Processing time estimation method, processing time estimation program, congestion information map generation method, and congestion information map generation program

Abstract

This processing time estimation method includes: allocating a plurality of tasks in an automated warehouse to each of a plurality of transport devices; generating a plurality of divided travel routes for each transport device by dividing, at a predetermined first time interval, an entire travel route during processing of the tasks for each transport device into N sections; and, for each of the N sections, executing input, to a trained model capable of outputting an estimated travel time for a divided travel route in a predetermined section for each transport device, of a divided travel route in an n-th section for each transport device, a map, and the first time interval.

Description

Processing Time Estimation Method, Processing Time Estimation Program, Traffic Jam Information Map Generation Method, and Traffic Jam Information Map Generation Program

[0006] ,

[0001] The present disclosure relates to a processing time estimation method, a processing time estimation program, a traffic jam information map generation method, and a traffic jam information map generation program.

[0002] Patent Document 1 discloses a warehouse management system provided with a host controller for managing the operation of a warehouse and a warehouse design device for designing the layout of the warehouse and the work flow of incoming and outgoing goods. The warehouse design device simulates work instructions or incoming and outgoing instructions based on the designed layout and work flow, and when a satisfactory result is obtained, it sends the designed layout and work flow to the host controller.

[0003] Japanese Patent Application Laid-Open No. 2008-94553

[0004] In recent years, there has been an increasing need to estimate the performance of an automated warehouse or formulate a work plan for the automated warehouse. The estimation of the performance of an automated warehouse or the formulation of a work plan may be performed by referring to the processing time of various operations in the automated warehouse. Therefore, in response to the increasing need, it is required to estimate the processing time of operations in the automated warehouse with higher accuracy. The processing time of an operation in an automated warehouse is, for example, estimated by simulating the operation. The warehouse design device of Patent Document 1 simulates operations such as incoming and outgoing or replenishment using the layout of the warehouse and the work flow of incoming and outgoing goods.

[0005] By the way, in an automated warehouse, a plurality of transport devices for transporting goods may transport the goods picked by an operator or move to a predetermined location based on an instruction from a user such as an operator. Therefore, traffic jams of transport devices may occur in an automated warehouse. However, Patent Document 1 does not consider the above possibility. In order to estimate the processing time of an operation in an automated warehouse with higher accuracy, and thus estimate the performance of the automated warehouse using the estimation result with higher accuracy, it is necessary to consider the traffic jams of transport devices in the automated warehouse.

[0006] This disclosure was devised in light of the conventional circumstances described above, and aims to provide a simple way to understand the impact of congestion on conveying equipment in automated warehouses.

[0007] This disclosure provides a processing time estimation method for an automated warehouse, which involves assigning a plurality of tasks constituting a work instruction to each of a plurality of transport devices, dividing the overall travel path when processing the tasks for each transport device into N (N: a natural number of 2 or more) sections at predetermined first time intervals to generate a plurality of divided travel paths for each transport device, and, in response to inputs of the divided travel paths in a predetermined section for each transport device, a map of the automated warehouse, and a predetermined time interval corresponding to the predetermined section, a trained model capable of outputting estimated travel times in the divided travel paths in the predetermined section for each transport device, is subjected to inputs of the divided travel paths in the nth (n: a natural number from 1 to N) section for each transport device, the map, and the first time interval for each of the N sections, and the processing time for the work instruction is estimated by summing the estimated travel times in the divided travel paths in each of the N sections for each transport device output from the trained model.

[0008] Furthermore, this disclosure provides a processing time estimation program that causes a computing device to assign a plurality of tasks constituting work instructions in an automated warehouse to each of a plurality of transport devices, divides the overall travel path when processing the tasks for each transport device into N (N: a natural number of 2 or more) sections at predetermined first time intervals, thereby generating a plurality of divided travel paths for each transport device, and, in response to inputs of the divided travel paths in a predetermined section for each transport device, a map of the automated warehouse, and a predetermined time interval corresponding to the predetermined section, causes a trained model capable of outputting estimated travel times in the divided travel paths in the predetermined section for each transport device to execute the inputs of the divided travel paths in the nth (n: a natural number from 1 to N) section for each transport device, the map, and the first time interval for each of the N sections, and estimates the processing time of the work instructions by summing the estimated travel times in the divided travel paths in each of the N sections for each transport device output from the trained model.

[0009] Furthermore, this disclosure provides a method for generating a traffic congestion map, which involves assigning a plurality of tasks constituting work instructions in an automated warehouse to each of a plurality of transport devices, dividing the overall travel route when processing the tasks for each transport device into N (N: a natural number of 2 or more) sections at predetermined first time intervals to generate a plurality of divided travel routes for each transport device, and, in response to inputs of the divided travel routes in a predetermined section for each transport device, a map of the automated warehouse, and a predetermined time interval corresponding to the predetermined section, a trained model capable of outputting a traffic congestion map of the automated warehouse at the predetermined time interval, and performing the input of the divided travel route in the nth (n: a natural number from 1 to N) section for each transport device, the map, and the first time interval for each of the N sections, and concatenating the N traffic congestion maps output from the trained model in a time series.

