Prediction device and prediction method

Abstract

To predict the working time of sorting work for taking out merchandise and sorting it for each shipping destination.SOLUTION: The prediction device 100 includes a prediction unit 113 that predicts a work time of sorting work, which is work of taking out and sorting articles from a storage that stores the articles, based on a plurality of orders related to the articles (commodities), and calculates the work time as a predicted work time. The prediction unit 113 calculates the predicted working time based on the input information including the type of article to be taken out from the storage, the number of articles to be taken out from the storage, and the number of articles included in each order. The storage may be a storage shelf that the transport robot carries to a work place of the sorting work. The item may be transported along a predetermined path to a work location of a sorting operation. Further, a worker who performs the sorting work may move to a place where the storage is placed and perform the sorting work.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to a prediction device and a prediction method for predicting the time required to sort items. [Background technology]

[0002] With the spread of e-commerce (online sales), it has become commonplace for ordered products to be delivered on the same day or the next day. As a result, logistics warehouses, which store, sort, pack, and ship products, are required to improve the efficiency of each task. The sorting process, which involves taking ordered products from storage and sorting them by destination, is still done manually, with little automation. Due to a serious labor shortage, there is a need to improve the efficiency of sorting work.

[0003] To improve the efficiency of sorting work, it is possible to allocate appropriate orders to workers who will perform the sorting work. For example, if orders with similar order contents are allocated to the same worker, it is thought that the sorting work time will be shortened. In other words, if it is possible to predict the time required for each sorting work, it is thought that it will be possible to allocate orders in a way that minimizes the total sorting work time.

[0004] A picking work management device described in Patent Document 1 is a technology for predicting the work time for sorting work. The picking work management device includes a destination information acquisition unit that acquires information indicating the location where the product is stored, which is the destination location to which the worker will be moved during the picking work. The picking work management device also includes a travel time prediction unit that predicts the travel time required for picking work based on the information indicating the location acquired by the destination information acquisition unit. The picking work management device also includes a post-travel work time prediction unit that predicts the post-travel work time required for the post-travel work to be performed by the worker after moving to the location. The picking work management device also includes a target time calculation unit that calculates a target time for the picking work based on the travel time predicted by the travel time prediction unit and the post-travel work time predicted by the post-travel work time prediction unit. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 2023-005770 Summary of the Invention [Problem to be solved by the invention]

[0006] The picking operation management device described in Patent Document 1 can predict the time it takes to retrieve (pick) products from a storage location. However, the time it takes to sort the retrieved products, including the time it takes to sort them by delivery destination (orderer, customer), is outside the scope of the prediction. The present invention has been made in view of the above background, and an object of the present invention is to provide a prediction device and a prediction method for predicting the work time required for sorting work, which involves picking out products and sorting them according to their delivery destinations. [Means for solving the problem]

[0007] In order to solve the above-mentioned problems, the prediction device of the present invention includes a prediction unit that predicts the working time of sorting work, which is the work of removing and sorting items from a storage warehouse that stores the items based on multiple orders for the items, and calculates the predicted working time, and the prediction unit calculates the predicted working time based on input information including the type of items to be removed from the storage warehouse, the number of items to be removed from the storage warehouse, and the number of items included in each order. [Effects of the Invention]

[0008] According to the present invention, it is possible to provide a prediction device and a prediction method for predicting the work time required for sorting work, which involves picking up products and sorting them according to their destinations. Problems, configurations, and effects other than those described above will become clear from the description of the following embodiments. [Brief explanation of the drawings]

