] By dividing the movements of the worker 1 into tasks and creating skeletal coordinates, feature amounts for each task are extracted (step S3). That is, the positions of the arms and hands, the bending angle of the upper body, the twisting of the body, the degree of arm extension, etc. are extracted. These feature amounts may be extracted automatically based on two-dimensional or three-dimensional images, or may be extracted manually while watching a video.

[0026] Based on the extracted feature quantities, the system analyzes overburden, waste, and unevenness (step S4). Unevenness is analyzed based on the maximum, minimum, and average values ​​of the work time for multiple tasks, so the analysis is performed on the work that follows the multiple tasks for which these measurements were obtained. For overburden and waste, standard operations (or reference values) that serve as the basis for judgment are prepared and stored in advance, so the work for all cycles is automatically analyzed.

[0027] The analysis results thus obtained are output (step S5), and are then confirmed and corrected as necessary (step S6). This is the process performed by the analysis result confirmation and correction unit 6h, and its necessity and content are as described above. The confirmed and corrected analysis results are then simultaneously displayed and output on the standard work combination chart (step S7).

[0028] An example of a standard work combination chart that simultaneously displays analysis results is shown in Figure 5. In the format shown in Figure 5, the task details (tasks A-I) and the actions (walking) that worker 1 should perform are displayed chronologically from top to bottom on the left side. The standard manual work time for each task is displayed to the right, and the walking time is displayed further to the right of that. Further to the right of these item columns is the so-called line diagram showing the task time. This task time column is graduated in 0.5-second increments, with elapsed time increasing to the right. The standard time for each task and walking is shown with a horizontally oriented thick line. Walking is shown with a wavy line sloping downward to the right, and tasks for which worker 1's posture or movement, such as excessive strain or waste, are the subject of analysis are shown with a horizontally oriented wavy line. Of course, the display format shown here is merely an example; the line thickness, shape, color, and orientation may be set as appropriate. Furthermore, any display may be selected, such as displaying symbols in addition to lines.

[0029] The analysis results are displayed simultaneously as follows. First, for unevenness, the maximum, minimum, and average times measured over multiple tasks are displayed numerically. To visualize the extent of the unevenness, a line (a short line in Figure 5 ) connecting the maximum and minimum values ​​is displayed parallel to the end of the thick line representing the standard time. In this case, if the difference between the maximum and minimum values ​​exceeds a predetermined tolerance, or if the maximum or average value exceeds the tolerance from the standard time, a so-called warning message is displayed to indicate an abnormality or to indicate that improvement should be considered. The warning message may be displayed in any manner, such as a different color from other displays or a special mark. A voice message alerting the user to the problem may also be output by hovering the cursor over or clicking the message on the output device 8. Note that a "☆" is indicated in Figure 5 .

[0030] Since overstress is not judged by time but by posture, arm angle, and other factors, it cannot be displayed as time in the work time column. Similarly, since waste is judged based on the posture and hand position of worker 1, it cannot be displayed as time in the work time column. Therefore, in the example shown in Figure 5, a thick horizontal line is displayed below the wavy line indicating standard work, and these lines are used as warning signs C1 and C2. Note that this warning sign may be a different color from the other signs. Special symbols may also be displayed alongside it. Furthermore, the degree of waste or overstress may be indicated by line thickness, color, or shade. Similarly to the warning sign for unevenness, a voice alert may be output by hovering the cursor over or clicking on the sign on the output device 8. Note that the standard work combination table shown in Figure 5 can simultaneously display appropriate information as needed. For example, so-called good work without unevenness, overstress, or waste may be visualized (or displayed as a graphic).

[0031] As described above, the analysis device according to the embodiment of the present invention determines the characteristics of worker 1 and the resulting inconsistencies, waste, and overstress in the worker's movements (tasks) based on the characteristics based on the video of the work performed on-site. The analysis and evaluation results are then simultaneously displayed on a standard work combination sheet, allowing the presence or absence of inconsistencies, waste, and overstress for each task to be clearly identified, and the extent of these inconsistencies and overstress can also be determined for each task. Therefore, the analysis device according to the present invention can identify the need for improvements in work performed on the production site, the items to be improved, the direction of improvement, and the degree of improvement, and can then promptly implement improvements, thereby improving not only work efficiency but also the work environment by reducing the burden on workers. Furthermore, the analysis device according to the present invention can automatically perform most of the analysis of worker 1's movements and comparison and analysis with standard work using electronic devices such as cameras and computers. This allows for rapid and efficient analysis of work, reducing the variance in analysis results, enabling stable and accurate analysis, and further reducing the labor required for work analysis.