[0010] Furthermore, this disclosure provides a traffic congestion information map generation program that causes a computing device to assign a plurality of tasks constituting work instructions in an automated warehouse to each of a plurality of transport devices, divides the overall travel route when processing the tasks for each transport device into N (N: a natural number of 2 or more) sections at predetermined first time intervals, thereby generating a plurality of divided travel routes for each of the transport devices, and, in response to inputs of the divided travel routes in predetermined sections for each of the transport devices, a map of the automated warehouse, and predetermined time intervals corresponding to the predetermined sections, causes a trained model capable of outputting a traffic congestion information map of the automated warehouse at predetermined time intervals to execute the inputs of the divided travel routes in the nth (n: a natural number from 1 to N) section for each of the transport devices, the map, and the first time interval for each of the N sections, and concatenates the N traffic congestion information maps output from the trained model in a time series.

[0011] Furthermore, any combination of the above components, as well as any conversion of the expressions of this disclosure between methods, apparatus, systems, storage media, computer programs, etc., are also valid as aspects of this disclosure.

[0012] According to this disclosure, the impact of congestion on conveying equipment in an automated warehouse can be easily understood.

[0013] Block diagram showing an example of the hardware configuration of a computing device according to one embodiment of this disclosure. Schematic diagram showing an example of a map of an automated warehouse according to one embodiment of this disclosure. Schematic diagram for explaining tasks constituting an order according to one embodiment of this disclosure. Schematic diagram for explaining task assignment to transport equipment according to one embodiment of this disclosure. Schematic diagram for explaining divided travel routes for each transport equipment according to one embodiment of this disclosure. Schematic diagram for explaining divided travel routes for each transport equipment according to one embodiment of this disclosure. Schematic diagram for explaining travel routes as input data to a trained model according to one embodiment of this disclosure. Schematic diagram for explaining the processing time estimation flow according to Embodiment 1. Flowchart showing the processing time estimation process according to Embodiment 1. Schematic diagram for explaining the learning data according to Embodiment 1. Schematic diagram for explaining the flow of trained model generation according to Embodiment 1. Flowchart showing the trained model generation process according to Embodiment 1. Schematic diagram for explaining the traffic congestion information map generation flow according to Embodiment 2. Flowchart showing the traffic congestion information map generation process according to Embodiment 2. Schematic diagram for explaining the learning data according to Embodiment 2. Schematic diagram for explaining the trained model generation flow according to Embodiment 2. Flowchart showing the trained model generation process according to Embodiment 2.

[0014] The various embodiments of this disclosure will be described in detail below, with reference to the drawings as appropriate. However, unnecessary details may be omitted. For example, detailed explanations of already well-known matters or redundant explanations of substantially identical configurations may be omitted. This is to avoid the following explanation becoming unnecessarily verbose and to facilitate understanding by those skilled in the art. The accompanying drawings and the following explanation are provided to enable those skilled in the art to fully understand this disclosure and are not intended to limit the subject matter described in the claims. Furthermore, in this specification, terms such as "first" and "second" are used merely to distinguish components for the sake of explanation and are not intended to be interpreted as limiting to specific components. Therefore, these expressions should be understood to be appropriately reinterpreted depending on the configuration to which the invention of this disclosure applies.

[0015] (Embodiment 1) [Device Configuration] Figure 1 is a block diagram showing an example of the hardware configuration of a computing device 10 according to one embodiment of the present disclosure. Embodiment 1 describes an example in which the computing device 10 estimates the processing time of an order corresponding to a work instruction in an automated warehouse. An order may include, for example, operations such as the dispatch or inbound of goods in an automated warehouse. In this specification, "goods" includes merchandise and cargo, and these terms may be read interchangeably. In Embodiment 1, for the sake of explanation, an order will refer to a dispatch operation. Furthermore, an order may be composed of multiple tasks, as will be described later with reference to Figure 3, etc.

[0016] The computing device 10 may be a general-purpose computer device such as a Personal Computer (hereinafter referred to as "PC") or a server computer, or it may be a mobile terminal such as a tablet terminal or a smartphone. Note that the configuration shown in Figure 1 is just one example, and one part may be divided into multiple parts, or multiple parts may be combined into one.

[0017] The arithmetic unit 10 comprises a processor 11, memory 12, communication device 13, input device 14, external interface device 15, and display device 16. Each component is configured to communicate via an internal interface device 17.

[0018] The processor 11 may be configured using, for example, a Central Processing Unit (hereinafter referred to as "CPU"), a Graphical Processing Unit (hereinafter referred to as "GPU"), a Micro Processing Unit (hereinafter referred to as "MPU"), a Digital Signal Processor (hereinafter referred to as "DSP"), or a Field Programmable Gate Array (hereinafter referred to as "FPGA"). The processor 11 realizes various functions by reading and executing various data and programs held in the memory 12.

