Work monitoring device and work monitoring method

Abstract

To provide a work monitoring device which estimates an abnormality which occurs in work conducted by a robot with high accuracy.SOLUTION: A work monitoring device 40 monitors work conducted by an industrial robot 11 attached with a work tool 12. The work monitoring device 40 includes an acquisition device 41 and an estimation device 42. The acquisition device 41 acquires at least a parameter regarding the work tool 12 during work and a parameter regarding motion of the robot 11. The estimation device 42 estimates whether an abnormality occurs during the work of the robot 11 based on a combination of the parameters acquired by the acquisition device 41.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present application mainly relates to a work monitoring device that monitors work performed by a robot. [Background technology]

[0002] Patent Document 1 discloses a welding method in which a welding torch is attached to an industrial robot to weld workpieces. In the welding method of Patent Document 1, a welding defect mode is estimated based on the welding current, welding voltage, number of short circuits, welding wire feed speed, etc., and confirmed by comparing it with the inspection results of the outer shape of the welded part. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2010-253538 Summary of the Invention [Problem to be solved by the invention]

[0004] In the welding method of Patent Document 1, the type of welding defect mode occurring is estimated based on the measurement results of the work environment and the parameters of the welding torch. However, because welding defects are caused by a combination of various factors, it may not be possible to accurately estimate the type of welding defect mode depending on, for example, differences in the robot's operation. Furthermore, this type of problem is not limited to welding, but is a common problem in various tasks performed using robots.

[0005] The present application has been made in view of the above circumstances, and its main object is to provide a work monitoring device that accurately estimates abnormalities that occur during work performed by a robot. [Means for solving the problem]

[0006] The problem to be solved by the present application is as described above. Next, the means for solving this problem and the effects thereof will be explained.

[0007] According to a first aspect of the present application, there is provided a work monitoring device for monitoring work performed by an industrial robot equipped with a work tool, the work monitoring device including an acquisition device and an estimation device. The acquisition device acquires at least parameters related to the work tool during work and parameters related to the operation of the robot. The estimation device estimates whether an abnormality has occurred during work by the robot based on a combination of the parameters acquired by the acquisition device.

[0008] According to a second aspect of the present application, there is provided the following work monitoring method. That is, the work monitoring method monitors work performed by an industrial robot equipped with a work tool. The work monitoring method acquires at least parameters related to the work tool during work and parameters related to the operation of the robot. The work monitoring method estimates whether an abnormality has occurred while the robot is working, based on a combination of the acquired parameters. [Effects of the Invention]

[0009] According to the present application, abnormalities occurring during work performed by a robot can be estimated with high accuracy. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is a block diagram of a work monitoring system. [Figure 2] FIG. 10 is an explanatory diagram showing the creation and use of a work inspection model. [Figure 3] A diagram comparing normal and abnormal conditions when the robot is performing arc welding. [Figure 4] FIG. 10 is a diagram conceptually showing normal and abnormal regions when there are three parameters. [Figure 5] 10 is a flowchart showing a process of monitoring and correcting a robot's work by the work monitoring system. [Figure 6] FIG. 10 is a block diagram of a modified example of a work monitoring system. DETAILED DESCRIPTION OF THE INVENTION

[0011] Next, an embodiment of the present application will be described with reference to the drawings. First, an overview of a work monitoring system 1 will be described with reference to FIG.

[0012] The work monitoring system 1 is a system that monitors work performed by industrial robots in real time. Work monitoring involves acquiring data related to the work and estimating whether or not an abnormality has occurred.

[0013] An industrial robot is a robot that performs work in a workplace such as a factory or warehouse. Industrial robots are of the teaching-and-playback type. The teaching-and-playback type means that the industrial robot is taught how to move in advance, and the industrial robot repeats the same movement in accordance with the teaching. An industrial robot is, for example, a vertically articulated or horizontally articulated arm robot. However, the industrial robot may be a robot other than an arm robot, such as a parallel link robot. Work performed by an industrial robot is, for example, assembly, welding, painting, machining, or transportation. Hereinafter, an industrial robot will be simply referred to as a "robot."

