Work support system

Abstract

The present application provides a technology capable of estimating excessive load applied to a worker's body with high precision. A work support system includes: one or more acquisition units that acquire one or more measurement values corresponding to a worker's movement; an input unit that inputs one or more personal parameters representing the worker's physical characteristics; an output unit that outputs an estimation result; and a control unit that includes correction information for correcting values of a plurality of evaluation indexes representing the degree of load on the worker's body in work performed based on the one or more measurement values according to each of the one or more personal parameters, acquires the measurement values from the one or more acquisition units, determines the values of the plurality of evaluation indexes based on the acquired measurement values, refers to the correction information based on the input personal parameters, thereby correcting the values of the plurality of evaluation indexes, and estimates a load risk.

Description

Technical Field

[0001] This invention relates to a job support system. Background Technology

[0002] Previously, as described in Patent Document 1, there is a known technique for estimating worker fatigue by analyzing the posture or movements of workers during an operation.

[0003] Patent Document 1: Japanese Patent Application Publication No. 2024-5974 Summary of the Invention

[0004] The inventors have developed a universal job support system applicable to various tasks. Furthermore, the inventors aim to develop a technique within this job support system that accurately estimates the excessive load exerted on the worker's body. Similar to the technology in Patent Document 1, this job support system estimates the load on the worker's body caused by the task by analyzing the worker's posture or movements.

[0005] However, in order to improve estimation accuracy, if the system is adjusted according to specific job content or conditions, there is a risk of losing versatility. Therefore, job support systems require a technology that can maintain versatility while accurately estimating the excessive load acting on the worker's body.

[0006] The present invention can be implemented in the following manner.

[0007] (1) According to one aspect of the present invention, a job support system comprises: one or more acquisition units that acquire one or more measurement values ​​corresponding to the actions of a worker; an input unit that inputs one or more personal parameters representing the physical characteristics of the worker; an output unit that outputs an estimation result of the job support system; and a control unit that controls the job support system, the control unit having correction information for correcting the values ​​of a plurality of evaluation indicators representing the degree of load on the worker's body during work based on the one or more measurement values, the one or more acquisition units acquiring the one or more measurement values ​​and determining the values ​​of the plurality of evaluation indicators based on the acquired one or more measurement values, correcting the values ​​of the plurality of evaluation indicators by referring to the correction information based on the one or more personal parameters pre-input by the input unit, and estimating a load risk as a risk caused by the load based on the corrected values ​​of the plurality of evaluation indicators.

[0008] In this approach, the load risk acting on the worker is estimated based on the value of a modified evaluation index. The modified evaluation index reflects individual parameters. Since these individual parameters represent physical characteristics, the estimated load risk reflects the worker's physical characteristics. Even for the same job, the degree of load that constitutes a risk varies depending on the worker's physical characteristics. However, by estimating load risk based on the worker's physical characteristics, the job support system can estimate the likelihood of excessive load acting on the worker's physical condition with higher accuracy compared to methods that do not consider the worker's physical characteristics.

[0009] (2) In the above-described job support system, the control unit may include: a machine learning model, which has been trained with the values ​​of the multiple evaluation indicators and the more than one personal parameter related to multiple disease experiencers who have experienced diseases caused by the job performed by the operator, and the control unit may use the machine learning model to estimate the load risk.

[0010] By configuring it in this way, the job support system can accurately estimate the high probability of workload risk that may arise from performing the same work as someone with a history of illness.

[0011] (3) In the above-described job support system, the more than one personal parameter may include at least one of the worker's height, weight and grip strength.

[0012] By configuring it in this way, the job support system can estimate the likelihood of load risk based on the physical condition of the workers.

[0013] (4) In the above-described operation support system, the one or more acquisition units may include: an inertial measurement device, which is installed on the worker's work clothes to acquire acceleration and angular velocity corresponding to changes in the worker's posture; and a pressure sensor, which is installed on the worker's gloves to acquire pressure values ​​applied to the fingers.