[0019] Memory 12 is a storage unit for storing various data and programs. Memory 12 may consist of volatile / non-volatile storage devices such as Random Access Memory (hereinafter referred to as "RAM"), Read Only Memory (hereinafter referred to as "ROM"), and Hard Disk Drive (hereinafter referred to as "HDD").

[0020] The communication device 13 is an interface for communicating with external devices via a network. The communication standards supported by the communication device 13 are not particularly limited and may support either wired or wireless communication standards. Furthermore, the communication device 13 may support multiple communication standards. Therefore, the network used by the communication device 13 may be composed of a combination of networks using multiple communication standards.

[0021] The input device 14 receives operations and instructions from a user, for example, who is selling goods in an automated warehouse. The input device 14 may consist of a mouse, keyboard, touch panel display, etc.

[0022] The external interface device 15 is an interface for sending and receiving data with an external device.

[0023] The display device 16 displays various user interfaces to the user. The display device 16 may consist of a liquid crystal display, a touch panel display, or the like.

[0024] As will be described in detail later, in Embodiment 1, the arithmetic unit 10 estimates the processing time of an order using a trained model 60 (see Figure 8). The trained model 60 may be stored and held in, for example, memory 12, or in an external device (not shown) accessible by the arithmetic unit 10.

[0025] Furthermore, the computing unit 10 may be connected to a management system that manages various data and information related to the automated warehouse in a communicative manner. The computing unit 10 may then acquire necessary data and information from the management system. For example, the computing unit 10 may acquire the automated warehouse map M1, described later, from the management system with reference to Figure 2. The management system may include, for example, a warehouse management system, sometimes referred to as a Warehouse Management System (hereinafter referred to as "WMS"), or a warehouse operation management system, sometimes referred to as a Warehouse Execution System (hereinafter referred to as "WES").

[0026] [Map of the automated warehouse] Next, with reference to Figure 2, the map M1 of the automated warehouse, which is used as input data to the trained model 60, will be described. Figure 2 is a schematic diagram showing an example of the map M1 of the automated warehouse according to one embodiment of the present disclosure. The arithmetic unit 10 may store and hold the map M1 as data in the memory 12. The arithmetic unit 10 may obtain the map M1 from a management system that manages various data related to the automated warehouse. Alternatively, the arithmetic unit 10 may obtain the map M1 from user input.

[0027] Map M1 shows various equipment and areas within the automated warehouse. In the example in Figure 2, Map M1 shows the locations of stations, garages, shelves, and shipping areas within the automated warehouse. Map M1 may also include X and Y coordinate information for any location within the automated warehouse. In Figure 2, the direction of the arrows on the X and Y axes is considered positive, and the direction opposite to the positive direction is considered negative.

[0028] In an automated warehouse, orders are assigned to stations. These orders are then processed at the station to which they were assigned. Further details will be provided later.

[0029] In an automated warehouse, multiple transport devices may transport items picked by workers or move to designated locations based on instructions or assigned tasks from users such as workers. Examples of tasks will be described later with reference to Figure 3. The transport devices transport bins in which items are stored. Examples of transport devices include Automatic Guided Vehicles (hereinafter referred to as "AGVs") and Autonomous Mobile Robots (hereinafter referred to as "AMRs"). In this specification, bins in which items are stored are distinguished into "inventory bins" for storing inventory, "empty bins" that do not store items, and "outbound bins" for storing items to be issued. When not specifically distinguished, they are simply referred to as "bins". At a station, for example, a worker or work robot may perform the task of moving items to be issued from inventory bins to outbound bins, in other words, the picking operation.

[0030] The automated warehouse incorporates multiple transport devices, which are stored in a garage and kept on standby. In Embodiment 1, each of the multiple transport devices is assigned to one of several stations. For example, if two transport devices are assigned to a station, these two devices operate to process the orders assigned to that station. More specifically, the multiple tasks included in the order are processed through the cooperation of the two transport devices and a worker or robot. This completes the processing of the order. At this time, the multiple tasks included in the order are linked to the station to which the order is assigned. This allows the worker or robot to perform picking and other tasks at the station to which the order is assigned. The two transport devices also transport bins to and from the station to which the order is assigned. In this way, an order is processed at the station to which it is assigned.

[0031] A brief example of the operation of conveying equipment in an automated warehouse is described below. It is assumed that the conveying equipment is assigned tasks corresponding to the operations described below. In an automated warehouse, for example, inventory bins stored on shelves are transported to a station by conveying equipment. Here, the inventory bins are assumed to contain items to be shipped. At the station, a worker or work robot transfers the items to be shipped from the inventory bins to the shipping bins. The conveying equipment transports the shipping bins containing the items to be shipped to the shipping area. The shipping bins transported to the shipping area may be shipped out of the automated warehouse in order.

[0032] Note that Map M1 shown in Figure 2 is a simplified map of an automated warehouse for illustrative purposes, and the layout of the automated warehouse is not limited to this. Therefore, although not shown in the figure, if the automated warehouse has multiple floors, Map M1 may further include information on elevators in the automated warehouse and information on each floor. Furthermore, Map M1 may also include information on locations where transport equipment can travel, in other words, information on pathways. The pathway information may further include information such as whether the pathway is one-way or not.