[0014] As shown in Fig. 1, the work monitoring system 1 of this embodiment is realized by various devices installed in a factory and a data center. The factory and the data center are connected via the Internet 2. This allows data to be transmitted and received between the devices installed in the factory and the devices installed in the data center. Note that the network connecting the factory and the data center may be a wide area network other than the Internet, or may be a local area network.

[0015] The factory is provided with a robot system 10, a sensor group 20, a management device 31, and a notification device 32. The robot system 10 also includes a robot 11, a work tool 12, and a control device 13.

[0016] The robot 11 of this embodiment is an arm robot installed in a factory. The work performed by the robot 11 is arc welding. The robot 11 has a plurality of arms 11a. The plurality of arms 11a are each operated independently by the power of an actuator such as a motor.

[0017] The work tool 12 performs work on the workpiece. In this embodiment, the work tool 12 is a welding torch that welds the workpiece. Note that, for example, when the work performed by the robot 11 is assembly or transportation, the work tool 12 is a hand that holds the workpiece.

[0018] The control device 13 includes a processing device such as a CPU and a storage device such as a hard disk, SSD, or flash memory. The processing device controls the robot 11 and the work tool 12 by executing programs stored in the storage device, causing the robot 11 and the work tool 12 to perform work. Specifically, the storage device of the control device 13 stores teaching data. The control device 13 operates the arm 11a by sending commands to an actuator based on the teaching data, and operates the work tool 12 by sending commands to the work tool 12 based on the teaching data. Hereinafter, work performed by the robot 11 equipped with the work tool 12 will be simply referred to as "work of the robot 11" or the like.

[0019] The sensor group 20 is a collective term for a plurality of sensors that detect various information related to the work of the robot 11. The sensor group 20 includes a clock 21, an ammeter 22, a voltmeter 23, a sound collection device 24, a camera 25, a workpiece thermometer 27, and an encoder 28. The sensors included in the sensor group 20 are examples, and some of the sensors may be omitted or other sensors may be added depending on the work content of the robot 11, the required accuracy of estimation, etc.

[0020] The clock 21 detects the time. Various pieces of information detected by each sensor in the sensor group 20 are associated with the time. The ammeter 22 detects the discharge current used in arc welding. The voltmeter 23 detects the discharge voltage used in arc welding. The sound pickup device 24 detects the volume and sound quality of the sound generated during arc welding. The sound quality is spectral data, in other words, data related to the amplitude of the sound according to its frequency. The camera 25 detects images of the welding location. The work thermometer 27 detects the temperature of the workpiece to be arc-welded, particularly the welding location. The welding location includes not only the location where the workpiece actually melts, but also the surrounding area. The encoder 28 is provided at each joint of the arm 11a and detects the rotation angle of each arm 11a.

[0021] The management device 31 is, for example, a PC, and includes a processing device such as a CPU, a storage device such as a hard disk, SSD, or flash memory, and a communication device such as a communication module. The management device 31 acquires information detected by each sensor of the sensor group 20 via the communication device. The management device 31 transmits the acquired information to the work monitoring device 40 in the data center via the Internet 2. Furthermore, the management device 31 acquires commands sent by the control device 13 to the robot 11 and the work tool 12, and transmits them to the work monitoring device 40 in the data center via the Internet 2. The management device 31 may transmit the information acquired from the sensor group 20 or the control device 13 to the work monitoring device 40 without processing it, or may process the information acquired from the sensor group 20 or the control device 13 and transmit it to the work monitoring device 40. Processing includes extracting necessary information or changing the data format, etc.

[0022] The notification device 32 notifies factory workers in response to commands from the management device 31. The notification device 32 is, for example, a display, a lamp, or a mobile terminal. The display, for example, displays information identifying the robot 11 that is waiting for a worker. Waiting for a worker means that an abnormality has occurred in the work of the robot 11, and the robot 11 needs assistance from a worker to resume the work. The notification device 32 may also be a lamp. For example, a lamp is provided for each robot 11. The lamp lights up or flashes when the corresponding robot 11 is waiting for a worker. Note that a lamp may also be provided for each area. The mobile terminal is capable of communicating with the management device 31 and is equipped with a small display. The mobile terminal displays the content received from the management device 31 on the small display.