[0014] By designing it in this way, the job support system can easily estimate the load risk for jobs involving whole-body movements and jobs involving fingertip movements. Therefore, this approach can further improve the effectiveness of the job support system of the present invention, which takes into account the physical characteristics of the worker when estimating load risk.

[0015] (5) In the above-described operation support system, the plurality of evaluation indicators may include a pressure accumulation indicator calculated by accumulating the pressure values.

[0016] By setting it up in this way, the job support system can easily estimate the risk of diseases caused by repetitive manual work. Attached Figure Description

[0017] Figure 1 This is an explanatory diagram showing the general structure of the work support system.

[0018] Figure 2 This is a flowchart illustrating a method for estimating the load risk caused by the job support system. Detailed Implementation

[0019] A. Implementation Method 1:

[0020] A-1. System Composition:

[0021] Figure 1 This is an explanatory diagram showing the general structure of the work support system 10. The work support system 10 estimates the workload risk of the worker WK. For example, the work support system 10 estimates the workload risk for a worker WK who manufactures vehicles.

[0022] Overload risk refers to the risk caused by the physical strain exerted on a worker's body during work. More specifically, overload risk is the risk that the accumulation of physical burden caused by work will adversely affect health or work efficiency. Examples of overload risks include excessive fatigue caused by prolonged work, and illnesses caused by improper movements or repetitive work. In this embodiment, examples of overload risks include lower back pain or trigger finger. Trigger finger is a type of tenosynovitis caused by repetitive manual work or excessive stress on the fingers. Overload risk is not limited to disease-related risks; it can also include risks such as overexertion or excessive stress.

[0023] The operation support system 10 has multiple acquisition units 100, input units 200, output units 400 and control units 300.

[0024] Multiple acquisition units 100 acquire multiple measurement values ​​corresponding to the actions of the worker WK. These multiple measurement values ​​are used to determine the values ​​of multiple evaluation indicators representing the degree of load exerted on the worker WK's body during work. Furthermore, the evaluation indicators represent the degree of load exerted on the worker WK's body during work based on the multiple measurement values. The multiple evaluation indicators include, for example, work posture indicators and stress accumulation indicators.

[0025] The occupational posture index is a scoring system that evaluates the workload of the entire body during work. Occupational posture indices are determined, for example, based on posture data collected experimentally to improve the accuracy of workload risk estimation and statistical methods. Occupational posture indices consider the frequency of postures and movements of the trunk or arms, and are expressed as scores ranging from 1 to 15. For example, a higher score indicates a greater workload on the body, thus requiring improvement in occupational posture. Parameters for occupational posture indices will be explained later. Furthermore, occupational posture indices are not limited to those described above; they can also be scores based on general posture assessment indicators, such as Rapid Entire Body Assessment (REBA) or Rapid Upper Limb Assessment (RULA).

[0026] The pressure accumulation index is an indicator used to evaluate the load applied to the fingers of an operator (WK). More specifically, the pressure accumulation index is expressed as the cumulative pressure value applied to the fingers of the operator (WK). In repetitive grasping actions or tasks requiring strong grip, the risk of trigger finger increases due to the cumulative pressure on the fingers. That is, the greater the total cumulative pressure value, the greater the load applied to the fingers, and therefore, for example, the need for work improvement.

[0027] The job support system 10 estimates load risk by combining multiple evaluation metrics using the machine learning model I322 described later. Thus, the job support system 10 can estimate load risk with higher accuracy than estimating load risk based on a single evaluation metric, thereby maintaining versatility.

[0028] Multiple acquisition units 100 are equipped with inertial measurement devices 110 and pressure sensors 120.

[0029] An inertial measurement unit 110 is installed on the work clothes WW worn by worker WK to acquire acceleration and angular velocity corresponding to changes in worker WK's posture. Hereinafter, the "inertial measurement unit" will also be referred to as an "inertial measurement unit (IMU)". The IMU 110 is installed, for example, on the back or arm of worker WK within the work clothes WW worn by worker WK. Thus, the IMU 110 acquires the acceleration and angular velocity generated on worker WK's back or arm. The control unit 300, described later, determines the angle of body bending based on changes in acceleration and angular velocity, for example, by calculating the angle between the vertical direction and the direction of back muscle extension. This angle, as the degree of body bending, is used to calculate work posture indicators.

