Work content estimating system and work content estimating method

Abstract

This work content estimating system comprises: an acquiring unit that acquires log information relating to a working machine, the log information being associated with a time; and an estimating unit that estimates a time series of work content of the working machine by inputting appearance / operation information based on the log information, the appearance / operation information being time series information corresponding to the appearance and operation of a movable portion of the work machine, into a trained machine learning model that accepts input of the appearance / operation information and outputs information indicating an estimated result of the work content of the work machine.

Description

Work content estimation system and work content estimation method

[0001] This application claims priority to Japanese Patent Application No. 2023-203125, filed on November 30, 2023, the contents of which are incorporated herein by reference.

[0002] Patent Document 1 describes an analysis support system including an estimation unit that estimates the work content of a work machine at each time from log information of the work machine. The estimation unit obtains a time series of likelihoods related to unit tasks by inputting the log information into a unit task prediction model in chronological order. The log information is time-series information indicating the status of the work machine acquired by multiple sensors equipped on the work machine. Patent Document 1 also lists, as examples of log information, position information, roll angle, pitch angle, swing angle, boom angle, arm angle, and bucket angle, as well as information indicating the proportional pressure control (PPC) pressure of the operating device, engine output, and information indicating instantaneous fuel consumption.

[0003] Japanese Patent Application Laid-Open No. 2020-183615

[0004] In the log information described above, information such as engine output and instantaneous fuel consumption is easily affected by differences in the class of the work machine. For this reason, if this information is used as input information for a machine learning model, it may be necessary to prepare multiple models according to differences in the class of the work machine.

[0005] The present disclosure aims to provide a work content estimation system and a work content estimation method that can improve the versatility of a machine learning model used to estimate work content.

[0006] The work content estimation system of the present disclosure includes an acquisition unit that acquires log information of a work machine associated with time, and an estimation unit that estimates the time series of the work content of the work machine by inputting external movement information, which is time series information corresponding to the external movement of a moving part of the work machine, based on the log information into a trained machine learning model that inputs the external movement information and outputs information representing an estimation result of the work content of the work machine.

[0007] The work content estimation method disclosed herein includes the steps of acquiring log information of a work machine associated with time, and estimating the time series of the work content of the work machine by inputting visual movement information, which is time series information corresponding to the visual movement of a moving part of the work machine, based on the log information into a trained machine learning model that inputs the visual movement information and outputs information representing an estimation result of the work content of the work machine.

[0008] According to the display control device and display control method of the present disclosure, it is possible to improve the versatility of the machine learning model used to estimate the work content.

[0009] FIG. 1 is a diagram showing the overall configuration of an analysis support system according to a first embodiment. FIG. 2 is a diagram showing the structure of a work machine according to a first embodiment. FIG. 3 is a diagram showing the functional configuration of a work content estimation system according to a first embodiment. FIG. 4 is a diagram showing the processing flow of the work content estimation system according to a first embodiment. FIG. 5 is a diagram showing an example of log information according to a first embodiment. FIG. 6 is a diagram showing an example of external movement information according to a first embodiment. FIG. 7 is a diagram showing a heat map used for estimating work content according to a first embodiment.

[0010] First Embodiment A display control device and a display control method according to a first embodiment will be described in detail below with reference to FIGS.

[0011] (Overall configuration of analysis support system) Figure 1 is a diagram showing the overall configuration of an analysis support system according to embodiment 1. The analysis support system 1 has a work content estimation system 10 and a data logger 20 mounted on each of a plurality of work machines 3.

[0012] The work machine 3 is the subject of work analysis by the work content estimation system 10. Examples of the work machine 3 include hydraulic excavators, wheel loaders, and bulldozers. In the following description, a hydraulic excavator will be used as an example of the work machine 3. Each work machine 3 is equipped with multiple sensors. The data logger 20 chronologically records and accumulates information obtained by the sensors indicating the state of the work machine 3. Hereinafter, the information recorded by the data logger 20 indicating the state of the work machine 3 at each time will also be referred to as log information. In addition, if the operation mechanism that operates the work machine 3 is configured to operate the work machine 3 using electrical operation signals, information on the operation signals of the work machine 3 may be recorded and accumulated chronologically and included in the log information. The data logger 20 transmits the recorded log information to the work content estimation system 10 via a wide area communication network at regular time intervals. In addition, the regular time intervals may be, for example, every five minutes. The work content estimation system 10 records the log information received from the data logger 20 on a recording medium. The functions of the work content estimation system 10 will be described later.