[0023] An operation monitoring device 40 is provided in the data center. The operation monitoring device 40 is, for example, a server device, and includes a processing device such as a CPU, a storage device such as a hard disk, SSD, or flash memory, and a communication device such as a communication module. The operation monitoring device 40 may be a single piece of hardware, or may be configured with multiple pieces of hardware working together. For example, the hardware that aggregates and stores information may be separate from the hardware that performs processing in response to information received from the management device 31. The operation monitoring device 40 may be realized by a cloud computing service.

[0024] The work monitoring device 40 includes an acquisition device 41 and an estimation device 42. The acquisition device 41 corresponds to a communication module included in the work monitoring device 40. The acquisition device 41 acquires information detected by each sensor in the sensor group 20, commands to the robot 11, and commands to the work tool 12 via the Internet 2. The estimation device 42 corresponds to a processing device and a storage device included in the work monitoring device 40. The estimation device 42 estimates whether an abnormality has occurred in the work of the robot 11 based on the information acquired by the acquisition device 41.

[0025] Next, the process by which the estimation device 42 estimates whether or not an abnormality has occurred in the work of the robot 11 will be described in detail with reference to FIGS.

[0026] In this embodiment, the work performed by the robot 11 is arc welding. The abnormality estimated by the estimation device 42 is incomplete fusion. Incomplete fusion occurs when welding is performed without the weld interface being sufficiently melted, resulting in a weak bond between the weld metal and the base material. Incomplete fusion occurs inside the workpiece, specifically at the interface between the weld metal and the base material. Therefore, the occurrence of incomplete fusion cannot be detected by visual inspection alone.

[0027] The work monitoring system 1 can also detect welding defects other than incomplete fusion. For example, the work monitoring system 1 can also be applied to undercuts, blowholes, etc. The work monitoring system 1 can also be applied to welding other than arc welding. Furthermore, the work monitoring system 1 can also be applied to the various other works mentioned above other than welding.

[0028] The estimation device 42 of this embodiment estimates and outputs whether the state of the workpiece is normal or whether malabsorption has occurred. The estimation result output by the estimation device 42 is not limited to two levels, normal or malabsorption. For example, the estimation device 42 may output an estimation result in three levels, such as normal, low probability of malabsorption occurring, or high probability of malabsorption occurring. Alternatively, the estimation device 42 may output a specific numerical value of the incidence rate of malabsorption as the estimation result.

[0029] Furthermore, the estimation device 42 may output estimation results from different perspectives, as long as they essentially mean whether or not an abnormality has occurred. For example, estimation results may be output from the perspectives of task success / task failure, or no correction required / correction required. All of the above cases are included in the "estimation of whether or not an abnormality has occurred in the task of the robot 11."

[0030] The estimation device 42 estimates whether or not a fusion defect has occurred during the work of the robot 11, using a work inspection model constructed by machine learning. As shown in FIG. 2, the work inspection model is a model constructed by machine learning inspection of inspection history data. The inspection history data is created based on multiple inspections. The inspection history data includes parameters and the results of ultrasonic testing. Note that the inspection history data may include the results of radiographic testing instead of or in addition to the results of ultrasonic testing. For example, when the work monitoring system 1 is applied to a blowhole, the estimation device 42 can estimate whether or not a welding defect has occurred by using the results of the radiographic testing.

[0031] Parameters are various data related to the work of the robot 11. Parameters can be divided into, for example, active parameters included in commands to the robot 11 or the work tool 12, and passive parameters acquired by sensors in the sensor group 20. In other words, active parameters are values ​​that can be directly specified by the control device 13. Passive parameters are values ​​that arise as a result of the work performed by the robot 11 and the environment.