[0030] Pressure sensor 120 is mounted on the glove WG of worker WK to acquire the pressure value applied to the fingers. Pressure sensor 120 is installed in the glove WG worn by worker WK at the fingertips of worker WK. For example, pressure sensor 120 acquires the pressure value applied to the fingertips of worker WK as worker WK grasps an object. However, pressure sensor 120 does not need to be installed on all fingers of worker WK; it can also be installed on a portion of the thumb, index finger, or middle finger.

[0031] The input unit 200 inputs multiple personal parameters representing the physical characteristics of the worker WK. The input unit 200 is, for example, a keyboard. That is, the user of the job support system 10 inputs the personal parameters of the worker WK into the control unit 300 by operating the keyboard, which serves as the input unit 200. Other personal parameters include, for example, the worker WK's height and grip strength.

[0032] Output unit 400 outputs the estimation results of work support system 10. More specifically, output unit 400 notifies worker WK of the load risk. Output unit 400 is, for example, a display. Output unit 400 prompts worker WK to take an appropriate break, for example, by displaying the level of load risk.

[0033] The control unit 300 controls the operation support system 10. The control unit 300 is a computer comprising a processor 310, a memory 320, an input / output interface 330, and an internal bus 340. The processor 310 is, for example, a central processing unit (CPU) and a graphics processing unit (GPU). The memory 320 includes, for example, storage devices including RAM, ROM, and a hard disk drive (HDD). The memory 320 stores programs (not shown). The processor 310 executes the programs stored in the memory 320. The input / output interface 330 is bidirectionally connected via the internal bus 340. A communication device 350, an input unit 200, and an output unit 400 are connected to the input / output interface 330. The communication device 350 can communicate directly or indirectly depending on the type of sensor.

[0034] Memory 320 stores correction information I321 and machine learning model I322.

[0035] Correction information I321 is information used to correct the values ​​of multiple evaluation indicators based on their respective individual parameters. Correction information I321 is, for example, a formula for calculating correction coefficients used to correct the values ​​of the evaluation indicators. The correction coefficients are calculated by substituting the individual parameters and the original values ​​of the evaluation indicators into this formula. This formula is experimentally determined considering the accuracy of load risk estimation.

[0036] For example, regarding work posture indicators, a formula is used where taller workers have higher evaluation values. This is because, in work environments designed for average height, taller workers (WK) are more likely to adopt inappropriate postures, thus increasing their risk of back pain. Regarding stress accumulation indicators, a formula is used where stronger grip strength has lower evaluation values. This is because stronger grip strength makes it easier to withstand loads, and even in the same manual work, workers with weaker grip strength have a higher risk of trigger finger compared to those with strong grip strength. These are just examples and do not apply to all jobs.

[0037] In addition, the correction information I321 is not limited to the correction coefficient. For example, it can also be mapping information that represents the value of the corrected evaluation index based on the individual parameters and the value of the evaluation index before correction.

[0038] Machine learning model I322 has been trained with values ​​of multiple evaluation indicators and multiple personal parameters related to multiple individuals with experience of illnesses caused by work performed by worker WK. Machine learning model I322 is composed of a neural network. Furthermore, the values ​​of the multiple evaluation indicators trained on machine learning model I322 are, for example, values ​​obtained immediately before the onset of illness in the individuals with experience of illness. Therefore, machine learning model I322 estimates workload risk, for example, by comparing the values ​​of the multiple evaluation indicators and the input personal parameters to machine learning model I322 with data from individuals with experience of illnesses that most closely resemble that data. For example, if the value of the evaluation indicator of the input data is greater than the value of the evaluation indicator of the data from individuals with experience of illnesses that most closely resemble that data, machine learning model I322 outputs a result indicating "high workload risk." By configuring it in this way, the job support system 10 of this embodiment can accurately estimate the high probability of workload risk arising from the same work as the individuals with experience of illness. Note that machine learning model I322 is not limited to data from individuals with experience of illness; it can also learn statistical averages of evaluation indicators or benchmark values ​​of standards on which each evaluation indicator is based.