[0009] [Figure 1]FIG. 1 is a plan view of a logistics warehouse where sorting work is carried out according to a first embodiment. [Figure 2] FIG. 2 is a diagram for explaining a sorting operation in a sorting area according to the first embodiment. [Figure 3] FIG. 1 is a functional block diagram of a prediction device according to a first embodiment. [Figure 4] FIG. 2 is a data configuration diagram of an order delivery information database according to the first embodiment. [Figure 5] FIG. 2 is a data configuration diagram of learning data according to the first embodiment. [Figure 6] 10 is a flowchart of a batch generation method selection process according to the first embodiment. [Figure 7] FIG. 10 is a data configuration diagram of an order delivery information database according to a modified example of the first embodiment. [Figure 8] FIG. 10 is a data configuration diagram of learning data according to a modified example of the first embodiment. [Figure 9] FIG. 10 is a diagram for explaining a sorting operation according to the second embodiment. [Figure 10] FIG. 10 is a data configuration diagram of an order delivery information database according to the second embodiment. [Figure 11] FIG. 11 is a data configuration diagram of learning data according to the second embodiment. [Figure 12] FIG. 11 is a diagram for explaining a sorting operation according to the third embodiment. [Figure 13] FIG. 11 is a data configuration diagram of an order delivery information database according to the third embodiment. [Figure 14] FIG. 11 is a data configuration diagram of learning data according to the third embodiment. [Figure 15] FIG. 11 is a data configuration diagram of an order delivery information database according to a modified example of the third embodiment. [Figure 16] FIG. 11 is a data configuration diagram of learning data according to a modified example of the third embodiment. [Figure 17] FIG. 11 is a data configuration diagram of learning data according to a modified example of the third embodiment. [Figure 18] FIG. 10 is a data configuration diagram of learning data according to a modified example of the first embodiment. [Figure 19]FIG. 10 is a hardware configuration diagram illustrating an example of a computer that realizes the functions of the prediction device according to the above-described embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, embodiments of the present invention will be described with reference to the drawings. The examples are illustrative for explaining the present invention, and appropriate omissions and simplifications have been made for clarity of explanation. The present invention can be implemented in various other forms. Unless otherwise specified, each component may be singular or plural. The position, size, shape, range, etc. of each component shown in the drawings may not represent the actual position, size, shape, range, etc., in order to facilitate understanding of the invention. Therefore, the present invention is not necessarily limited to the position, size, shape, range, etc. disclosed in the drawings.

[0011] Although various types of information may be described using expressions such as "table," "list," and "queue" as examples, the various types of information may also be expressed using data structures other than these. For example, various types of information such as "XX table," "XX list," and "XX queue" may also be referred to as "XX information." When describing identification information, expressions such as "identification information," "identifier," "name," "ID," and "number" are used, but these are interchangeable.

[0012] When there are multiple components having the same or similar functions, they may be described by using the same reference numeral with different subscripts. When there is no need to distinguish between these multiple components, the subscripts may be omitted.

[0013] <<Outline of the prediction device>> An overview of a prediction device in a mode (embodiment) for carrying out the present invention will be described below. The prediction device predicts the work time required for sorting work, for example, in a logistics warehouse (fulfillment center), in which stored products (items) are removed and sorted according to delivery destination. The products are stored in storage boxes placed on shelves of a storage shelf (storage warehouse) to which a transport robot brings them, for example. A worker removes the products from the storage boxes and places them in a sorting box for each delivery destination. The sorting boxes are placed on shelves of the sorting shelf.

[0014] The prediction device predicts the work time using input information such as the product, the number of products to be picked, and the number of products for each shipping destination. The input information may also include the position of the product storage box on the product storage shelf and the position of the sorting box on the sorting shelf. Machine learning technology is used for the prediction.

[0015] Such a prediction device makes it possible to predict the time required to sort products. When evaluating an algorithm for assigning orders to workers, an evaluation value can be calculated using the work time predicted by the prediction device (predicted work time). The evaluation value is, for example, the total work time required for sorting work for all orders. By assigning orders to workers using the algorithm with the best evaluation value, efficient sorting work and, ultimately, efficient operation of the logistics warehouse can be achieved.

[0016] <Sorting work> Before describing the configuration of the prediction device 100 (see FIG. 3 described later), the sorting work will be described. FIG. 1 is a plan view of a logistics warehouse 400 where the sorting work according to the first embodiment is performed. FIG. 2 is a diagram for explaining the sorting work in a sorting area 450 according to the first embodiment.

[0017] The warehouse management system 210 (see FIG. 3 described later) is an information processing system that assigns to workers the task of sorting products to delivery destinations according to the order details (sorting task). The warehouse management system 210 also instructs the transport robot 440 to transport the storage shelf 420 (storage warehouse) containing the products from the storage area 410 to the sorting area 450 in accordance with the timing when the worker sorts the products. The storage area 410 is the area (location) where the storage shelf 420 is located. The storage shelf 420 containing the products to be sorted is transported to the sorting area 450 when the products are sorted, and is returned to the storage area 410 after the sorting task is completed.

[0018] 1, the storage shelves 420 being transported by the transport robot 440 and the storage shelves 420 in the sorting area 450 are hatched. The transport robot 440 is, for example, an AMR (Autonomous Mobile Robot) or an AGV (Automatic Guided Vehicle). The transport robot 440 crawls into the space below the storage shelves 420 and lifts and transports the storage shelves 420.

[0019] Once a sorting task is assigned, the worker picks up a sorting box 465 carried by the belt conveyor 470 and places it on the sorting shelf 460. The sorting box 465 is a box prepared for each delivery destination, and the sorted products are placed in the sorting box 465. As will be described later, the worker removes the products from the storage box 425 and places them in the sorting box 465. The position of the sorting shelf 460 on which the sorting box 465 is placed may be specified by the warehouse management system 210, and the worker places the sorting box 465 in the specified position on the sorting shelf 460. The position of the sorting shelf 460 on which the sorting box 465 is placed is indicated by a row and a column. In addition to the row and column, the position of the sorting box 465 on the sorting shelf 460 may also be indicated by including the surface of the sorting shelf 460.