[0032] In this embodiment, the tool attitude, movement speed, and welding applied voltage are active parameters. The tool attitude is the attitude of the work tool 12 with respect to the welding point. The tool attitude can also be described as the orientation of the work tool 12 with respect to the welding point. The tool attitude is a parameter related to the work tool 12. The movement speed is the speed at which the robot 11 moves the work tool 12 during welding. The movement speed is a parameter related to the robot 11. The welding applied voltage is the voltage applied to the work tool 12 for arc welding. The welding applied voltage is a parameter related to the work tool 12. In this way, the parameters in this embodiment include parameters related to the robot 11 and parameters related to the work tool 12.

[0033] The passive parameters include the welding current, welding voltage, volume, sound quality, image parameters, workpiece temperature, and tool usage time. The welding current is the value detected by the ammeter 22. The welding voltage is the value detected by the voltmeter 23. The volume and sound quality are the values ​​detected by the sound pickup device 24. The weld distortion is the degree of cracking in the weld. The image parameters are parameters created by analyzing the image captured by the camera 25, such as weld distortion, weld line posture, arc light, weld pool shape, and workpiece size. The weld line is the trajectory of the welding location. The weld line posture is a value related to how the weld line changes, such as whether it is ascending or descending. Instead of analyzing the image to calculate the workpiece size, the workpiece size may be obtained based on data created during teaching of the robot 11. In this case, information regarding the workpiece material may also be obtained based on the data created during teaching of the robot 11. The tool usage time is the continuous usage time or cumulative usage time of the work tool 12. The tool usage time is determined based on logs from the management device 31 or the control device 13. The image parameters and the workpiece temperature are parameters related to the workpiece, while the other parameters are parameters related to the work environment.

[0034] If the workpiece moves or its orientation changes during welding, the tool posture and movement speed may be calculated using images captured by a camera in addition to command values ​​for the robot 11. The above-described parameters are merely examples, and at least one of them may be omitted, or other parameters may be used. For example, when the work monitoring system 1 is applied to welding defects other than incomplete fusion, suitable parameters are used to estimate the occurrence of the applicable welding defect. Another parameter is information about the surrounding environment, such as the temperature and humidity of the workplace. The temperature and humidity of the workplace are related to the dew point temperature, which is the temperature at which condensation occurs on the base material. When welding areas wetted by condensation, incomplete fusion or blowholes are more likely to occur. Therefore, information about the surrounding environment, such as the temperature and humidity, can be used as parameters for estimating welding defects.

[0035] Next, the differences in various parameters between normal and abnormal conditions where incomplete fusion occurs will be briefly explained with reference to Fig. 3. Fig. 3 shows the position of the work tool 12, the arc light, and the molten pool in normal and abnormal conditions.

[0036] The tool attitude, movement speed, and applied welding voltage are parameters that are directly related to arc welding, and it is clear that these parameters are related to the occurrence of incomplete fusion.

[0037] Next, as shown in Figure 3, the arc discharge distance is longer under abnormal conditions compared to normal conditions, which tends to result in a lower welding current. On the other hand, the welding voltage tends to be higher under abnormal conditions compared to normal conditions. The volume and frequency of the noise tend to be lower under abnormal conditions compared to normal conditions. Weld distortion or an upturned weld line position tend to increase the likelihood of incomplete fusion. As shown in Figure 3, the size and shape of the arc light differ between normal and abnormal conditions, as well as the size and shape of the molten pool. Furthermore, as the workpiece becomes larger or the thermal conductivity of the workpiece increases, heat escapes more easily from the weld, resulting in a faster cooling rate of the weld metal, which tends to increase the likelihood of incomplete fusion. Under abnormal conditions, the gaps created by incomplete fusion make it difficult for heat to be transmitted, which tends to delay the temperature rise compared to normal conditions. Tool usage time affects the quality of the weld and is therefore related to the occurrence of incomplete fusion.

[0038] As described above, each parameter is related to the occurrence of malabsorption. Furthermore, there is no dominant parameter that can predict the occurrence of malabsorption. In other words, whether or not malabsorption occurs is determined not by each parameter alone, but by the interaction of multiple parameters.