[0039] A-2. Estimation Method:

[0040] Figure 2This is a flowchart illustrating a method for estimating the load risk caused by the work support system 10. In step S100, the user inputs multiple personal parameters representing the physical characteristics of the worker WK into the control unit 300 via the input unit 200. That is, the user inputs the height and grip strength of the worker WK into the control unit 300 via the keyboard. After step S100, the worker WK, wearing work clothes WW or gloves WG equipped with multiple acquisition units 100, begins work. Furthermore, the multiple personal parameters input are stored in the control unit 300, so that step S100 can be omitted when starting this process again.

[0041] In step S200, the control unit 300 acquires multiple measurement values ​​corresponding to the actions of the worker WK through multiple acquisition units 100. Specifically, the control unit 300 acquires the acceleration and angular velocity generated on the back or arm of the worker WK through the inertial measurement device 110. Furthermore, the control unit 300 acquires the pressure value applied to the fingertips of the worker WK through the pressure sensor 120. Additionally, the control unit 300 repeatedly acquires multiple measurement values ​​within a predetermined time period. That is, the predetermined time period is longer than the interval between acquisitions of the measurement values. For example, the predetermined time period is one day of work time. Therefore, after the work is completed, the next step S300 begins. Alternatively, the predetermined time period can be two hours or several minutes.

[0042] In step S300, the control unit 300 determines the values ​​of multiple evaluation indicators based on the acquired multiple measurement values. That is, the control unit 300 determines the values ​​of the work posture indicator and the pressure accumulation indicator.

[0043] More specifically, the control unit 300 determines the change in the relative position of the body part equipped with the inertial measurement device 110 based on the acceleration and angular velocity acquired multiple times in step S100. Furthermore, the control unit 300 determines a value representing the worker's posture (WK) based on this change in relative position. For example, the control unit 300 determines the angle for bending or extending the body by calculating the angle between the vertical direction and the direction of back muscle extension based on the change in the relative position of the back. Similarly, the control unit 300 determines the angle for raising or lowering the arm by calculating the angle between the direction of back muscle extension and the line connecting the shoulder to the elbow based on the change in the relative position of the arm and the back. The control unit 300 also determines the frequency of body bending and extension, arm raising and lowering, etc. As described above, the control unit 300 determines the value of the work posture index in fractional form by determining the parameters required for the work posture index.

[0044] Furthermore, the control unit 300 determines the value of the pressure accumulation index by accumulating the pressure values ​​acquired multiple times in step S100.

[0045] In step S400, the control unit 300 corrects the values ​​of multiple evaluation indicators by referring to correction information I321 based on multiple personal parameters pre-input through the input unit 200. For example, the control unit 300 determines the respective correction coefficients for the values ​​of the work posture indicator and the pressure accumulation indicator by substituting the values ​​of the multiple personal parameters and the multiple evaluation indicators into the calculation formula used as correction information I321. Furthermore, the control unit 300 calculates the corrected values ​​of the multiple evaluation indicators by multiplying the correction coefficients by the values ​​of each indicator.

[0046] In step S500, the control unit 300 estimates the load risk based on the values ​​of multiple evaluation indicators corrected by the correction information I321. That is, the control unit 300 outputs the load risk estimation result by inputting the corrected values ​​of multiple evaluation indicators into the machine learning model I322. The estimation result is represented, for example, by three levels: low, medium, and high, regarding lower back pain or trigger finger.

[0047] In step S600, the control unit 300 outputs the estimated load risk from the output unit 400. The control unit 300 may, for example, prompt the operator WK to take an appropriate break by displaying the load risk level on a monitor.

[0048] As described above, the job support system 10 considers multiple individual parameters to estimate load risk.