[0020] Incidentally, the sorting shelf 460 illustrated in FIG. 2 has four shelves in the vertical direction on which sorting boxes 465 are placed. Each shelf of the sorting shelf 460 can accommodate a total of six sorting boxes 465, three in the left-right direction and two in the depth direction. This is the same for the storage shelf 420. The number of shelves and the width of each shelf (the number of storage boxes 425 and sorting boxes 465 that can be placed on a shelf) are set appropriately depending on the use, purpose, etc. The dimensions of the storage boxes 425 and sorting boxes 465 placed on the shelves are also set appropriately depending on the use, purpose, etc.

[0021] When the storage shelf 420 is carried to the sorting area 450, the worker presses a button on the work recording device 480 to start the sorting work. The worker removes the products from the storage boxes 425 on the storage shelf 420 and places them in the sorting boxes 465 on the sorting shelf 460. The position of the storage boxes 425 on the storage shelf 420 is also indicated by row and column, and in some cases by surface (for example, the front, back, left, or right surfaces of the storage shelf 420 or the sorting shelf 460).

[0022] The worker may simultaneously remove products for multiple delivery destinations. For example, when sorting two products for each of three delivery destinations, the worker may remove six products from the storage box 425 and place them two by two into three sorting boxes 465.

[0023] When sorting is complete, the worker presses a button on the work recording device 480. The worker places the sorting box 465 after sorting onto the belt conveyor 470. The transport robot 440 also returns the storage shelf 420 to the storage area 410.

[0024] As described above, the storage (see storage shelf 420) is the storage shelf 420 that the transport robot 440 transports to the work location for the sorting work (see sorting area 450). The sorted items (products) are placed in sorting boxes 465 placed on the sorting shelves 460.

[0025] <Configuration of prediction device> 3 is a functional block diagram of a prediction device 100 according to the first embodiment. The prediction device 100 is a computer and includes a control unit 110, a storage unit 120, and an input / output unit 180. User interface devices such as a display, keyboard, and mouse are connected to the input / output unit 180. The input / output unit 180 includes a communication device, and is capable of transmitting and receiving data to and from devices such as the warehouse management system 210 and the work recording device 480. A media drive may also be connected to the input / output unit 180, enabling data exchange using a recording medium.

[0026] The storage unit 120 is configured to include storage devices such as a read-only memory (ROM), a random access memory (RAM), and a solid-state drive (SSD). The storage unit 120 stores an order delivery information database 130, a sorting time database 140, a prediction model 121, learning data 150, and a program 128. The program 128 includes descriptions of processes to be executed by functional units of the control unit 110, which will be described later. Note that the various storage contents of the storage unit 120 may be stored in an external storage device, such as a cloud server, and read as needed.

[0027] <<Storage section: Order delivery information database>> FIG. 4 is a data configuration diagram of the order delivery information database 130 according to the first embodiment. The order delivery information database 130 is tabular data and includes details of the sorting work for ordered items, generated by the warehouse management system 210. Each row (record) of the order delivery information database 130 includes identification information for the record, the product name, the number of items to be removed from the storage box 425, and identification information for the sorting work (referred to as "batch" in FIG. 4). The record also includes identification information for the storage shelf 420 (referred to as "ID" in FIG. 4), the position (side, row, column) of the storage box 425 on the storage shelf 420, identification information for the sorting shelf 460 (referred to as "ID" in FIG. 4), and the position (side, row, column) of the sorting box 465 on the sorting shelf 460. The record also includes identification information for the shipping destination, the number of items to be placed in the sorting box 465 (referred to as "quantity" in FIG. 4), and identification information for the worker. A "batch" is a collection of sorting tasks that one worker is responsible for at one time. In Figure 4, batch "B34" is a collection of sorting tasks for product "A," "B," "C," and "D."

[0028] For example, the records with identification information "841", "842", and "843" indicate that 10 products with product name "A" are to be taken out from the storage box 425 on the "1" side, "1" row, and "1" column of the storage shelf 420 with identification information "1", and that they are to be sorted and placed into sorting boxes 465 as follows: Two products are to be placed into the sorting box 465 on the "1" side, "2" row, and "1" column of the sorting shelf 460 with identification information "1" (the shipping destination for "D81"); Three products are to be placed into the sorting box 465 on the "2" side, "3" row, and "1" column of the sorting shelf 460 with identification information "1" (the shipping destination for "D82"); Five products are to be placed into the sorting box 465 on the "3" side, "2" row, and "1" column of the sorting shelf 460 with identification information "1" (the shipping destination for "D83").