[0039] When creating the inspection history data for creating the work inspection model, these parameters are stored in association with the results of the ultrasonic testing. The results of the ultrasonic testing are the results of conducting an ultrasonic testing to check whether or not a fusion defect has occurred. The work inspection model is a model constructed by machine learning the inspection history data. Because the inspection history data includes the results of the ultrasonic testing, this machine learning is supervised learning.

[0040] The work inspection model is a model that estimates whether malabsorption will occur when multiple parameter values ​​are input. Figure 4 is a conceptual diagram of the work inspection model when the number of parameters is reduced to three for simplicity. Figure 4 shows a virtual space with the three parameters as coordinate axes, and normal and abnormal regions are defined. If a point in the virtual space determined based on the three parameters is in the normal region, it is estimated that malabsorption has not occurred and the model is normal. If a point in the virtual space determined based on the three parameters is in the abnormal region, it is estimated that malabsorption has occurred and the model is abnormal.

[0041] For example, if the inspection standard for malabsorption is defined as "the length of malabsorption per unit length is equal to or less than a threshold," the inspection standard is met even if malabsorption occurs only slightly as long as it is below the threshold. Therefore, even if malabsorption occurs, the inspection standard can be met by immediately changing the operating conditions to return to normal. On the other hand, if the threshold for the inspection standard for malabsorption is very small, it becomes difficult to meet the inspection condition at the time malabsorption occurs. In this case, the range of the normal region can be narrowed or the range of the abnormal region can be expanded. Since there is a high possibility that malabsorption has not occurred immediately after entering the abnormal region, immediate correction can further suppress the occurrence of malabsorption. In this specification, the abnormal region includes not only a region where an abnormality is suspected, but also a region where an abnormality is suspected or a region that may lead to the occurrence of an abnormality.

[0042] It is also possible to estimate whether or not malabsorption has occurred without modeling the examination history data. For example, when there are three parameters, a large number of examination data are plotted in the virtual space shown in Figure 4. Next, a boundary is created to separate the set of points where malabsorption has occurred from the set of points where malabsorption has not occurred. This makes it possible to create normal and abnormal regions. It is also possible to perform this process regardless of the number of parameters.

[0043] Next, with reference to FIG. 5, a process for monitoring and correcting the work of the robot 11 by the work monitoring system 1 will be described.

[0044] First, the robot 11 starts working based on instructions from the management device 31 (S101). Specifically, the management device 31 instructs the control device 13 to start working with the robot 11. The control device 13 sends commands to the robot 11 and the work tool 12 based on the teaching data. As a result, the robot 11 and the work tool 12 perform welding on the workpiece.

[0045] Next, the management device 31 determines whether the work has been completed (S102). If the work has not been completed, the acquisition device 41 of the work monitoring device 40 acquires parameters in real time (S103). Specifically, while the work is in progress, the management device 31 receives information detected by each sensor of the sensor group 20 and commands sent by the control device 13 to the robot 11 and work tool 12, processes the information as necessary, and sends it. The acquisition device 41 receives the information, i.e., the parameters, sent by the management device 31.

[0046] Next, the estimation device 42 of the work monitoring device 40 estimates whether or not an abnormality has occurred based on the multiple parameters acquired by the acquisition device 41 (S104). Specifically, the estimation device 42 inputs the multiple parameters acquired by the acquisition device 41 into the work inspection model. This provides an estimation result of whether the current state is in the normal region or the abnormal region. If the current state is normal, no abnormality has occurred, and so the process of step S102 is performed again.

[0047] If the current state is an abnormal state, the estimation device 42 determines whether automatic correction is permitted (S106). Correction means changing the work conditions from an abnormal region to a normal region by correcting the work conditions, as shown in Fig. 4. The work monitoring system 1 of this embodiment is capable of performing automatic correction, in which the work monitoring system 1 corrects the work conditions, and manual correction, in which a factory worker manually corrects the work conditions.

[0048] If automatic correction is permitted, the estimation device 42 instructs the control device 13 via the management device 31 to correct commands for at least one of the robot 11 and the work tool 12. This changes at least one of the tool posture, movement speed, and applied welding voltage.