[0013] (Structure of Work Machine) Figure 2 is a diagram showing the structure of a work machine according to the first embodiment. The work machine 3, which is a hydraulic excavator, excavates earth and sand and levels the ground at work sites and the like. As shown in Figure 2, the work machine 3, which is a hydraulic excavator, comprises a lower traveling body 31 for traveling, and a rotatable upper rotating body 32 that is installed above the lower traveling body 31. The upper rotating body 32 is also provided with a driver's cab 32A, a work implement 32B, and two GNSS (Global Navigation Satellite System) antennas G1, G2. Note that the upper rotating body 32 will also be referred to as the vehicle body hereinafter.

[0014] The undercarriage 31 has a left crawler track CL and a right crawler track CR. The work machine 3 moves forward, turns, and reverses by rotation of the left crawler track CL and the right crawler track CR.

[0015] The operator's cab 32A is a place where an operator of the work machine 3 gets in and operates the work machine 3. The operator's cab 32A is installed, for example, on the left side of the front end of the upper rotating body 32. The internal configuration of the operator's cab 32A will be described later.

[0016] The work implement 32B consists of a boom BM, an arm AR, and a bucket BK. The boom BM is attached to the front end of the upper rotating body 32. An arm AR is attached to the boom BM. A bucket BK is attached to the arm AR. A boom cylinder SL1 is attached between the upper rotating body 32 and the boom BM. The boom BM can be moved relative to the upper rotating body 32 by driving the boom cylinder SL1. An arm cylinder SL2 is attached between the boom BM and the arm AR. The arm AR can be moved relative to the boom BM by driving the arm cylinder SL2. A bucket cylinder SL3 is attached between the arm AR and the bucket BK. The bucket BK can be moved relative to the arm AR by driving the bucket cylinder SL3. The upper rotating body 32, boom BM, arm AR, and bucket BK described above that are equipped on the work machine 3, which is a hydraulic excavator, are one aspect of the movable parts of the work machine 3. Further, the lower traveling body 31 is also one aspect of the movable part of the work machine 3. Furthermore, the bucket BK is one example of a configuration of a "work tool" according to the present disclosure. The bucket BK is equipped with a cutting edge BKT used for excavation, etc. Note that in other types of work machines, for example, tracks, blades, etc. are movable parts, and blades, rippers, etc. are the working tools.

[0017] (Functional Configuration of the Work Content Estimation System) FIG. 3 is a diagram showing the functional configuration of the work content estimation system according to the first embodiment. Hereinafter, functions of the work content estimation system 10 according to the first embodiment will be described with reference to FIG. 3. As shown in FIG. 3, the work content estimation system 10 includes a CPU 100, a memory 101, a display unit 102, an operation reception unit 103, a communication interface 104, and storage 105. Note that the CPU (Central Processing Unit) 100 may be a processor such as an FPGA or GPU instead of a CPU.

[0018] The CPU 100 is a processor that controls the overall operation of the task content estimation system 10. The various functions of the CPU 100 will be described later.

[0019] The memory 101 is a so-called main storage device, and stores instructions and data necessary for the CPU 100 to operate based on a program.

[0020] The display unit 102 is a display device capable of visually displaying information, and is, for example, a liquid crystal display or an organic EL display.

[0021] The operation reception unit 103 is an input device, such as a general mouse, keyboard, or touch sensor.

[0022] The communication interface 104 is a communication interface for communicating with the data logger 20 .

[0023] The storage 105 is a so-called auxiliary storage device, such as a hard disk drive (HDD) or solid state drive (SSD). The storage 105 records log information TL received from the data logger 20, a work machine model TM which is a 3D model prepared in advance for each vehicle type and model of the work machine 3, and the like. The work machine model TM will be described later. The storage 105 also records a unit work estimation model PM1 and an element work estimation model PM2 which are trained machine learning models used when estimating the work content of the work machine 3, heat maps (H1, H2) generated during the estimation process, estimated work content R of the work machine 3, and the like. The unit work estimation model PM1, element work estimation model PM2, and heat maps (H1, H2) will be described later.

