JP2020030539A - Working vehicle determination device and working vehicle determination method - Google Patents

Abstract

To determine a working vehicle suitable to an operator.SOLUTION: A working vehicle determination device (1) comprises an evaluation part (103) determining evaluation for a result of a work when an operator works on with a fork lift using a learnt model that has been made to conduct machine learning of correlation between evaluation of a result of an operation and a combination of an operator, fork lift, and a work, and a working vehicle determination part (104) determining a fork lift that is made to be used by an operator based on the evaluation.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a work vehicle determination device that determines a work vehicle suitable for a worker, and the like.

  2. Description of the Related Art There is known a conventional technique for controlling a transfer schedule of a transport vehicle when a plurality of transport operations are performed in parallel while sharing a plurality of transport vehicles (see Patent Document 1).

JP-A-5-94211

  However, the above-described prior art focuses on improving the transfer efficiency by reviewing the transfer schedule in real time. For this reason, even if such a conventional technique is used, when a certain operation is assigned to the operator of the transport vehicle, the transport vehicle suitable for the operator cannot be determined.

  For this reason, conventionally, when a plurality of work vehicles exist as selection candidates, the worker using the work vehicle has selected the work vehicle used by himself from the candidates based on intuition. . As a result, if a work vehicle that is incompatible with the worker is selected (for example, a vehicle that is difficult for the worker to drive or has had an accident in the past), the efficiency of the work is reduced. However, there is a problem that safety, comfort, etc. may be reduced.

  The present invention has been made in view of the above problems, and has as its object to provide a work vehicle capable of determining a work vehicle suitable for a worker in terms of work efficiency, safety, comfort, and the like. Vehicle determination device and the like.

  In order to solve the above problems, a working vehicle determination device according to one aspect of the present invention includes an evaluation of a result of a work performed by a worker using the working vehicle, the worker, the working vehicle, , And using the learned model obtained by machine learning the correlation with the combination of the work, the worker evaluates the result of the predetermined work when performing the predetermined work using a certain work vehicle. An evaluation unit for determining is provided, and a work vehicle determining unit for determining the work vehicle to be used by the worker based on the evaluation determined by the evaluation unit.

  In addition, in order to solve the above-described problem, a work vehicle determination method according to one embodiment of the present invention includes an evaluation of a result of a work performed by a worker using a work vehicle, the work, and the worker, Vehicle, and using a learned model obtained by machine learning the correlation with the combination of the work, the worker performs a predetermined work using a certain work vehicle with respect to a result of the predetermined work. An evaluation step of determining an evaluation; and a work vehicle determination step of determining the work vehicle to be used by the worker based on the evaluation determined in the evaluation step.

  According to one embodiment of the present invention, a work vehicle suitable for a worker can be determined in terms of work efficiency, safety, comfort, and the like.

It is a block diagram showing an example of the important section composition of the working vehicle decision device concerning Embodiment 1 of the present invention. It is a figure showing the outline of the above-mentioned work vehicle decision device. It is a flowchart which shows an example of the process which the said work vehicle determination apparatus performs.

[Embodiment 1]
(Device overview)
An outline of the working vehicle determination device of the present embodiment will be described with reference to FIG. FIG. 2 is a diagram illustrating an outline of the work vehicle determination device 1. The work vehicle determination device 1 is a device that determines a work vehicle suitable for the worker with respect to a work performed by the worker operating the work vehicle. Hereinafter, an example in which the work vehicle is a forklift will be described. However, the work vehicle may be any vehicle used for work, and is not limited to a forklift.

  In the example of FIG. 2, a worker who performs a work using a forklift inputs a work code indicating the work and a worker ID which is identification information of the worker to the work vehicle determination device 1. The operations performed using a forklift include, for example, unloading from shelves and trucks, loading of loads on shelves and trucks, moving luggage, and a combination thereof (for example, performing both unloading and loading). And the like.