[0033] [Orders and Tasks] Next, orders and tasks in an automated warehouse will be described with reference to Figures 3 and 4. Figure 3 is a schematic diagram illustrating the tasks that constitute an order according to one embodiment of the present disclosure. Note that the multiple tasks included in the order shown in Figure 3 are examples, and the tasks are not limited to these.

[0034] The order shown in the example in Figure 3 consists of Task OS, Task BR1, Task PK1, Task RT1, Task BR2, Task PK2, Task RT2, Task BR3, Task PK3, Task RT3, and Task SP.

[0035] Task OS is a task that instructs the transport equipment to transport empty bottles, which have been designated as outbound bottles, to the station. Although not shown in map M1, the automated warehouse may also have an area where empty bottles are stored.

[0036] Task BR1 is the task of having a conveying machine transport inventory bins from the shelves to the station.

[0037] Task PK1 is a task in which a worker or work robot is instructed to transfer items to be issued from the inventory bins into the issue bins, in other words, to perform a picking operation.

[0038] Task RT1 is the task of having the conveying equipment return the inventory bins, for which the picking process has been completed, to the shelves.

[0039] Tasks BR1, PK1, and RT1 are tasks for a specific stock bin. Therefore, in task PK1, picking is performed using the stock bin that was transported to the station by the execution of task BR1. Then, in task RT1, the stock bin from which the items were removed in task PK1 is returned to the shelf.

[0040] Tasks BR2, PK2, and RT2 are also tasks for specific inventory bins, different from the specific inventory bins mentioned above. The same applies to tasks BR3, PK3, and RT3. Tasks BR2, PK2, RT2, BR3, PK3, and RT3 are described in the same way as tasks BR1, PK1, and RT1, only differing in the target inventory bins, so their descriptions are omitted.

[0041] Task SP is a task that causes the transport equipment to transport the bins containing the completed items for tasks PK1, PK2, and PK3 from the station to the retrieval area. As a result, the bins containing all the items to be retrieved as specified in the order shown in Figure 3 are retrieved from the automated warehouse.

[0042] Thus, each task included in the order is performed by transport equipment and a worker or a work robot.

[0043] Figure 4 is a schematic diagram illustrating the assignment of tasks to a transport device according to one embodiment of the present disclosure. In the example in Figure 4, it is assumed that the order shown in Figure 3 is processed collaboratively by a worker or work robot and transport devices a and b. Referring to Figure 4, an example of assigning tasks other than picking (i.e., task OS, task BR1, task RT1, task BR2, task RT2, task BR3, task RT3, task SP) to two transport devices will be explained.

[0044] The processor 11 of the computing device 10 may, for example, allocate each task included in the order to each transport device so that the processing time of the order is minimized. To allocate each task to each transport device, the computing device 10 may, for example, acquire necessary data and information, etc. from a management system that manages various data and information related to an automated warehouse. The necessary data and information may be, for example, the performance of the transport devices introduced in the automated equipment or the map M1 of the automated warehouse described with reference to FIG. 2. Alternatively, the computing device 10 may allocate each task to each transport device based on an input from the user. Here, the input from the user may be, for example, a condition setting such as "shorten the processing time of the order" or "shorten the travel distance".

[0045] In the example of FIG. 4, task OS, task BR3, task RT1, task RT2, and task SP are allocated to transport device a among the two transport devices, and task BR1, task BR2, and task RT3 are allocated to transport device b. Task PK1 and task PK2 are performed on the inventory bin transported by transport device b to the station, and task PK3 is performed on the inventory bin transported by transport device a to the station. The movements of transport device a and transport device b when each task is allocated in this way will be described in chronological order. In the description with reference to FIG. 4, the station refers to one of the plurality of stations (that is, the station to which the order shown in FIG. 3 is allocated).

[0046] First, transport device a starts task BR3, and transport device b starts task BR1. As a result, each of the two transport devices transports the inventory bin to the station. Transport device a completes task BR3 earlier than transport device b completes task BR1 and starts task OS. As a result, an empty bin is transported to the station as an outbound bin.

[0047] At the station, task PK1 and task PK3 are started. As a result, the articles to be shipped, which are stored in each of the two inventory bins transported to the station by each of the two transport devices, are transferred to the shipping bin. Task PK1 completes earlier than task PK3. While task PK3 is being executed, transport device b starts task BR2. As a result, an inventory bin is transported to the station by transport device b. When task PK3 is completed, transport device a starts task RT1. As a result, the inventory bin transported to the station by the execution of task BR1 is returned to the shelf. While task RT1 is being executed, transport device b completes task BR2. Then, at the station, task PK2 is started.