[0024] The functions of the CPU 100 of the work content estimation system 10 according to the first embodiment will be described in detail. The CPU 100 operates based on a predetermined program to function as an acquisition unit 1000 and an estimation unit 1001. The predetermined program may be configured to implement some of the functions of the work content estimation system 10. For example, the program may be configured to implement the functions in combination with other programs already stored in the storage 105 or in combination with other programs implemented in other devices. In other embodiments, the work content estimation system 10 may include a custom large-scale integrated circuit (LSI) such as a programmable logic device (PLD) in addition to or instead of the above configuration. Examples of PLDs include programmable array logic (PAL), generic array logic (GAL), complex programmable logic device (CPLD), and field programmable gate array (FPGA). In this case, some or all of the functions implemented by the processor may be implemented by the integrated circuit.

[0025] The acquisition unit 1000 acquires log information TL to be analyzed from among multiple pieces of log information TL recorded and accumulated in the storage 105. Here, it is assumed that the multiple pieces of log information TL are recorded in the storage 105 as files recorded with different file names. The acquisition unit 1000 acquires, for example, one piece of log information TL to be analyzed.

[0026] The estimation unit 1001 estimates the time series of work content at each time of the work machine 3 from the acquired log information TL.

[0027] (Processing Flow of the Work Content Estimation System) Hereinafter, a specific processing flow performed by the work content estimation system 10 will be described in detail with reference to FIGS.

[0028] 4 starts when a dedicated application is launched by an operator of the work content estimation system 10. When the dedicated application is launched by the operator's operation, the acquisition unit 1000 of the CPU 100 expands and acquires the designated log information TL1 designated as the target of analysis in the memory 101 (step S00).

[0029] Here, the log information TL will be described with reference to FIG.

[0030] As shown in Fig. 5, the log information TL includes work machine identification information. Specifically, the work machine identification information is an individual identification number for individually identifying the work machine 3. In Fig. 5, the work machine identification information is allocated so as to correspond to the vehicle type, model, type, serial number, etc. of the work machine 3, indicating a hydraulic excavator, wheel loader, bulldozer, etc. Note that the work machine identification information may be numbers, alphabetic characters, symbols, or a combination of these, in addition to numbers.

[0031] As shown in Figure 5, the log information TL includes information indicating the position and attitude of the work machine 3 at each time, and angle information of the movable parts of the work machine 3. Specifically, the log information TL records the position of the work machine 3, the roll angle of the work machine 3, which is the left-right tilt of the machine body, the pitch angle, which is the fore-and-aft tilt of the machine body, the slewing angle, boom angle, arm angle, and bucket angle for each time. Here, the data logger 20 mounted on the work machine 3 identifies and records the position of the work machine 3 based on positioning information indicating latitude, longitude, and altitude, which is information obtained by receiving from GNSS antennas G1, G2, for example. The data logger 20 also calculates and records the roll angle and pitch angle of the work machine 3 based on measurement results from an IMU (Inertial Measurement Unit) mounted on the work machine 3. The data logger 20 also calculates and records the rotation angle of the upper rotating body 32 based on positioning information obtained from each of the GNSS antennas G1 and G2 provided on the upper rotating body 32. The rotation angle is the change in the orientation of the upper rotating body 32 per predetermined time. The data logger 20 also calculates and records the boom angle, arm angle, and bucket angle based on the extension and retraction degrees of the boom cylinder SL1, arm cylinder SL2, and bucket cylinder SL3. The boom angle, arm angle, bucket angle, and rotation angle may be acquired, for example, by attaching IMUs to the boom, arm, bucket, and upper rotating body and using these IMUs.

[0032] The position, roll angle, and pitch angle are information necessary to identify the position and posture of the work machine 3 itself. The roll angle, pitch angle, swing angle, boom angle, arm angle, and bucket angle are posture information that represent the posture of the work machine 3.

[0033] 5 includes the position information, roll angle, pitch angle, swing angle, boom angle, arm angle, and bucket angle of the work machine 3. This information has the advantage that it is not easily affected by differences in the class of the work machine 3 as information used when estimating the work content.

[0034] Returning to FIG. 4, the estimation unit 1001 of the CPU 100 estimates the work content of the work machine 3 at each time based on the designated log information TL1 acquired in step S00 (step S01).