  Then, the work vehicle determination device 1 selects a forklift suitable for the worker from the selection candidates of the three forklifts based on the input work code and the worker ID. As described above, the worker only inputs his / her worker ID and work code to the work vehicle determination device 1 and causes the work vehicle determination device 1 to select an appropriate forklift according to the worker and the work. be able to.

(Main configuration of work vehicle determination device)
A configuration of a main part of the work vehicle determination device 1 will be described with reference to FIG. FIG. 1 is a block diagram illustrating an example of a main configuration of the work vehicle determination device 1. As shown in the drawing, the work vehicle determination device 1 includes a control unit 10 that integrally controls each unit of the work vehicle determination device 1, and a storage unit 20 that stores various data used by the work vehicle determination device 1. Have. Further, the work vehicle determination device 1 includes an input unit 30 that receives an input operation on the work vehicle determination device 1, and an output unit 40 that allows the work vehicle determination device 1 to output information.

  Further, the control unit 10 includes an input data generation unit 101, a candidate extraction unit 102, an evaluation unit 103, a work vehicle determination unit 104, a teacher data generation unit 105, and a learning unit 106. The storage unit 20 stores empty vehicle information 201. The empty vehicle information 201 is information indicating a currently available forklift, that is, a forklift that is a selection candidate. The empty vehicle information 201 is appropriately updated according to the use situation of the forklift.

  The input data generation unit 101 generates input data to be input to the evaluation unit 103. Although the details will be described later, the input data is data indicating a worker who uses the forklift, a forklift that is a selection candidate, and work to be performed using the forklift.

  The candidate extracting unit 102 extracts a forklift as a selection candidate. Specifically, the candidate extracting unit 102 extracts a currently available forklift as a selection candidate based on the empty vehicle information 201 stored in the storage unit 20. In the present embodiment, an example will be described in which a forklift as a selection candidate is identified by a vehicle ID. It should be noted that, even for the same vehicle type, the operational feeling and the like slightly differ from vehicle to vehicle depending on past traveling history and the like, and the difference may affect the ease of driving of the forklift. For this reason, it is preferable to assign a unique vehicle ID to each forklift.

  The evaluation unit 103 uses a learned model in which the correlation between the evaluation of the result of the work performed by the worker using the forklift and the worker, the forklift, and the combination of the work is machine-learned. When a worker performs a predetermined operation using a certain forklift, an evaluation on the result of the predetermined operation is determined. Details of the learned model will be described later.

  The work vehicle determination unit 104 determines a forklift to be used by the worker based on the evaluation determined by the evaluation unit 103. Specifically, the work vehicle determination unit 104 determines a forklift having the highest evaluation by the evaluation unit 103 among the forklifts to be selected as a forklift to be used by the worker.

  Further, the work vehicle determination unit 104 causes the output unit 40 to output the determined forklift. This allows the worker to recognize a forklift suitable for the worker. Note that the output mode of the output unit 40 is not particularly limited. For example, if the output unit 40 is a display device, display output may be performed, and if the output unit 40 is a sound output device, audio output may be performed. Further, the work vehicle determination unit 104 may notify the portable terminal owned by the worker of the determined forklift, for example. If the forklift can be moved remotely, the work vehicle determining unit 104 may move the determined forklift to the worker.

  The teacher data generation unit 105 generates teacher data used for updating the learned model. The learning unit 106 updates the learned model using the teacher data. Details of the learned model will be described below, and generation of teacher data and updating of the learned model will be described later.

(About the trained model)
The learned model used by the evaluation unit 103 for evaluation will be described. In the following, an example in which a trained model of a neural network (NN: Neural Network) is used will be described. However, if a trained model that can be used for evaluating a forklift as a selection candidate can be constructed, other models can be used. It is also possible to apply the following algorithm. When the NN trained model is used, it is preferable to use a multilayer NN trained model capable of performing highly accurate evaluation.