[0048] When task PK2 is completed, that is, when all the articles to be shipped have been transferred from the inventory bin to the shipping bin, transport device a starts task SP and transport device b starts task RT3. As a result, the shipping bin is transported from the station to the shipping area, and the shipping bin is shipped out of the automated warehouse. Also, the inventory bin transported to the station by the execution of task BR3 is returned to the shelf. After transport device a completes task SP, it starts task RT2. As a result, the inventory bin transported to the station by the execution of task BR2 is returned to the shelf.

[0049] So that each task shown in FIG. 3 is processed in this way, the processor 11 of the arithmetic unit 10 may allocate each task to the two transport devices.

[0050] [Divided Travel Route] Next, referring to FIGS. 5, 6, and 7, the flow in which the processor 11 of the arithmetic unit 10 generates a divided travel route and a time interval as input data to the learned model 60 to be described later will be described. FIG. 5 is a schematic diagram for explaining the divided travel route for each transport device according to an embodiment of the present disclosure.

[0051] As described above, in an automated warehouse, orders are assigned to stations. In the example in Figure 5, the processor 11 assigns all orders to stations S1, S2, and S3, respectively. The method of assigning orders to stations is not particularly limited.

[0052] In the example shown in Figure 5, the processor 11 assigns transport devices V1 and V2 to station S1, transport devices V3 and V4 to station S2, and transport devices V5 and V6 to station S3. As a result, for example, an order assigned to station S1 is processed by transport devices V1 and V2. For the sake of simplicity, the involvement of human operators or robotic operators is omitted.

[0053] As explained with reference to Figures 3 and 4, the multiple tasks included in an order assigned to a station are assigned to the transport equipment assigned to that station. For example, an order assigned to station S1 is processed by transport equipment V1 and transport equipment V2.

[0054] For example, the processor 11 assigns multiple tasks, such as task TSK1 and task TSK2, included in order OD1 assigned to station S1, to either transport device V1 or transport device V2.

[0055] In the example shown in Figure 5, station S1 is assigned orders OD1, OD2, OD3, and OD4. Each task included in each order is also assigned to transport equipment V1 and transport equipment V2, respectively. Similarly, stations S2 and S3 are assigned multiple orders, and tasks are further assigned to each of their transport equipment.

[0056] The processor 11 can generate travel paths for each order and each transport device by assigning each task to each transport device. For example, the processor 11 can generate the travel paths for transport device V1 and transport device V2 when processing each task of order OD1.

[0057] Furthermore, the processor 11 can generate an overall travel path for each transport device by linking the generated travel paths of each transport device in the order of order processing. For example, the processor 11 can generate the overall travel path VR1 of transport device V1 by linking the travel paths of transport device V1 as it processes orders OD1, OD2, OD3, and OD4 in the order of order processing. In this case, the order of order processing is OD1, OD2, OD3, and OD4. The processor 11 can generate the overall travel path VR2 of transport device V2, the overall travel path VR3 of transport device V3, the overall travel path VR4 of transport device V4, the overall travel path VR5 of transport device V5, and the overall travel path VR6 of transport device V6, similar to the overall travel path VR1 of transport device V1.

[0058] Figure 6 shows the overall travel path VR5 of the transport equipment V5, and the explanation will continue with reference to this figure. Figure 6 is a schematic diagram for explaining the divided travel paths for each transport equipment according to one embodiment of the present disclosure.

[0059] When the processor 11 generates a travel path for the conveying equipment V5 to process orders, it can calculate the displacement of the conveying equipment V5 in the X-axis direction over time. The processor 11 can also calculate the displacement of the conveying equipment V5 in the Y-axis direction over time. Furthermore, based on the displacement of the conveying equipment V5 in the X-axis direction and the displacement of the conveying equipment V5 in the Y-axis direction over time, the processor 11 can generate the travel path of the conveying equipment V5 in the XY plane, i.e., in the automated warehouse. The processor 11 can connect the generated travel paths in the order of order processing to generate the overall travel path VR5 of the conveying equipment V5.

[0060] Alternatively, the processor 11 may calculate the displacement in the X-axis direction and the displacement in the Y-axis direction of the transport device V5 as time progresses while all orders are being processed, and generate the overall travel path VR5 of the transport device V5 from these. As shown in Figure 6, the processor 11 may generate the overall travel path VR5 based on a displacement P1 representing the displacement in the X-axis direction of the transport device V5 as time progresses, and a displacement P2 representing the displacement in the Y-axis direction of the transport device V5 as time progresses.

[0061] The processor 11 can generate divided travel routes by dividing the overall travel route into arbitrary time intervals. In the example shown in Figure 6, the processor 11 divides the overall travel route VR5 into time intervals TS1 to generate multiple divided travel routes, including divided travel route DR50, divided travel route DR51, and divided travel route DR52.

[0062] To divide the overall travel route into arbitrary time intervals and generate multiple divided travel routes means, more precisely, dividing the total travel time corresponding to the overall travel route into arbitrary time intervals and generating multiple travel routes, i.e., divided travel routes, for each divided time interval. The time intervals for dividing the overall travel route may be set in advance by the user, or they may be set by known technologies such as Artificial Intelligence (hereinafter referred to as "AI") technology. For example, the time intervals for dividing the overall travel route may be the minimum processing time of an order of magnitude or the average processing time estimated by AI technology.