[0035] Here, the procedure by which the estimation unit 1001 estimates the work content of the work machine 3 from the log information TL will be described with reference to Figure 7. The estimation unit 1001 estimates the work content of the work machine 3 for both unit tasks and element tasks. A unit task is a task that accomplishes one work purpose. An element task is an element that makes up a unit task, and is a task that represents a series of actions or tasks categorized by purpose.

[0036] Examples of unit task classifications include "digging and loading," "plowing," "slope (from below)," "load collection," "traveling," and "parking / restoring" as shown in FIG. 7 , as well as "ditch digging," "backfilling," and "slope (from above)." Excavation and loading is the task of digging and scraping away soil or rocks and loading the scraped soil or rocks onto the bed of a transport vehicle. Excavation and loading is a unit task consisting of excavation, loading and unloading, soil removal, empty loading and waiting for soil removal, and bed holding. Plowing is the task of scraping away excess ground unevenness to a specified height. Plowing is a unit task consisting of excavation and soil removal, or excavation, loading and unloading, soil removal, and empty loading and may include leveling and brooming. Slope (from below) is the task of creating a slope using a work machine 3 positioned below the target area. Compacting a slope (from below) is a unit operation consisting of rolling, excavation, loaded rotation, soil removal, and empty rotation, and may include pushing and leveling. Load collection is the work of collecting soil and sand excavated by excavation, etc. before loading it onto a transport vehicle. Load collection is a unit operation consisting of excavation, loaded rotation, soil removal, and empty rotation, and may include pushing and leveling. Traveling is the work of moving the work machine 3. Traveling as a unit operation is a unit operation consisting of traveling as an element operation. Stopping / idling is a state in which the bucket BK is free of soil and rocks and is stopped for a predetermined period of time or more. Stopping / idling as a unit operation is a unit operation consisting of stopping as an element operation. Trench excavation is the work of digging a long, narrow trench in the ground and scraping it away. Trench excavation is a unit operation consisting of excavation, loaded rotation, soil removal, and empty rotation, and may include pushing and leveling. Backfilling is the process of filling existing trenches or holes in the ground with soil and sand to fill them flat. Backfilling is a unit operation consisting of excavation, loading and turning, soil removal, compaction, and empty turning, and may also include leveling and brooming. Slope (from above) is the process of creating a slope using a work machine 3 positioned above the target area. Slope (from above) is a unit operation consisting of compaction, excavation, loading and turning, soil removal, and empty turning, and may also include leveling.

[0037] Examples of classifications of elemental work include "excavation," "loaded rotation," "waiting for soil discharge," "soil discharge," "empty load rotation," and "load carrier holddown" shown in FIG. 7 , as well as "compaction," "leveling," and "brooming." Excavation is the work of digging up and scraping away soil or rocks using a bucket BK. Loaded rotation is the work of rotating the upper rotating body 32 while the bucket BK holds the scraped soil or rocks. Waiting for soil discharge is the work of waiting for a transport vehicle to load the scraped soil or rocks while the bucket BK holds the scraped soil or rocks. Soil discharge is the work of lowering the scraped soil or rocks from the bucket BK onto a transport vehicle or a predetermined location. Empty load rotation is the work of rotating the upper rotating body 32 when the bucket BK is empty of soil or rocks. Load carrier holddown is the work of pressing down the soil loaded on the load carrier of a transport vehicle from above with the bucket BK to level it. Compaction is the process of pushing soil and sand into disturbed ground with a bucket BK to shape and strengthen the ground. Leveling is the process of sweeping and leveling the soil and sand with the bottom of a bucket BK. Brooming is the process of sweeping and leveling the soil and sand with the side of a bucket BK.

[0038] The estimation unit 1001 obtains a time series of likelihoods related to a unit task using the unit task estimation model PM1. The unit task estimation model PM1 is a trained machine learning model that inputs, as an explanatory variable, external movement information, which is time series information corresponding to the external movement of the moving parts of the work machine 3, and outputs information that represents the estimation result of the work content of the work machine 3. Here, the external movement of the moving parts of the work machine 3 refers to movement that is visible from outside the moving parts of the work machine 3. The external movement of the moving parts of the work machine 3 refers to the movement of the upper rotating body 32, the boom BM, the arm AR, the bucket BK, and the lower traveling body 3, for example.