[0049] Next, the acquisition device 41 performs the process of step S103 again to acquire new parameters. The estimation device 42 performs the process of step S104 again to estimate whether or not an abnormality has occurred. In the second estimation, the parameters have changed compared to the previous estimation, so there is a possibility that the abnormal state has returned to a normal state. The work monitoring system 1 repeats the process of changing the parameters until the estimation device 42 estimates that the state is normal.

[0050] In this embodiment, the command is modified according to the following rules. As described above, the parameters that the control device 13 can change are the tool attitude, the travel speed, and the applied welding voltage. The first parameter, which has the smallest effect on the welding quality of arc welding, is the tool attitude, the second parameter, which has the next smallest effect, is the travel speed, and the third parameter, which has the largest effect, is the applied welding voltage. In this embodiment, the priority of change is highest for the first parameter, followed by the second parameter and the third parameter. Specifically, the command is modified so that the first parameter is changed first and the other parameters are not changed. Then, if the estimation result is not normalized by changing the first parameter alone, a change in the second parameter is permitted next, and finally a change in the third parameter is permitted.

[0051] Instead of completely fixing the second and third parameters when the first parameter is changed, the condition may be that the amount of change in the second and third parameters is within a threshold value. In other words, any rule may be defined as long as the priority is defined in the order of the first parameter, the second parameter, and the third parameter.

[0052] If it is determined in step S106 that automatic correction is not permitted, i.e., manual correction is permitted, the estimating device 42 operates the notification device 32 via the management device 31 to request a factory worker to make a correction (S108). The worker corrects the command by, for example, operating the management device 31 or the control device 13. Even in the case of manual correction, the estimating device 42 again performs the process of step S104 at an appropriate timing to estimate whether an abnormality has occurred. In other words, even in the case of manual correction, the worker repeats the process of changing the parameters until the estimating device 42 estimates that the parameter is normal.

[0053] By performing these processes, even if an abnormality occurs, it is possible to immediately return to normal, thereby preventing or minimizing the occurrence of poor fusion.

[0054] Next, a modified example of the above embodiment will be described with reference to Fig. 6. In the description of this modified example, the same or similar components as those in the above embodiment will be denoted by the same reference numerals in the drawings, and the description thereof may be omitted.

[0055] In the above embodiment, the work monitoring device 40 is installed in a data center, in other words, on the cloud. Instead, in this modified example, the work monitoring device 40 is installed in a factory. In this case, the work monitoring device 40 may also have the functions of the management device 31. In this way, the work monitoring system 1 can also be configured to be completed within the factory.

[0056] As described above, the work monitoring device 40 of this embodiment performs a work monitoring method for monitoring work performed by an industrial robot 11 equipped with a work tool 12. The work monitoring device 40 includes an acquisition device 41 and an estimation device 42. The acquisition device 41 acquires at least parameters related to the work tool 12 during work and parameters related to the operation of the robot 11. The estimation device 42 estimates whether or not an abnormality has occurred while the robot 11 is working, based on a combination of multiple parameters acquired by the acquisition device 41.

[0057] Since an abnormality in the work of the robot 11 is estimated based on multiple types of parameters, the occurrence of an abnormality can be estimated with high accuracy. Furthermore, by performing an abnormality estimation during work, if an abnormality occurs or a situation close to an abnormality occurs, it can be identified or dealt with early.

[0058] In the work monitoring device 40 of this embodiment, the acquisition device 41 further acquires parameters related to the workpiece that is the work target.

[0059] This increases the number of types of parameters that can be used for estimation, allowing for accurate abnormality estimation.

[0060] In the work monitoring device 40 of this embodiment, the parameters acquired by the acquisition device 41 include active parameters included in commands to the robot 11 or work tool 12, and passive parameters acquired by sensors.

[0061] This allows estimation from both an active and a passive perspective, making it possible to accurately estimate an abnormality.

[0062] In the work monitoring device 40 of this embodiment, the estimation device 42 is capable of communicating with the notification device 32 that notifies that an abnormality has occurred. When the estimation device 42 estimates that an abnormality has occurred, it instructs the notification device 32 to notify the occurrence of the abnormality.