[0029] <Memory section: sorting time database> Returning to FIG. 3, the description of the memory unit 120 continues. The sorting time database 140 stores the work time of the sorting work measured by the measurement unit 111, which will be described later. In the first embodiment, the work time of the sorting work for each product is recorded in association with the product name and the sorting work (see "batch" in FIG. 4). In FIG. 4, worker "W34" sorts products "A", "B", "C", and "D" into delivery destinations "D81", "D82", "D83", and "D84". In this case, the sorting time for each of products "A", "B", "C", and "D" is recorded in the sorting time database 140.

[0030] <Memory section: Prediction model> The prediction model 121 is a machine learning model used to predict the time required to sort products. The explanatory variables of the prediction model 121 are the product name, the position (side, row, column) of the storage box 425 on the storage shelf 420, the number of products to be removed from the storage box 425, and the number of products by sorting destination. The sorting destination is indicated by the position (side, row, column) of the sorting box 465 on the sorting shelf 460. The objective variable of the prediction model 121 is the work time required for the sorting work.

[0031] <Memory section: learning data> Fig. 5 is a data configuration diagram of the training data 150 according to the first embodiment. The training data 150 is training data for the prediction model 121. For example, the training data in the first row in Fig. 5 is data obtained from the sorting work of product "A" indicated by three records with identification information "841," "842," and "843" in the order delivery information database 130 (see Fig. 4).

[0032] <Prediction device: control unit> 3, the control unit 110 will be described. The control unit 110 is configured to include a CPU (Central Processing Unit) and is equipped with a measurement unit 111, a learning unit 112, a prediction unit 113, a batch generation unit 114, and a batch generation method selection unit 115. The control unit 110 may be configured to include a GPU (Graphics Processing Unit), an NPU (Neural (network) Processing Unit), an FPGA (Field Programmable Gate Array), an ASIC (Application Specific Integrated Circuit), etc.

[0033] <Control unit: Measurement unit> The measurement unit 111 measures the work time of the sorting work and records it in the sorting time database 140. The measurement unit 111 measures the work time based on the time when the button on the work recording device 480 is pressed and records it in the sorting time database 140.

[0034] As described above, the prediction device 100 includes a measurement unit that measures the task time.

[0035] <Control unit: Learning unit> The learning unit 112 generates learning data 150 (see FIG. 5) based on the order delivery information database 130 (see FIG. 4) and the sorting time database 140. Next, the learning unit 112 trains and generates a prediction model 121 using the learning data 150.

[0036] <Control unit: Prediction unit> The prediction unit 113 predicts the work time for the sorting work using the prediction model 121 based on the product name, the position of the storage box 425 on the storage shelf 420, the number of products to be removed from the storage box 425, and the number of products by sorting destination. The sorting destination is indicated by the position of the sorting box 465 on the sorting shelf 460. The work time predicted by the prediction unit 113 is also referred to as the predicted work time.

[0037] As described above, the prediction device 100 includes a prediction unit 113 that predicts the work time of sorting work, which is the work of removing and sorting items from a storage warehouse (see storage shelf 420) that stores items based on multiple orders for the items (products), and calculates it as a predicted work time. The prediction unit 113 calculates the predicted work time based on input information (see explanatory variables of the prediction model 121 and explanatory variables of the learning data 150) including the type of item (product name) to be taken out of the storage warehouse, the number of items to be taken out of the storage warehouse, and the number of items included in each order.

[0038] The prediction unit 113 makes a prediction using a machine learning model (see the prediction model 121) in which the explanatory variables include input information and the objective variable is the work time. The input information includes the position where the item is stored on the storage shelf 420 (see the storage shelf in the learning data 150 shown in FIG. 5). The input information includes the position of the sorting box 465 on the sorting shelf 460 (see the number of sorts in the learning data 150). The input information includes identification information of the worker who will be performing the sorting work.

[0039] <Control unit: Batch generation unit> The batch generation unit 114 determines the sorting tasks to be assigned to workers based on one or more orders. There are one or more allocation methods (algorithms), and Figure 3 shows multiple batch generation units 114 corresponding to each method. Allocation methods include, for example, a random allocation method, or a method that minimizes the number of workers (total number, number of sorting tasks ("batches")) by keeping the number of shipping destinations below a predetermined number. In the following description, the allocation method and the batch generation unit 114 will be considered to be the same thing.

[0040] <Control unit: Batch generation method selection unit> The batch generation method selection unit 115 calculates the predicted work time for the entire sorting task (predicted total work time) based on the work time predicted by the prediction unit 113 for each sorting task when allocating one or more sorting tasks to workers. The batch generation method selection unit 115 selects the batch generation unit 114 that uses an algorithm that minimizes the predicted total work time. The warehouse management system 210 allocates sorting tasks to workers using this algorithm.