  The generation of the trained model includes, as teacher data, information on the work performed using the forklift in the past, evaluation of the result of the work, the forklift used for the work, the operator thereof, and information indicating the work. The associated data can be used. Among these data, the information indicating the used forklift, its operator and the work is input data, and the evaluation is correct answer data.

  Hereinafter, an example will be described in which the result of the work is evaluated in four stages of “A” to “D” (“A” has the highest evaluation and “D” has the lowest evaluation). Specifically, the evaluation of any one of "A" to "D" is determined based on at least one of the following (1) to (4) for work performed using a forklift in the past. Then, the forklift used for the work, the operator thereof, and information indicating the work are used as input data, and teacher data is used in which the above-described evaluation determined for the work is used as correct data. The four-level evaluation is an example, and the evaluation may be performed in less than three levels, five or more levels, or a numerical value (evaluation value).

(1) Whether there is an accident in the work,
(2) Details of the accident in the work,
(3) The operation contents of the fork list in the work (the amount of sudden turning, rapid acceleration / deceleration, etc.),
(4) Time from the start to the end of the work In this way, a forklift suitable for the worker to perform the work can be determined from at least one of the above points (1) to (4). . For example, when the evaluation based on the above (1) is adopted, a forklift having the lowest possibility of causing an accident can be determined from among the candidate forklifts in view of the result of the past work.

  When evaluating based on the above (1), for example, if there is no accident, it may be evaluated as “A”, and if there is an accident, it may be evaluated as “D”. The evaluation may be performed in consideration of the number or frequency of accidents. For example, the evaluation of “A” to “D” may be determined according to the number of accidents per year.

  In the case of evaluation based on the above (2), for example, if the content of the accident is minor, it may be evaluated as "B", and if the content of the accident is serious, it may be evaluated as "D". Alternatively, the evaluation may be made based on whether or not the cause of the accident is a forklift. For example, a collision accident caused by an erroneous operation may be evaluated as “D” because there is a possibility that use of a forklift that is difficult for the operator to operate is one of the causes. On the other hand, if the accident is a collision with another forklift while the vehicle is stopped, the cause of the accident is not considered to be the forklift, so the evaluation may be “B”.

  When the evaluation is performed based on the above (3), for example, the evaluation may be performed based on at least one of the number of times of rapid acceleration, the number of times of sudden deceleration, the number of times of sudden turning, and the number of erroneous operations. For example, if the number of times is equal to or more than the threshold, the evaluation may be “D”, and if the number is less than the threshold, the evaluation may be “A” to “C”. It should be noted that any of the evaluations “A” to “C” may be determined based on the threshold.

  In the case of evaluation based on the above (4), for example, if the time is significantly shorter than the standard work time of the work, "A" is evaluated, and if the work time is about the same as the standard work time, "A" is evaluated. B ", and if it is significantly longer than the standard operation time, it may be evaluated as" C ". Further, the evaluation may be determined in consideration of a plurality of the above (1) to (4).

  By generating such teacher data for work performed using a forklift in the past, and performing machine learning using the generated teacher data, it is possible to generate a learned model used by the evaluation unit 103. it can.

  Then, when the input data generated by the input data generation unit 101 is input to the trained model, the probability corresponding to each of the evaluations “A” to “D” is output. Then, the evaluation unit 103 determines the evaluation having the highest probability as the evaluation corresponding to the input data. For example, when the probabilities corresponding to “A” to “D” are 80%, 15%, 3%, and 2%, respectively, the evaluation unit 103 determines the evaluation as “A”.

(Processing flow)
The flow of processing executed by the work vehicle determination device 1 will be described with reference to FIG. FIG. 3 is a flowchart illustrating an example of a process (work vehicle determination method) executed by the work vehicle determination device 1.

  First, the input unit 30 receives an input of an ID of a worker who operates a forklift (S1). The input unit 30 receives an input of a work code indicating the work content assigned to the worker (S2).