[0063] This allows the user to be notified that an abnormality may have occurred.

[0064] In the work monitoring device 40 of this embodiment, the estimation device 42 is capable of communicating with the control device 13 that controls at least one of the work tool 12 and the robot 11. When the estimation device 42 estimates that an abnormality has occurred, it instructs the control device 13 to correct the command for at least one of the work tool 12 and the robot 11.

[0065] This makes it possible to change the situation in which an abnormality may occur, thereby preventing the robot 11 from failing in its work.

[0066] In the work monitoring device 40 of this embodiment, the estimation device 42 repeatedly estimates whether or not an abnormality has occurred after correcting the command. The estimation device 42 repeatedly instructs the correction of the command until it estimates that no abnormality has occurred.

[0067] This allows multiple corrections to be made even if a failure cannot be prevented by a single correction, thereby increasing the probability that a failure in the work of the robot 11 can be prevented in advance.

[0068] In the work monitoring device 40 of this embodiment, the parameters include a first parameter and a second parameter. The command is corrected so that the first parameter is changed with priority over the second parameter.

[0069] This makes it possible to suppress the amount of change in the second parameter, so that, for example, by using a parameter that affects the quality of work or other items as the second parameter, it is possible to reduce the impact of command correction.

[0070] In the work monitoring device 40 of this embodiment, the work performed by the robot 11 is welding. The first parameter is a parameter related to the posture of the work tool 12. The second parameter is a parameter related to the robot 11 that moves the work tool 12.

[0071] The movement speed of the work tool 12 is likely to affect the welding quality, so by setting the movement speed of the work tool 12 as the second parameter, the frequency of correction of the movement speed of the work tool 12 can be reduced.

[0072] In the work monitoring device 40 of this embodiment, the estimation device 42 estimates whether an abnormality has occurred while the robot 11 is working by using a work inspection model constructed by machine learning of multiple parameters and results indicating whether an abnormality has occurred in a combination of the parameters.

[0073] This allows for accurate anomaly estimation even in situations where multiple parameters interact with each other and affect the success or failure of the work.

[0074] In the work monitoring device 40 of this embodiment, the estimation device 42 estimates whether or not an abnormality has occurred inside the workpiece that is the target of work by the robot 11.

[0075] Since it is difficult to accurately estimate abnormalities that are not visible to the naked eye, the effects of this embodiment can be effectively utilized.

[0076] The preferred embodiment and modifications of the present application have been described above, but the above configurations can be modified, for example, as follows.

[0077] The flowcharts shown in the above embodiments are merely examples, and some processes may be omitted, the contents of some processes may be changed, or new processes may be added.

[0078] For example, the correction process may be omitted when the estimation device 42 estimates an abnormality. In this case, the work may be stopped or restarted at the point when the estimation device 42 estimates an abnormality.

[0079] Alternatively, when the estimation device 42 estimates an abnormality, it may determine whether or not correction is possible. If it determines that correction is possible, the estimation device 42 may instruct the worker to make the correction, or if correction is not possible, the estimation device 42 may instruct the worker to stop or redo the work. For example, when the estimation device 42 estimates that an abnormality has occurred in a welding location where even slight lack of fusion is not acceptable, the estimation device 42 determines that correction is not possible. On the other hand, when the estimation device 42 estimates that an abnormality has occurred in a welding location where a certain degree of lack of fusion is acceptable, the estimation device 42 determines that correction is possible.

[0080] In the above embodiment, the presence or absence of only one type of abnormality is estimated. Alternatively, the work monitoring system 1 may estimate the presence or absence of multiple types of abnormalities. In this case, a model may be created for each type of abnormality. Alternatively, a model capable of estimating the type of abnormality may be constructed by machine learning.