[0063] The generated segmented travel paths and time intervals (i.e., seconds, minutes, or hours) are used as input data to the trained model 60, which will be described later. The conversion process for using the segmented travel paths as input data to the trained model 60 will be explained with reference to Figure 7. Figure 7 is a schematic diagram illustrating the travel paths used as input data to the trained model 60 according to one embodiment of this disclosure.

[0064] Figure 7 shows the segmented travel path DR generated by the processor 11. The processor 11 converts the segmented travel path DR into segmented path data DRD. The segmented path data DRD is the segmented travel path DR converted into a format that can be input into the trained model 60. The segmented path data DRD includes data on the starting position, arrival position, and intermediate path between the starting position and arrival position of the transport equipment in the segmented travel path DR. In the example in Figure 7, the segmented path data DRD shows "1" for the cell corresponding to the starting position of the transport equipment among the cells included in the grid corresponding to the map M1 of the automated warehouse. Also, cells corresponding to the intermediate path between the starting position and arrival position of the transport equipment are shown as "2", and cells corresponding to the arrival position are shown as "3".

[0065] Note that the segmented route data DRD shown in Figure 7 is just one example of a segmented travel route as input data, and the format of the segmented travel route that can be input to the trained model 60 is not limited to this.

[0066] [Processing Time Estimation Process] Next, a method for estimating the processing time of an order of magnitude using the trained model 60 will be described with reference to Figures 8 and 9. Figure 8 is a schematic diagram illustrating the flow of processing time estimation according to Embodiment 1.

[0067] In the example shown in Figure 8, it is assumed that the overall travel path of each of the six transport devices shown in Figure 5 is divided at time intervals TS1, generating multiple divided travel paths. Here, as an example for explanation, let's assume that the overall travel path of each transport device is divided N times, generating N divided travel paths. Note that N is a natural number greater than or equal to 2.

[0068] We focus on the nth segment of the N segmented travel paths of each transport device: segmented travel path DR1, segmented travel path DR2, segmented travel path DR3, segmented travel path DR4, segmented travel path DR5, and segmented travel path DR6. Here, n is a natural number from 1 to N.

[0069] The divided travel paths DR1, DR2, DR3, DR4, DR5, and DR6 are the respective travel paths of the transport equipment V1, V2, V3, V4, V5, and V6. The processor 11 of the arithmetic unit 10 converts each divided travel path into divided path data DRD1, DRD2, DRD3, DRD4, DRD5, and DRD6, respectively, in order to use them as input data for the trained model 60.

[0070] Furthermore, the processor 11 uses the time interval TS1 and the automated warehouse map M1 as input data to the trained model 60. In other words, the processor 11 inputs the data of the divided travel path in the nth section for each transport device, the map M1, and the time interval TS1 to the trained model 60.

[0071] The trained model 60 is generated in a way that allows it to output the estimated travel time for each transport device in a predetermined section of its divided travel path, based on inputs of data for divided travel paths in predetermined sections for each transport device, the automated warehouse map M1, and predetermined time intervals corresponding to those predetermined sections. The predetermined time intervals corresponding to the predetermined sections are the time intervals used when dividing the overall travel path in order to generate the divided travel paths. The method for generating the trained model 60 will be described later with reference to Figures 10, 11, and 12.

[0072] In the example shown in Figure 8, as described above, the processor 11 inputs the data of the divided travel path in the nth section for each transport device, the map M1, and the time interval TS1 into the trained model 60. The processor 11 then obtains the estimated travel time in the divided travel path for the nth section for each transport device, which is output from the trained model 60. In other words, the processor 11 obtains the estimated travel time in the divided travel path DR1 for transport device V1, the estimated travel time in the divided travel path DR2 for transport device V2, the estimated travel time in the divided travel path DR3 for transport device V3, the estimated travel time in the divided travel path DR4 for transport device V4, the estimated travel time in the divided travel path DR5 for transport device V5, and the estimated travel time in the divided travel path DR6 for transport device V6.

[0073] The processor 11 can obtain estimated travel times for each of the N divided travel paths in each of the N sections from the first section to the Nth section for each transport device by repeatedly inputting the data into the trained model 60 in this manner. The processor 11 can then estimate the processing time for an order by summing the estimated travel times for each of the N divided travel paths for each transport device. More specifically, the processor 11 sums the estimated travel times for each of the N divided travel paths for transport device V1 and transport device V2 to calculate the estimated travel time for the entire travel path of transport device V1 and the estimated travel time for the entire travel path of transport device V2. This allows the processor 11 to calculate the processing time for an order assigned to station S1. This is because transport devices V1 and V2 are assigned to station S1. In this embodiment, it is assumed that the transport devices operate (travel) to process the assigned task, and the travel time of the transport devices corresponds to the time required to process the order (task). Here, the travel time of the transport equipment includes time when it is temporarily stopped, for example, to process a task, and is not actually traveling.