  Next, the candidate extracting unit 102 extracts a plurality of forklift selection candidates currently available to the worker with reference to the empty vehicle information 201 (S3). Specifically, the candidate extracting unit 102 extracts the vehicle ID of the selected forklift from the empty vehicle information 201.

  Then, the input data generation unit 101 generates input data including a combination of the worker ID, the work code, and the vehicle ID for each of the above selection candidates (S4). Then, the input data generation unit 101 inputs the generated input data to the evaluation unit 103. The format of the input data is the same as the teacher data used for the machine learning. For example, when the input data in the teacher data is data obtained by combining a vehicle ID of a forklift, an operator ID, and an operation code, the input data generation unit 101 outputs the vehicle ID, the operator ID, and the operation code. Generate combined input data.

  Next, the evaluation unit 103 determines the evaluation of each selection candidate based on the output value of the learned model (S5, evaluation step). Specifically, the evaluation unit 103 determines the evaluation of each selection candidate to be one of “A” to “D”.

  Finally, the work vehicle determination unit 104 assigns the highest evaluation selection candidate to the worker (S6, work vehicle determination step). Thus, the process of FIG. 3 ends. Note that the work vehicle determination unit 104 may assign a selection candidate whose evaluation is equal to or higher than the reference to the worker. In this case, if there are a plurality of selection candidates having evaluations equal to or higher than the reference, the work vehicle determination unit 104 may present the plurality of selection candidates to the operator and allow the operator to select a desired selection candidate.

  In addition, when assigning a selection candidate whose evaluation is equal to or higher than the reference to the worker, the evaluation may be sequentially determined for a plurality of selection candidates, and the evaluation may be terminated when the evaluation equal to or higher than the reference is determined. As a result, it is possible to quickly complete the assignment of the selection candidates without determining the evaluation of all the selection candidates.

(About generating teacher data and updating trained models)
After the work using the forklift of the selection candidate determined as described above is performed, the teacher data generation unit 105 evaluates the work and the input data generated by the input data generation unit 101 for the selection candidate. To generate teacher data. Note that the above evaluation may be input via the input unit 30. In addition, the teacher data generation unit 105 may determine the evaluation by analyzing the work content using the forklift. For example, an acceleration sensor is mounted on the forklift, and the teacher data generation unit 105 receives the output value of the acceleration sensor, specifies the number of times of rapid acceleration or deceleration from the output value, and The evaluation may be determined. As a result, teacher data can be generated without using manual labor.

  As described above, since the teacher data generation unit 105 generates the teacher data in accordance with the execution of the work by the forklift determined by the work vehicle determination unit 104, the teacher data increases each time the forklift is determined and the work is performed. Go. The learning unit 106 updates the learned model of the evaluation unit 103 by using the teacher data every time a predetermined number of the teacher data generated in this way is accumulated. Thereby, the accuracy of the evaluation by the evaluation unit 103 can be maintained or improved.

(Summary of Embodiment 1)
As described above, the work vehicle determination device 1 causes the machine learning of the correlation between the evaluation of the result of the work performed by the worker using the forklift and the worker, the forklift, and the combination of the work. An evaluation unit 103 is provided for determining the evaluation of the result of the predetermined work when the worker performs a predetermined work using a certain forklift using the learned model. Further, the work vehicle determination device 1 includes a work vehicle determination unit 104 that determines the work vehicle to be used by the worker based on the evaluation determined by the evaluation unit. Therefore, according to the work vehicle determination device 1, a work vehicle suitable for the worker can be determined.

[About distributed processing]
Part of the processing performed by the work vehicle determination device 1 may be performed by one or more devices that are connected to and communicate with the work vehicle determination device 1. For example, the processing performed by the evaluation unit 103 may be executed by an AI server that is communicatively connected to the work vehicle determination device 1. In this case, the work vehicle determination device 1 generates input data, transmits the input data to the AI server, receives output data from the AI server, and determines a forklift suitable for the worker.