[0041] As described above, the prediction device 100 includes the batch generation unit 114 that allocates a plurality of orders to a plurality of workers who will perform the sorting work. The prediction device 100 also includes a batch generation method selection unit 115 that determines, from among the multiple batch generation units 114 included in the prediction device 100, the batch generation unit 114 with the shortest predicted total work time, which is the sum of the predicted work times for sorting work related to multiple orders.

[0042] <Batch generation method selection process> 6 is a flowchart of the batch generation method selection process according to the first embodiment. The batch generation method selection process is executed, for example, when the method for allocating sorting work is changed or a new method is devised, to select the best method (batch generation unit 114) (with the shortest predicted total work time) at a predetermined interval. The entire set of orders to be allocated is referred to as the total orders.

[0043] In step S11, the batch generation method selection unit 115 starts the process of repeating steps S12 to S17 for each batch generation unit 114. In step S12, the batch generation unit 114 generates batches by allocating the total orders to workers.

[0044] In step S13, the batch generation method selection unit 115 starts the process of repeating steps S14 to S16 for each batch generated in step S12. In step S14, the batch generation method selection unit 115 starts the process of repeating step S15 for each product included in the batch. In step S15, the prediction unit 113 predicts the work time required for sorting the products included in the batch.

[0045] In step S16, the batch generation method selection unit 115 calculates the sum of the work times calculated in step S15 to calculate the work time of the batch (estimated work time). In step S17, the batch generation method selection unit 115 sums up the predicted work times calculated in step S16 to calculate the work time (predicted total work time) for sorting work corresponding to the total orders. In step S18, the batch generation method selection unit 115 selects the batch generation unit 114 that minimizes the predicted total work time as the best.

[0046] <Features of the prediction device> The prediction device 100 can predict the work time required for sorting items using machine learning technology. When evaluating an algorithm that assigns orders to workers, the prediction device 100 can calculate a predicted total work time based on the work time predicted and use this as an evaluation value. By assigning sorting tasks to workers using the algorithm with the best evaluation value (batch generation unit 114), efficient sorting work and, ultimately, efficient operation of the logistics warehouse can be achieved.

[0047] <<Variation: Sorting box position>> The order delivery information database 130 (see FIG. 4) and the learning data 150 (see FIG. 5) include the positions of the sorting boxes 465 on the sorting shelves 460. The order delivery information database 130 and the learning data 150 may be configured not to include the positions of the sorting boxes 465 on the sorting shelves 460.

[0048] Fig. 7 is a data configuration diagram of the order delivery information database 130A according to a modified example of the first embodiment. Fig. 8 is a data configuration diagram of the learning data 150A according to a modified example of the first embodiment. The number of sorts in the learning data 150A shows only the number of items to be sorted from largest to smallest.

[0049] The explanatory variables of the prediction model 121 according to this modification are the product name, the position of the storage box 425 on the storage shelf 420, the number of products to be taken out of the storage box 425, and the number of products to be sorted. The learning unit 112 trains and generates the prediction model 121 using the learning data 150A.

[0050] <<Variation: Storage Box Location>> The learning unit 112 may generate training data that does not include the position of the storage bin 425 on the storage shelf 420 in the order delivery information database 130 (see FIG. 3). In this case, the explanatory variables of the prediction model 121 are the product name, the number of products removed from the storage bin 425, and the number of products by sorting destination. The sorting destination is indicated by the position (face, row, column) of the sorting bin 465 on the sorting shelf 460. The learning unit 112 trains and generates the prediction model 121 using training data that does not have the storage shelf attribute of the training data 150 (see FIG. 5).

[0051] <<Variation: Location of storage box and sorting box>> The learning unit 112 may generate learning data that does not include the storage shelf of the learning data 150A (see FIG. 8). In this case, the explanatory variables of the prediction model 121 are the product name, the number of products taken out from the storage box 425, and the number of items to be sorted. The learning unit 112 trains and generates the prediction model 121 using the learning data that does not include the storage shelf attribute of the learning data 150A (see FIG. 8).

[0052] Second Embodiment In the first embodiment, the products to be sorted are stored in storage boxes 425 on storage shelves 420 brought by a transport robot 440 (see FIG. 2). In the second embodiment, storage boxes 425B are carried by a belt conveyor 470B (see FIG. 9, which will be described later).