[0081] The functions of the elements disclosed herein can be performed using circuits or processing circuitry, including general-purpose processors, special-purpose processors, integrated circuits, ASICs (Application Specific Integrated Circuits), conventional circuits, and / or combinations thereof, configured or programmed to perform the disclosed functions. A processor is considered a processing circuit or circuitry because it includes transistors and other circuitry. In this disclosure, a circuit, unit, or means is hardware that performs the recited functions or hardware that is programmed to perform the recited functions. The hardware may be hardware disclosed herein or other known hardware that is programmed or configured to perform the recited functions. Where the hardware is a processor, which is considered a type of circuit, the circuit, means, or unit is a combination of hardware and software, and the software is used to configure the hardware and / or processor. [Explanation of symbols]

[0082] 1. Work monitoring system 10 Robot Systems 11. Robot 12 Work Tools 40 Work monitoring device 41 Acquisition device 42 Estimation device

Claims

1. A work monitoring device for monitoring work performed by an industrial robot equipped with a work tool, an acquisition device that acquires at least parameters related to the work tool during work and parameters related to the operation of the robot; an estimation device that estimates whether a specific abnormality has occurred during work by the robot by inputting the multiple parameters acquired by the acquisition device into a work inspection model constructed by machine learning the multiple parameters and a result indicating that a specific abnormality has occurred due to a combination of the parameters; A work monitoring device comprising:

2. The work monitoring device according to claim 1, The acquisition device further acquires parameters related to the workpiece, which is the object of the work.

3. The work monitoring device according to claim 1 or 2, A work monitoring device, wherein the parameters acquired by the acquisition device include active parameters included in commands to the robot or the work tool, and passive parameters acquired by sensors.

4. The work monitoring device according to any one of claims 1 to 3, the estimation device is capable of communicating with a notification device that notifies that an abnormality has occurred; When the estimation device estimates that an abnormality has occurred, the estimation device instructs the notification device to notify the occurrence of the abnormality.

5. A work monitoring device that monitors work performed by an industrial robot equipped with a work tool, an acquisition device that acquires at least parameters related to the work tool during work and parameters related to the operation of the robot; an estimation device that estimates whether or not an abnormality has occurred during work by the robot based on a combination of the plurality of parameters acquired by the acquisition device; Equipped with the estimation device is capable of communicating with a control device that controls at least one of the work tool and the robot, When the estimation device estimates that an abnormality has occurred, the estimation device instructs the control device to modify a command to at least one of the work tool and the robot.

6. The work monitoring device according to claim 5, the estimation device repeatedly estimates whether an abnormality has occurred after the command has been corrected; The estimation device repeatedly instructs the correction of the command until it estimates that no abnormality has occurred.

7. The work monitoring device according to claim 6, the parameters include a first parameter and a second parameter; The work monitoring device modifies the command so that the first parameter is changed with priority over the second parameter.

8. The work monitoring device according to claim 7, The work performed by the robot is welding, the first parameter is a parameter related to the attitude of the work tool, The work monitoring device, wherein the second parameter is a parameter related to a speed at which the robot moves the work tool.

9. The work monitoring device according to any one of claims 1 to 8, The estimation device is a work monitoring device that estimates whether or not an abnormality has occurred inside the workpiece that is the target of work by the robot.

10. The work monitoring device according to any one of claims 1 to 9, The estimation device is a work monitoring device that estimates whether or not an abnormality has occurred in the welding performed by the robot.

11. 1. A work monitoring method for monitoring work performed by an industrial robot equipped with a work tool, comprising: Acquire at least parameters related to the work tool during work and parameters related to the operation of the robot; A work monitoring method in which a work inspection model is constructed by machine learning the plurality of parameters and a result indicating that a specific abnormality has occurred for a combination of the parameters, and the obtained plurality of parameters are input into the model, thereby estimating whether or not a specific abnormality has occurred while the robot is working.

12. 1. A work monitoring method for monitoring work performed by an industrial robot equipped with a work tool, comprising: Acquire at least parameters related to the work tool during work and parameters related to the operation of the robot; Based on a combination of the acquired plurality of parameters, it is estimated whether or not an abnormality has occurred during the work of the robot; A work monitoring method in which, when it is estimated that an abnormality has occurred, a control device that controls at least one of the work tool and the robot is instructed to correct a command to at least one of the work tool and the robot.

Images