[About worker identification and selection candidate detection]
In the above embodiment, the worker is specified by receiving the input of the worker ID, but the method of specifying the worker is not limited to this example. For example, the work vehicle determination device 1 may specify a worker by analyzing an image of the worker. In this case, it is preferable to use, for the image analysis, a learned model constructed by machine learning using an image of the worker as teacher data. Particularly, it is preferable to use a trained model of a convolutional neural network (CNN: Convolutional Neural Network) because it is possible to specify a worker with high accuracy.

  In the above-described embodiment, an example in which the selection candidate forklift is specified (extracted) from the empty vehicle information 201 has been described. However, the specification method of the selection candidate forklift is not limited to this example. For example, a candidate forklift may be specified by analyzing an image of a parking space of a forklift where a currently available forklift is parked. In this case, for the image analysis, it is preferable to use a learned model constructed by machine learning using an image of each forklift as teacher data. As a result, a parked forklift can be detected from the image of the parking space, and the vehicle ID of the detected forklift can be specified. Also for this learned model, it is preferable to apply CNN from the viewpoint of specific accuracy. When an image of a parking space includes a worker, both the worker and the selection candidate forklift can be specified from the image.

[Example of software implementation]
A control block (particularly, each unit included in the control unit 10) of the work vehicle determination device 1 may be realized by a logic circuit (hardware) formed in an integrated circuit (IC chip) or the like, or may be realized by software. You may.

  In the latter case, the work vehicle determination device 1 includes a computer that executes instructions of a program that is software for realizing each function. This computer includes, for example, one or more processors and a computer-readable recording medium storing the program. Then, in the computer, the object of the present invention is achieved by the processor reading the program from the recording medium and executing the program. As the processor, for example, a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit) can be used. Examples of the recording medium include “temporary tangible medium” such as a ROM (Read Only Memory), a tape, a disk, a card, a semiconductor memory, and a programmable logic circuit. Further, a RAM (Random Access Memory) for expanding the program may be further provided. Further, the program may be supplied to the computer via an arbitrary transmission medium (a communication network, a broadcast wave, or the like) capable of transmitting the program. Note that one embodiment of the present invention can also be realized in the form of a data signal embedded in a carrier wave, in which the program is embodied by electronic transmission.

  The present invention is not limited to the embodiments described above, and various modifications are possible within the scope shown in the claims, and embodiments obtained by appropriately combining technical means disclosed in different embodiments. Is also included in the technical scope of the present invention.

1 work vehicle determination device 103 evaluation unit 104 work vehicle determination unit

Claims (3)

Using the learned model obtained by machine learning the correlation between the evaluation of the result of the work performed by the worker using the work vehicle and the correlation between the worker, the work vehicle, and the combination of the work, An evaluation unit that determines an evaluation for a result of the predetermined work when the worker performs a predetermined work using a certain work vehicle,
A work vehicle determining unit that determines the work vehicle to be used by the worker based on the evaluation determined by the evaluation unit,
A work vehicle determination device, comprising: Evaluation of the result of the work performed by the worker using the work vehicle,
(1) Whether there is an accident in the work,
(2) Details of the accident in the work,
(3) the operation content of the work vehicle in the work,
(4) determined based on at least one of the time from the start to the end of the work;
The work vehicle determination device according to claim 1, wherein: Using the learned model obtained by machine learning the correlation between the evaluation of the result of the work performed by the worker using the work vehicle and the correlation between the worker, the work vehicle, and the combination of the work, An evaluation step of determining an evaluation for a result of the predetermined work when the worker performs a predetermined work using a certain work vehicle,
A work vehicle determination step of determining the work vehicle to be used by the worker based on the evaluation determined in the evaluation step,
A work vehicle determination method, comprising:

Images