[0049] In this approach, the load risk acting on worker WK is estimated based on the value of a modified evaluation index. The modified evaluation index reflects individual parameters. Since these individual parameters represent physical characteristics, the estimated load risk reflects the physical characteristics of worker WK. Even for the same job, the degree of load that constitutes a risk varies depending on the physical characteristics of worker WK. However, the job support system 10 of this embodiment estimates load risk based on the physical characteristics of worker WK, which, compared to methods that do not consider the physical characteristics of worker WK, allows for a higher level of accuracy in estimating the likelihood of excessive load acting on the worker WK's body.

[0050] Furthermore, the control unit 300 uses a machine learning model I322 to estimate workload risk. By configuring it in this way, the job support system 10 of this embodiment can accurately estimate the high probability of workload risk that may arise from performing the same work as someone with a disease experience.

[0051] Furthermore, by setting multiple personal parameters, including the height and grip strength of the worker WK, the work support system 10 of this embodiment is able to estimate the likelihood of load risk based on the physique of the worker WK.

[0052] Furthermore, the multiple acquisition units 100 include an inertial measurement unit 110 and a pressure sensor 120. By configuring it in this way, the job support system 10 can easily estimate load risk for tasks involving whole-body movements and tasks involving fingertip movements. More specifically, the load risk associated with these tasks is easily affected by physical characteristics such as height, weight, and grip strength. Therefore, this approach further improves the effectiveness of the job support system 10 in estimating load risk by considering the physical characteristics of the worker (WK).

[0053] Furthermore, several evaluation indicators include a pressure accumulation index, which is calculated by accumulating the pressure values ​​applied to the operator WK's fingers. By configuring it in this way, the job support system 10 of this embodiment can easily estimate the risk of diseases arising from repetitive manual labor. That is, the job support system 10 of this embodiment can easily estimate the risk of trigger finger.

[0054] B. Other implementation methods:

[0055] (B1) In the above embodiment, the job support system 10 includes multiple acquisition units 100. However, the job support system 10 may also include only one acquisition unit 100. In addition, one acquisition unit 100 may acquire one measurement value or multiple measurement values.

[0056] (B2) In the above embodiment, the acquisition unit 100 is an inertial measurement device 110 or a pressure sensor 120. However, the acquisition unit 100 can also be other sensors. For example, the acquisition unit 100 can be a camera installed at the work site to capture images of the worker WK, or it can be a camera installed on the head of the worker WK to capture the worker WK's field of vision. Based on the images from these cameras, by analyzing the posture of the worker WK, a measurement value representing the change in the posture of the worker WK is acquired. Furthermore, the acquisition unit 100 can also be a bending sensor installed on the glove WG. The bending sensor acquires a resistance value corresponding to the action of opening or closing the hand. That is, the bending sensor acquires a measurement value corresponding to the action accompanying manual work.

[0057] (B3) In the above embodiment, the personal parameter is height or grip strength. However, the personal parameter is not limited to these. In addition, the personal parameter may also be weight, muscle strength other than grip strength, average time required for a specified task, past medical history, etc. For example, the machine learning model I322 may indicate that a longer average time required increases the workload risk, or that a worker WK with a history of medical conditions has a higher workload risk. As a result, the job support system 10 may be able to improve the accuracy of workload risk estimation.

[0058] (B4) In the above embodiments, multiple personal parameters are exemplified. However, there may also be only one personal parameter. That is, the personal parameter may be only height, only grip strength, or only weight.

[0059] For example, in the job support system 10, only one or more personal parameters need to be used, including at least one of height, weight, and grip strength. In this way, the job support system 10 can also estimate the likelihood of load risk based on the worker's (WK) physique.

[0060] (B5) In the above embodiment, the control unit 300 includes a machine learning model I322. However, the control unit 300 may also not include a machine learning model I322. In this case, the control unit 300 includes a rule-based estimation model for estimating load risk. The rule-based estimation model determines the level of load risk, for example, based on whether the values ​​of individual parameters and modified evaluation indicators meet predetermined conditions.