[0053] Second embodiment: sorting work 9 is a diagram illustrating a sorting operation according to the second embodiment. A worker removes products from a storage box 425B carried by a belt conveyor 470B and places them in a sorting box 465. The belt conveyor 470B is controlled by the warehouse management system 210. When the storage box 425B containing the products to be removed comes in front of the worker, the warehouse management system 210 instructs the worker to display the number of products to be removed on a display 480B of the DPS (Digital Picking System). The worker removes the displayed number of products, presses a button on the display 480B, and places the products in the sorting box 465.

[0054] Second Embodiment: Order Delivery Information Database Figure 10 is a data configuration diagram of the order delivery information database 130B according to the second embodiment. Compared to the order delivery information database 130 according to the first embodiment (see Figure 4), the order delivery information database 130B does not include identification information relating to storage shelves and sorting shelves, or attributes such as face, row, and column. However, since the sorting box 465 is located on the sorting shelf 460, the order delivery information database 130B may include attributes such as face, row, and column that indicate its position.

[0055] Second Embodiment: Training Data 11 is a data configuration diagram of learning data 150B according to the second embodiment. Unlike the learning data 150 according to the first embodiment (see FIG. 5), learning data 150B does not have a storage shelf attribute. The sorting number indicates only the number of items to be sorted into in descending order of size.

[0056] Second Embodiment: Learning Unit The explanatory variables of the prediction model 121 in the second embodiment are the product name, the number of products taken out of the storage box 425B, and the number of sorts. The learning unit 112 generates learning data 150B based on the order delivery information database 130B and the sorting time database 140. Next, the learning unit 112 trains and generates the prediction model 121 using the learning data 150B.

[0057] As described above, the articles (products) are transported along a predetermined route (see belt conveyor 470B in FIG. 9) to the sorting work area.

[0058] <Features of the second embodiment> In the sorting work in which the containers 425B are transported by the belt conveyor 470B, the prediction unit 113 can predict the work time for the sorting work of the products, as in the first embodiment.

[0059] Third Embodiment In the first and second embodiments, the storage shelves 420 storing the storage boxes 425 and the storage boxes 425B are carried by the transport robot 440 to a location where a worker is present (see sorting area 450). In the third embodiment, the worker moves to the location (area) where the products are stored, picks up the products, and sorts them.

[0060] Third embodiment: sorting work FIG. 12 is a diagram illustrating a sorting operation according to the third embodiment. A worker moves a sorting shelf 460C (a movable shelf with wheels) carrying a sorting box 465 to a location (area) where a storage box 425C is located. In FIG. 12, the storage box 425C is located on a storage shelf 420C (a fixed shelf) with two shelves. A display 480C is installed on each storage box 425C on the storage shelf 420C, and displays the number of products to be removed from the storage box 425C. The worker removes the displayed number of products, presses a button on the display 480C, and places the products in the sorting box 465. Note that the sorting shelf 460C may be carried by a transport robot instead of a worker.

[0061] Third Embodiment: Order Delivery Information Database 13 is a data configuration diagram of the order delivery information database 130C according to the third embodiment. Compared to the order delivery information database 130 according to the first embodiment (see FIG. 4), the order delivery information database 130C does not have identification information or attributes of faces and columns related to storage shelves and sorting shelves.

[0062] Third Embodiment: Training Data Fig. 14 is a data configuration diagram of learning data 150C according to the third embodiment. For example, the learning data in the first row in Fig. 14 is data obtained from the sorting work of product "A" indicated by the records with identification information "531" and "532" in Fig. 13. The storage shelf indicates the identification information, row, and column of storage shelf 420C.

[0063] Third Embodiment: Learning Unit The explanatory variables of the prediction model 121 in the third embodiment are the product name, the position of the storage box 425C (shelf identification information, row, and column), the number of products removed from the storage box 425C, and the number of products by sorting destination. The sorting destination is indicated by the position (row) of the sorting box 465 on the sorting shelf 460C. The learning unit 112 generates learning data 150C based on the order delivery information database 130C and the sorting time database 140. Next, the learning unit 112 trains and generates the prediction model 121 using the learning data 150C.

[0064] <Features of the third embodiment> Even if the worker moves to the location of the storage box 425C containing the products, removes the products, and sorts them, the prediction unit 113 can predict the working time for sorting the products, just as in the first embodiment.

[0065] <<Modification: Order delivery information database, learning data>> 15 is a data configuration diagram of an order delivery information database 130D according to a modification of the third embodiment. Unlike the order delivery information database 130C (see FIG. 13), the order delivery information database 130D does not have an attribute for the sorting shelf level (sorting shelf level).

[0066] 16 is a data configuration diagram of learning data 150D according to a modified example of the third embodiment. Unlike the learning data 150C (see FIG. 14), the learning data 150D does not have the attribute of "sorting shelf" that indicates the position of the sorting box 465. Furthermore, the sorting number indicates only the number of items to be sorted into, in descending order.