[0074] As described above, the processor 11 can calculate the processing time for orders assigned to station S1, and similarly for orders assigned to stations S2 and S3, respectively. Therefore, the processor 11 can estimate the total processing time for all orders by summing the processing times for orders assigned to each of the multiple stations.

[0075] The multiple segmented travel paths (more precisely, segmented path data) input to the trained model 60 are segmented travel paths for the same section (for example, between 1:00 and 2:00) from among multiple segmented travel paths divided at predetermined time intervals for each transport device. This allows the trained model 60 to estimate the travel time for each transport device as a travel time that takes into account congestion among the transport devices. In other words, in Embodiment 1, the order processing time takes into account congestion among the transport devices in the automated warehouse.

[0076] Next, with reference to Figure 9, the processing time estimation process will be explained along with the flowchart. Figure 9 is a flowchart of the processing time estimation process according to Embodiment 1. Note that each process in the flowchart shown in Figure 9 is performed by the processor 11 of the arithmetic unit 10. Furthermore, it is assumed that there are multiple orders in the automated warehouse, multiple stations are provided in the automated warehouse, and multiple transport devices are installed. In addition, it is assumed that the processor 11 assigns multiple transport devices to each of the multiple stations.

[0077] The processor 11 assigns all orders to each of the multiple stations located in the automated warehouse (step St 100).

[0078] In step St100, the processor 11 assigns the multiple tasks included in the multiple orders assigned to each station to each of the multiple transport devices, and generates a travel route for each order and each transport device (step St101). At this time, the tasks included in the orders assigned to a station are assigned to the transport device assigned to that station.

[0079] The processor 11 generates the overall travel path for each transport device by concatenating the travel paths generated in step St 101 in the order of processing (step St 102).

[0080] The processor 11 divides the overall travel path into N segments at arbitrary time intervals for each transport device and generates N segmented travel paths (step St103).

[0081] The processor 11 sets the value i, which is used to determine the end of the estimation process that starts from step St105, to 1 (step St104).

[0082] The processor 11 starts the estimation process and executes the processes from step St106 to step St108 while i is less than or equal to N (step St105). At this time, the processor 11 focuses on the i-th section of the divided travel paths for each transport device, which are generated in step St103 and consist of sections from the first section to the N-th section.

[0083] The processor 11 converts each of the i-th section's divided travel paths for each transporter, generated in step St 103, into input data for the trained model 60 (step St 106). This generates divided path data. Since the value of i is set to 1 in step St 104, the divided travel path of the first section out of the N sections of divided travel paths for each transporter is converted into divided path data.

[0084] The processor 11 inputs the divided travel paths for each transport device, which were converted into divided route data in step St 106, the automated warehouse map M1, and the time interval used when dividing the overall travel path in step St 103, into the trained model 60 to estimate the travel time for each divided travel path for each transport device (step St 107). Since the value of i is set to 1 in step St 104, the processor 11 can obtain the estimated travel time for each divided travel path in the first section for each transport device.

[0085] The processor 11 adds 1 to the value of i (step St108). For example, if the value of i is set to 1, step St108 makes the value of i 2.

[0086] The processor 11 repeatedly executes the processes from step St106 to step St108 as long as i is less than or equal to N. This allows the processor 11 to have the trained model 60 output the estimated travel time for each of the N sections of the divided travel path for each transport device.

[0087] The processor 11 calculates the processing time for each order by summing the estimated travel times for each of the N divided travel paths for each transport device (step St 109). Then, the processor 11 terminates this processing flow.

[0088] As described above, in Embodiment 1, the calculated order processing time takes into account congestion of the conveying equipment in the automated warehouse. Therefore, if a user is considering multiple orders in the automated warehouse, they can have the calculation unit 10 estimate the total processing time for all orders, taking into account congestion of the conveying equipment. In this way, the user can easily grasp the impact of congestion of the conveying equipment in the automated warehouse.

[0089] [Trained Model] Next, a method for generating a trained model 60 will be described with reference to Figures 10, 11, and 12. Figure 10 is a schematic diagram illustrating the training data according to Embodiment 1.

[0090] The processor 11 generates a trained model 60 by training the model using multiple pairs of input data and ground truth data as training data. The input data and ground truth data will be described below. The training algorithm for generating the trained model 60 is not particularly limited, but machine learning including neural network technologies such as Multi-Layer Perceptron (hereinafter referred to as "MLP") and Convolutional Neural Network (hereinafter referred to as "CNN") and deep learning technologies may be used.

[0091] The processor 11 uses, as input data, a time interval for dividing the overall travel route, the divided travel routes obtained by dividing the overall travel route at the said time interval, and the automated warehouse map M1. Here, the divided travel routes used are the divided travel routes of each of the multiple transport devices (see Figure 11). This is to train the trained model 60 so that it can output an estimated travel time that takes into account congestion of the transport devices. More precisely, the divided travel routes used are those that have been converted into divided route data.