[0039] The external operation information is information including the results of detection by sensors of the external operation of the movable parts, calculated values ​​based on the detection results, etc. For example, as shown in Fig. 6 , the external operation information D1 includes posture information D2, processing information D3, and other operation information D4. As described above, the posture information D2 is information that represents the posture of the work machine 3, and includes, for example, a boom angle D21, an arm angle D22, a bucket angle D23, a swing angle D24, a pitch angle D25, and a roll angle D26. Furthermore, the processing information D3 includes a derivative value D31 of the posture information D2, bucket cutting edge information D32, and a derivative value D33 of the bucket cutting edge information D32. The derivative value D31 of the posture information D2 includes a boom angular velocity D311, an arm angular velocity D312, a bucket angular velocity D313, and a swing angular velocity D314. The bucket cutting edge information D32 includes the bucket cutting edge position (height in the vertical direction relative to the vehicle body) D321, the bucket cutting edge position (distance in the horizontal direction relative to the vehicle body) D322, and the bucket cutting edge angle (angle relative to the vehicle body) D323.

[0040] The bucket cutting edge position (height in the vertical direction relative to the vehicle body) D321 can be calculated using the following equations (1) to (3).

[0041]

[0042]

[0043]

[0044] Here, boom_h, arm_h, and bucket_h are the heights of the boom, arm, and bucket, boom_deg, arm_deg, and bucket_deg are the angles of the boom, arm, and bucket, and BOOM, ARM, and BUCKET are the lengths of the boom, arm, and bucket, which are model-specific values.

[0045] The bucket cutting edge position (distance from the vehicle body in the horizontal direction) D322 can be calculated using the following equations (4) to (6).

[0046]

[0047]

[0048]

[0049] Here, boom_d, arm_d, and bucket_d are the distances of the boom, arm, and bucket to the vehicle body.

[0050] The cutting edge angle (angle relative to the vehicle body) D323 (edge_deg) can be calculated using the following formula (7).

[0051]

[0052] The derivative value D33 of the bucket cutting edge information D32 includes the bucket cutting edge speed (speed in a direction vertical to the vehicle body) D331, the bucket cutting edge speed (speed in a direction horizontal to the vehicle body) D332, and the bucket cutting edge angular velocity (angular velocity to the vehicle body) D333. The other operation information D4 includes, for example, the vehicle body travel distance D41. Note that the appearance operation information D1 may include all of the information exemplified in FIG. 6 or may not include all of it.

[0053] The external operation information D1 includes, for example, part or all of the attitude information D2 that indicates the attitude of the work machine shown in FIG. 6 , and part or all of the processing information D3 calculated based on the attitude information D2. That is, the attitude information D2 includes, for example, information that indicates at least one of the roll angle, pitch angle, swing angle, boom angle, arm angle, or bucket angle of the work machine 3. The processing information D3 calculated based on the attitude information D2 includes, for example, information that indicates at least one of the position, distance, speed, angle, or angular velocity of a movable part or work implement of the work machine 3. The information that is included in the processing information D3 and that indicates at least one of the position, speed, angle, or angular velocity and that is calculated based on the attitude information D2 includes information that indicates at least one of the angular velocity of a movable part, and the position, speed, angle, or angular velocity of the work implement of the work machine 3, calculated based on the roll angle, pitch angle, swing angle, boom angle, arm angle, or bucket angle of the work machine 3. The vehicle body travel distance D41 is information calculated based on the position information shown in FIG. 5. Other examples of operation information D4 include the moving speed of the vehicle body, the relative position (distance) of the working implement such as a bucket or blade with respect to the lower running body 31, the amount of change in the relative position (speed), the angle of the working implement with respect to the ground, and the amount of change in the angle with respect to the ground (angular velocity).

[0054] The unit task estimation model PM1 is a machine learning model that has been trained in advance by supervised learning or unsupervised learning. When external movement information based on the log information TL is input to the unit task estimation model PM1, the unit task estimation model PM1 outputs a likelihood related to the unit task as information representing the estimation result of the task content of the work machine 3. The output information may be stored in the storage 105, for example.

[0055] The estimation unit 1001 also obtains a time series of likelihoods related to element tasks by inputting the log information TL into the element task estimation model PM2 in chronological order. The element task estimation model PM2 is a trained machine learning model that has been trained by, for example, learning using teacher data or learning without using teacher data, and is a model that outputs likelihoods related to element tasks when external movement information based on the log information TL is input, and the output information may be stored in the storage 105, for example.