[0061] (B6) In the above embodiments, the multiple evaluation indicators include work posture indicators and stress accumulation indicators. However, the multiple evaluation indicators are not limited to these indicators. In addition to these indicators, or in place of these indicators, other evaluation indicators may also be included. Other evaluation indicators may be, for example, indicators of changes in body load or indicators of the time required for one cycle.

[0062] Body load variation indicators (BVR) are, for example, indicators used to evaluate changes in workload during work. More specifically, BVR analyzes the temporal variation of pressure applied to the fingertips to indicate how the worker's workload (WK) shifts. BVR parameters are, for example, the differences or points of change in pressure values ​​obtained from the waveform changes of pressure values ​​over one cycle. BVR uses these parameters to represent the degree of abrupt increases or decreases in workload applied to the body. Abrupt increases or decreases in workload can potentially increase the risk of illness. Therefore, BVR is used as an indicator for reviewing work performance to avoid abrupt increases or decreases in workload.

[0063] The time required for one cycle is an indicator used to evaluate the efficiency of a work process. More specifically, the time required for one cycle is the time required for one work cycle. Parameters of the time required for one cycle include, for example, the start and end times of the work obtained based on the location information of the worker (WK) or the results of work identification. Therefore, to obtain the parameters of the time required for one cycle, the acquisition unit 100 may include a surveillance camera capturing images of the work site or a wearable terminal capable of measuring work time based on whether the worker (WK) has moved. A higher time required for one cycle indicates a greater burden on the worker (WK). Therefore, the time required for one cycle is used for work standardization or efficiency improvement.

[0064] This invention is not limited to the embodiments described above, and can be implemented in various configurations without departing from its spirit. For example, in order to solve some or all of the above-described problems, or to achieve some or all of the above-described effects, the technical features of the embodiments corresponding to the technical features in the various methods described in the summary section of the invention can be appropriately replaced or combined. Furthermore, if the technical features are not required to be described in this specification, they can be appropriately deleted.

[0065] Symbol Explanation

[0066] 10-Work support system, 100-Acquisition unit, 110-Inertial measurement unit, 120-Pressure sensor, 200-Input unit, 300-Control unit, 310-Processor, 320-Memory, 330-Input / output interface, 340-Internal bus, 350-Communication device, 400-Output unit, I321-Correction information, I322-Machine learning model, WG-Gloves, WK-Operator, WW-Work clothes.

Claims

1. A job support system, characterized in that, have: One or more acquisition units acquire one or more measurement values ​​corresponding to the actions of the operator; The input unit receives one or more personal parameters representing the physical characteristics of the worker. The output unit outputs the estimation results of the operation support system; and The control unit controls the aforementioned work support system. The control unit includes correction information for correcting the values ​​of multiple evaluation indicators representing the degree of physical stress experienced by the worker during work based on the one or more individual parameters, according to the individual parameters of each of the aforementioned individual parameters. The one or more acquisition units acquire the one or more measurement values, and determine the values ​​of the plurality of evaluation indicators based on the acquired one or more measurement values. The values ​​of the plurality of evaluation indicators are corrected by referring to the correction information based on one or more personal parameters pre-input by the input unit, and the load risk as a risk caused by the load is estimated based on the corrected values ​​of the plurality of evaluation indicators.

2. The job support system according to claim 1, characterized in that, The control unit is equipped with a machine learning model, which has been trained with values ​​of multiple evaluation indicators and one or more individual parameters related to multiple individuals who have experienced diseases caused by the work performed by the workers. The control unit uses the machine learning model to estimate the load risk.

3. The job support system according to claim 2, characterized in that, The more than one personal parameter includes at least one of the worker's height, weight, and grip strength.

4. The job support system according to claim 3, characterized in that, The one or more acquisition units include: An inertial measurement unit, mounted on the worker's work clothes, acquires acceleration and angular velocity corresponding to changes in the worker's posture; and A pressure sensor, which is mounted on the worker's glove, acquires the pressure value applied to the fingers.

5. The job support system according to claim 4, characterized in that, The plurality of evaluation indicators include a pressure accumulation indicator, which is calculated by accumulating the pressure values.

Images