[0067] The order delivery information database 130D and training data 150D may be configured as described above. In this modification, the explanatory variables of the prediction model 121 are the product name, the position of the storage bin 425C (shelf identification information, row, and column), the number of products to be removed from the storage bin 425C, and the number of items to be sorted. The learning unit 112 generates the training data 150D based on the order delivery information database 130D and the sorting time database 140. Next, the learning unit 112 trains and generates the prediction model 121 using the training data 150D.

[0068] <<Variation: Sorting Shelf>> The learning data may further include the position of the sorting shelf 460C in the location (area) where the storage shelf 420C is located. The position is indicated, for example, by coordinates in the location where the storage shelf 420C is located. FIG. 17 is a data configuration diagram of learning data 150E according to a modified example of the third embodiment. Compared to learning data 150D (see FIG. 16), learning data 150E includes an attribute of the sorting shelf position.

[0069] The position of the storage box 425C (see FIG. 12) from which the product is taken and the position of the sorting shelf 460C are generally considered to be close to each other, but when taking products from multiple storage boxes 425C, they may be far apart. The explanatory variables of the prediction model 121 in this modified example include the position of the sorting shelf 460C. The learning unit 112 trains and generates the prediction model 121 using the learning data 150E. By using such a prediction model 121, it becomes possible to accurately predict the work time for the sorting task even when the storage box 425C and the sorting shelf 460C are far apart.

[0070] As described above, the worker who performs the sorting work moves to the location where the storage warehouse (see storage shelf 420C and storage box 425C in FIG. 12) is located and performs the sorting work. The input information (see the explanatory variables of the prediction model 121 and the explanatory variables of the training data 150C) includes the storage facility (see the identification information of the storage shelf 420C in the training data 150C) and the position where the item is stored in the storage facility (see the row and column of the storage shelf 420C in the training data 150C). The input information includes the position of the sorting shelf on which the sorted items for each order are placed (see shelf 1 and shelf 2 for the sorting number in the learning data 150C).

[0071] <<Variation: Product packaging>> In the embodiments described so far, when products are purchased, they are unpackaged and stored individually in storage boxes 425, 425B, and 425C. There are also logistics warehouses where products are placed on storage shelves 420 without being unpackaged. There is a large difference in the time it takes to remove products when the boxes containing the products have been opened and when they have not, which is thought to have a large impact on the work time required for sorting. In such cases, the opening status may be included in the learning data.

[0072] 18 is a data configuration diagram of learning data 150F according to a modified example of the first embodiment. Compared to learning data 150 (see FIG. 5), learning data 150F adds an attribute of "opened." "Opened" indicates the state of opening of the box in which the product was packed at the time of purchase (arrival). "Opened" indicates that the box has been opened, and "Not yet" indicates that the box has not been opened.

[0073] The explanatory variables of the prediction model 121 according to this modification are the product name, the position (face, row, column) of the storage box 425 on the storage shelf 420, the number of products to be removed from the storage box 425, the opening state of the box, and the number of products by sorting destination. The learning unit 112 trains and generates the prediction model 121 using the learning data 150F. Using such a prediction model 121 makes it possible to accurately predict the work time for sorting work, including cases where opening is required.

[0074] As described above, the input information (see the explanatory variables of the prediction model 121 and the explanatory variables of the training data 150F) includes the opening state of the box in which the item is packed (see the opening state of the training data 150F).

[0075] Other variations Although several embodiments of the present invention have been described above, these embodiments are merely examples and do not limit the technical scope of the present invention. For example, in the first embodiment, the work time for sorting work is measured using the work recording device 480. Alternatively, the work time may be measured by analyzing images taken by a camera that captures the sorting area 450. In the above-described embodiments, the sorting work is performed by a worker, but it may also be performed by a robot that performs the sorting work.

[0076] The present invention can take on various other embodiments, and various modifications such as omissions and substitutions can be made without departing from the spirit of the present invention. These embodiments and modifications are included in the scope and spirit of the invention described in this specification, etc., and are also included in the invention described in the claims and their equivalents.

[0077] <Hardware configuration> The prediction device 100 according to the embodiment described above is realized by a computer 900 having a configuration as shown in FIG. 19, for example. FIG. 19 is a hardware configuration diagram showing an example of the computer 900 that realizes the functions of the prediction device 100 according to the embodiment described above. The computer 900 includes a CPU 901, a ROM 902, a RAM 903, an SSD 904, an input / output interface 905 (referred to as an input / output I / F (Interface) in FIG. 19), a communication interface 906 (referred to as a communication I / F in FIG. 19), and a media interface 907 (referred to as a media I / F in FIG. 19). The computer 900 may include a hard disk drive (HDD) instead of the SSD 904, or may include a HDD in addition to the SSD 904.