[0056] The estimation unit 1001 smooths the time series of likelihoods for unit tasks and element tasks by applying a time averaging filter to each of them, and generates a unit task heat map H1 representing the time series of likelihoods for the smoothed unit tasks and an element task heat map H2 representing the time series of likelihoods for the smoothed element tasks, as shown in FIG. 7 . The heat maps H1 and H2 are maps in which colors representing the likelihoods of task tasks are applied to a plane with task tasks on the vertical axis and time on the horizontal axis, based on the time series of smoothed likelihoods. For example, the colors of the heat maps may be closer to blue as the likelihood of a task task is lower, and closer to red as the likelihood of a task task is higher. The estimation unit 1001 stores the heat maps H1 and H2 in the storage 105.

[0057] The estimation unit 1001 identifies a time period during which the likelihood of a unit task is dominant, based on the time series of the smoothed likelihood, and estimates the work content of the work machine 3 for that time period. For example, in a time period during which the likelihood of the unit task "digging and loading" is dominant, the estimation unit 1001 estimates the work content of the work machine 3 to be "digging and loading". Similarly, the estimation unit 1001 identifies a time period during which the likelihood of an element task is dominant, based on the time series of the smoothed likelihood, and estimates the work content of the work machine 3 for that time period. For example, in a time period during which the likelihood of the element task "digging" is dominant, the estimation unit 1001 estimates the work content of the work machine 3 to be "digging". The estimation unit 1001 stores information on the work content R estimated for the work machine 3 in the storage 105, and displays it on the display unit 102 (step S02).

[0058] (Actions and Effects) As described above, the work content estimation system 10 according to the first embodiment comprises an acquisition unit 1000 that acquires log information TL of the work machine 3 associated with time, and an estimation unit 1001 that estimates the time series of the work content of the work machine 3 by inputting the appearance movement information D10 based on the log information TL into trained machine learning models (unit work estimation model PM1 and element work estimation model PM2) that input appearance movement information D10, which is time series information corresponding to the appearance movement of the moving parts of the work machine 3, and output information representing the estimation results of the work content of the work machine 3. With this configuration, there is no need to input information that is easily affected by differences in vehicle class, and therefore the versatility of the machine learning model used to estimate the work content can be improved.

[0059] (Modification) The contents of the log information TL according to the first embodiment are not limited to those in other embodiments. For example, if the work machine 3 is not a hydraulic excavator but a different vehicle type, log information TL according to that vehicle type is recorded. Examples of other vehicle types include a wheel loader, a bulldozer, etc.

[0060] Furthermore, the work content estimation system 10 according to the first embodiment has been described as being installed at a location away from the work machine 3 and connected to the data logger 20 mounted on the work machine 3 via a wide area communication network, but other embodiments are not limited to this configuration.

[0061] For example, in a work content estimation system 10 according to another embodiment, part or all of the configuration of the work content estimation system 10 may be installed inside the work machine 3. In this case, the data logger 20 may transmit the log information TL to the work content estimation system 10 via a network inside the work machine 3, rather than via a wide area communication network. In this way, the operator on board the work machine 3 can check the work content estimation results on the spot.

[0062] Note that the work content estimation system 10 installed inside the work machine 3 may acquire log information TL of other work machines 3 via a wide area communication network or the like. In this way, it is possible to estimate the work content of work machines 3 other than the work machine 3 on which the work content estimation system 10 is installed.

[0063] The various processes of the above-described task content estimation system 10 are stored in the form of a program on a computer-readable recording medium, and the above-described various processes are performed by a computer reading and executing the program. The computer-readable recording medium refers to a magnetic disk, a magneto-optical disk, a CD-ROM, a DVD-ROM, a semiconductor memory, etc. Alternatively, the computer program may be distributed to a computer via a communication line, and the computer that receives the program may execute the program.

[0064] Furthermore, in the first embodiment, the unit task estimation model PM1 outputs a likelihood related to a unit task, and the element task estimation model PM2 outputs a likelihood related to an element task. However, for example, the unit task estimation model PM1 may input external movement information based on the log information TL and output both a likelihood related to a unit task and a likelihood related to an element task.