[0032] The above-described task analysis and display of the analysis results can be performed for each worker 1. Therefore, the standard task combination chart shown in FIG. 5 , which simultaneously displays the analysis results, is unique to the worker 1 who performed the analysis. In other words, even if the tasks to be performed are the same, different workers 1 will obtain different analysis results. The analysis results displayed on each standard task combination chart include both those due to the individual characteristics of the worker 1 and those due to the work environment. Therefore, comparing the standard task combination charts for each worker 1 will reveal tasks that are both considered to have either unevenness, overburden, or waste, as well as tasks in which one worker 1 has either unevenness, overburden, or waste but the other worker has none of these. If the analysis results are consistent, the analysis results can be considered to be due to the work environment. Tasks with different analysis results are considered to reflect the individual characteristics of the worker 1 (such as physique or experience). In this way, the analysis device of the present invention can analyze and identify the need for improvements to the work environment, or the need for replacement or training of the worker 1. In this respect, too, it becomes possible to improve the work and environment at the production site quickly and efficiently.

[0033] The analysis results obtained using the standard work combination chart can be compared not only between workers 1 but also by comparing the standard work combination charts before and after the improvement. That is, when an improvement is made to a task, any of the unevenness, overburden, or waste in the improved task should change, so comparing the standard work combination charts before and after the improvement makes it possible to determine whether the expected change has occurred. If the expected change is observed, the task improvement has been carried out appropriately. However, if unevenness, overburden, or waste still remains, it indicates that there is room for further improvement in the work procedures, work environment, etc. In this way, the analytical device of the present invention can be used not only to analyze unevenness, overburden, and waste in tasks and to determine whether the associated improvements are necessary, but also to determine the appropriateness of the implemented improvements, or whether they are excessive or insufficient.

[0034] It should be noted that the present invention is not limited to the above-described embodiment, and for example, instead of extracting feature amounts from work images, sensors may be attached to on-site workers and signals from the sensors may be used to extract feature amounts of the worker's movements or feature amounts from skeletal coordinates. Furthermore, the standard work combination sheet is not limited to the format shown in Fig. 5 above, as long as it can simultaneously display, for each work task, standard amounts (standard values) such as work time and posture, and any of unevenness, overburden, and waste based on the actually measured work time and posture. [Explanation of symbols]

[0035] 1. Worker 2. Conveying equipment 3 Work 4 tables 5. Camera 6 Controller 6a Input section 6b Annotation section 6c Skeleton estimation section 6d Muri Analysis Department 6e Mura analysis department 6F Waste Analysis Department 6g Analysis result output section 6h correction part 6i display output section 7 Input Device 8 Output Device

Claims

1. A work analysis device that extracts feature amounts of worker movements from a work video and evaluates work improvement based on the extracted feature amounts, an analysis unit that classifies the worker's movements into tasks, extracts feature values ​​of the movements for each task, and determines unevenness, waste, and strain in the movements for each task based on the feature values; a display output unit that simultaneously displays any of the unevenness, waste, and unreasonableness along with the characteristic amount of each task on a standard task combination sheet in which the content and order of each task and the standard time required to complete each task are arranged in chronological order; A work analysis device comprising:

2. The work analysis device according to claim 1 , the feature amount includes skeletal coordinates of the worker obtained from the work video, The unevenness is the amount of deviation of the maximum, minimum, and average values ​​of the time required from the start to the end of each of the multiple times the work is performed from a predetermined standard value, The waste is a difference between the movement of the worker during the work and a predetermined standard movement, The strain is a difference between the worker's posture during the work and a predetermined standard posture. A work analysis device characterized by:

Images