[0078] The CPU 901 operates based on a program stored in the ROM 902 or the SSD 904, and performs control by the control unit 110 in Fig. 3. The ROM 902 stores a boot program executed by the CPU 901 when the computer 900 starts up, programs related to the hardware of the computer 900, and the like.

[0079] The CPU 901 controls an input device 910 such as a mouse or keyboard, and an output device 911 such as a display or printer, via an input / output interface 905. The CPU 901 acquires data from the input device 910 via the input / output interface 905, and outputs generated data to the output device 911.

[0080] The SSD 904 stores programs executed by the CPU 901 and data used by the programs. The communication interface 906 receives data from other devices (not shown) (such as the warehouse management system 210) via a communication network and outputs the data to the CPU 901, and also transmits data generated by the CPU 901 to other devices via the communication network.

[0081] The media interface 907 reads a program or data stored in the recording medium 912 and outputs it to the CPU 901 via the RAM 903. The CPU 901 loads the program from the recording medium 912 onto the RAM 903 via the media interface 907 and executes the loaded program. The recording medium 912 is an optical recording medium such as a DVD (Digital Versatile Disk), a magneto-optical recording medium such as an MO (Magneto Optical disk), a magnetic recording medium, a conductive memory tape medium, a semiconductor memory, or the like.

[0082] For example, when the computer 900 functions as the prediction device 100 according to the above-described embodiment, the CPU 901 of the computer 900 executes a program 128 (see FIG. 3) loaded onto the RAM 903, thereby realizing the functions of the prediction device 100. The CPU 901 reads the program from a recording medium 912 and executes it. Alternatively, the CPU 901 may read the program from another device via a communication network, or may install the program 128 from the recording medium 912 onto the SSD 904 and execute it. [Explanation of symbols]

[0083] 100 Prediction Device 111 Measurement Unit 112 Learning Department 113 Prediction Department 114 Batch Generation Unit 115 Batch generation method selection section 130, 130A, 130B, 130C, 130D Order delivery information database 140 Sorting Time Database 150, 150A, 150B, 150C, 150D, 150E, 150F Training data 121 Predictive Model 420,420C Storage Shelf (Storage) 425, 425B, 425C storage box 460,460C Sorting shelf 465 Sorting Box 480 Work Recording Device 480B,480C Display

Claims

1. a prediction unit that predicts a work time for sorting work, which is work of taking out the items from a storage warehouse that stores the items and sorting them based on a plurality of orders for the items, and calculates the predicted work time; The prediction unit The predicted work time is calculated based on input information including the type of item to be taken out from the storage warehouse, the number of items to be taken out from the storage warehouse, and the number of items included in each order. Prediction device.

2. The storage facility includes: The storage shelf is carried by the transport robot to the sorting work location and has multiple shelves in the vertical direction. The prediction device according to claim 1 .

3. The input information is The position where the item is stored on the storage shelf is included. The prediction device according to claim 2 .

4. The sorted articles are They are placed in sorting boxes on the sorting shelves, The input information is The position of the sorting box on the sorting shelf is included. The prediction device according to claim 1 .

5. The work time may be measured by a measuring unit. The prediction device according to claim 1 .

6. The prediction unit Prediction is performed using a machine learning model in which explanatory variables include the input information and objective variables include the task time. The prediction device according to claim 1 .

7. a batch generation unit that allocates the plurality of orders to a plurality of workers who will perform the sorting work; a batch generation method selection unit that determines, from among the plurality of batch generation units included in the prediction device, the batch generation unit with the shortest predicted total operation time, which is the sum of predicted operation times for sorting operations related to the plurality of orders. The prediction device according to claim 1 .

8. The article comprises: The items are transported along a predetermined route to the sorting work location. The prediction device according to claim 1 .

9. The worker who performs the sorting work is Move to the location where the storage warehouse is located and perform the sorting work, The input information is The storage facility and the location where the item is stored in the storage facility are included. The prediction device according to claim 1 .

10. The worker who performs the sorting work is Move to the location where the storage warehouse is located and perform the sorting work, The input information is The location of the sorting shelves on which the items sorted for each order are placed The prediction device according to claim 1 .

11. The input information is Including identification information of the worker who performs the sorting work The prediction device according to claim 1 .

12. The input information is Including the state of the box in which the item is packed The prediction device according to claim 1 .

13. A prediction device that predicts the work time for sorting work, Executing a step of predicting a work time for sorting work, which is work of taking out the items from a storage warehouse that stores the items and sorting them based on a plurality of orders for the items, and calculating the predicted work time; The predicted work time is The calculation is based on input information including the type of item to be removed from the storage, the number of items to be removed from the storage, and the number of items included in each order. Forecasting methods.

Images