[0065] The program may be one that realizes part of the above-mentioned functions, or may be one that realizes the above-mentioned functions in combination with a program already recorded in the computer system, such as a so-called differential file or differential program.

[0066] Some or all of the functions of the work content estimation system 10 described above may be assigned to the work machine 3. For example, some or all of the functions of the acquisition unit 1000, the estimation unit 1001, the memory 101, the display unit 102, the operation reception unit 103, the communication interface 104, and the storage 105 may be assigned to the work machine 3.

[0067] Although several embodiments of the present disclosure have been described above, these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their modifications are included in the scope of the inventions described in the claims and their equivalents, as well as in the scope and spirit of the inventions.

[0068] (Additional Note) The task content estimation system 10 according to the present disclosure can be understood, for example, as follows.

[0069] (1) A work content estimation system 10 according to a first aspect includes an acquisition unit 1000 that acquires log information TL of a work machine 3 associated with time, and an estimation unit 1001 that estimates a time series of the work content of the work machine by inputting visual movement information D1 based on the log information into trained machine learning models (a unit work estimation model PM1 and an element work estimation model PM2) that input visual movement information D1, which is time-series information corresponding to the visual movement of a moving part of the work machine, and output information representing an estimation result of the work content of the work machine. According to this aspect and each of the following aspects, it is possible to increase the versatility of the machine learning model used to estimate work content.

[0070] (2) A work content estimation system 10 according to a second aspect is the work content estimation system 10 of (1), wherein the external movement information includes posture information representing the posture of the work machine and information representing at least one of position, speed, angle, or angular velocity calculated based on the posture information.

[0071] (3) A work content estimation system 10 according to a third aspect is the work content estimation system 10 according to (1) or (2), wherein the posture information includes information representing at least one of the roll angle, pitch angle, slewing angle, boom angle, arm angle, and bucket angle of the work machine.

[0072] (4) A work content estimation system 10 according to a fourth aspect is the work content estimation system 10 of (2) or (3), wherein the information representing at least one of the position, speed, angle, or angular velocity calculated based on the posture information includes information representing at least one of the angular velocity of the movable part, and the position, speed, angle, or angular velocity of the implement of the work machine, calculated based on the roll angle, pitch angle, slewing angle, boom angle, arm angle, or bucket angle of the work machine.

[0073] According to the display control device and display control method of the present disclosure, it is possible to improve the versatility of the machine learning model used to estimate the work content.

[0074] 1 Analysis support system, 10 Work content estimation system, 100 CPU, 1000 Acquisition unit, 1001 Estimation unit, 101 Memory, 102 Display unit, 103 Operation reception unit, 104 Communication interface, 105 Storage, 20 Data logger, 3 Work machine, H1 to H2 Heat map, TL Log information, PM1 Unit work estimation model, PM2 Element work estimation model, R Estimated work content

Claims

1. A work content estimation system comprising: an acquisition unit that acquires log information of a work machine associated with time; and an estimation unit that estimates a time series of the work content of the work machine by inputting visual movement information based on the log information into a trained machine learning model that inputs the visual movement information, which is time series information corresponding to the visual movement of a moving part of the work machine, and outputs information representing an estimation result of the work content of the work machine.

2. The work content estimation system according to claim 1, wherein the external movement information includes posture information representing the posture of the work machine and information representing at least one of position, speed, angle, or angular velocity calculated based on the posture information.

3. A work content estimation system as described in claim 2, wherein the posture information includes information representing at least one of the roll angle, pitch angle, swing angle, boom angle, arm angle and bucket angle of the work machine.

4. The work content estimation system according to claim 3, wherein the information representing at least one of the position, speed, angle, or angular velocity calculated based on the posture information includes information representing at least one of the angular velocity of the movable part, and the position, speed, angle, or angular velocity of a work implement of the work machine, calculated based on a roll angle, pitch angle, slewing angle, boom angle, arm angle, or bucket angle of the work machine.

5. A method for estimating work content, comprising: a step of acquiring log information of a work machine associated with time; and a step of estimating a time series of work content of the work machine by inputting visual movement information based on the log information into a trained machine learning model which inputs the visual movement information, which is time series information corresponding to the visual movement of a moving part of the work machine, and outputs information representing an estimation result of the work content of the work machine.

Images