[0092] The processor 11 may divide the overall travel route at various time intervals. For example, Figure 6 shows an example where the overall travel route VR5 is divided at time interval TS1. However, as shown in Figure 10, the processor 11 may use divided travel routes DR53, generated by dividing the overall travel route VR5 at time interval TS2, for example, as input data.

[0093] The processor 11 uses the estimated travel time calculated by simulation as the ground truth data. Here, the estimated travel time used is the estimated travel time for each of the multiple transport devices (see Figure 11).

[0094] In the example shown in Figure 10, as described above, the processor 11 divides the overall travel path VR5 into time intervals TS2 to generate divided travel paths DR53. To generate training data (ground truth data), the processor 11 calculates the estimated travel time of the transport equipment V5 on the divided travel paths DR53 by performing a simulation of operations in an automated warehouse. The processor 11 may convert the divided travel paths into divided path data for use in the simulation. In this case, the processor 11 may use, for example, discrete simulation techniques. For example, the processor 11 may perform discrete simulation based on an event-driven calculation method. Under an event-driven calculation method, system changes are captured event by event, and the system state is updated only when an event occurs. As a result, the behavior of the entire system is not constantly tracked, but rather calculations are performed only when important events occur, enabling efficient simulation. For example, the processor 11 may apply discrete simulation techniques based on an event-driven calculation method when simulating a case where multiple transport equipment handle tasks related to the transport of goods in an automated warehouse. This allows the processor 11 to individually track the movement path (travel path) of the transport equipment and the time it takes to reach specific locations such as stations, and to update the simulation each time an event occurs in which a transport piece of equipment reaches a specific location.

[0095] In this way, the processor 11 can track events that occur discretely through discrete simulation, and can therefore simulate in detail the actual operation of the automated warehouse, taking into account congestion or competition at intersections between multiple transport devices. As a result, the processor 11 can calculate the estimated travel time of the transport device V5 on the divided travel path DR53. At this time, the processor 11 may also calculate the estimated travel time of each transport device other than the transport device V5 on the divided travel path generated by dividing it at the time interval TS2.

[0096] The estimated travel time of the transport equipment on the divided travel path calculated by the simulation is used as ground truth data. Next, the flow of generating the trained model 60 will be explained with reference to Figure 11. Figure 11 is a schematic diagram illustrating the flow of generating the trained model according to Embodiment 1.

[0097] In the example shown in Figure 11, it is assumed that the overall travel path of each of the six transport devices shown in Figure 5 is divided at time intervals TS2, generating multiple divided travel paths. Here, as an example for explanation, let's assume that the overall travel path of each transport device is divided M times, generating M divided travel paths. Here, M is a natural number greater than or equal to 2.

[0098] We focus on the m-th section of the M divided travel paths of each transport device: divided travel paths DR7, DR8, DR9, DR10, DR11, and DR12. Here, m is a natural number from 1 to M.

[0099] The divided travel paths DR7, DR8, DR9, DR10, DR11, and DR12 are the travel paths of transport equipment V1, V2, V3, V4, V5, and V6, respectively. The processor 11 of the arithmetic unit 10 converts each divided travel path into divided path data DRD7, DRD8, DRD9, DRD10, DRD11, and DRD12, respectively, in order to use them as input data for model 60a. Model 60a is the model in the training phase of the trained model 60. In other words, once the training of model 60a is complete, the trained model 60 is generated.

[0100] Furthermore, the processor 11 uses the time interval TS2 and the automated warehouse map M1 as input data to the model 60a. In other words, the processor 11 inputs the data of the divided travel path in the mth section for each transport device, the map M1, and the time interval TS2 to the model 60a.

[0101] Model 60a outputs an estimated travel time for each conveying device on its divided travel path in a predetermined section, based on inputs of data on the divided travel path in a predetermined section for each conveying device, the automated warehouse map M1, and a predetermined time interval corresponding to the predetermined section. Therefore, Model 60a outputs an estimated travel time for each conveying device on its divided travel path in the m-th section, based on inputs of data on the divided travel path in the m-th section for each conveying device, the map M1, and the time interval TS2.

[0102] The processor 11 calculates the estimated travel time for each m-th section of the divided travel path for each transport device through simulation. The processor 11 uses the estimated travel time calculated by simulation as ground truth data.

[0103] The processor 11 adjusts the parameters of model 60a so that the difference between the estimated travel time in the m-th section of the divided travel path for each transport device, output from model 60a, and the estimated travel time in the m-th section of the divided travel path for each transport device, calculated by simulation, becomes smaller. The processor 11 generates a trained model 60 by repeatedly adjusting the parameters of model 60a in this way.

[0104] Next, the trained model generation process will be explained following the flowchart with reference to Figure 12. Figure 12 is a flowchart of the trained model generation process according to Embodiment 1. Note that each process in the flowchart shown in Figure 12 is performed by the processor 11 of the arithmetic unit 10. Furthermore, it is assumed that there are multiple orders in the automated warehouse, multiple stations are provided in the automated warehouse, and multiple transport devices are installed. In addition, it is assumed that the processor 11 assigns multiple transport devices to each of the multiple stations.

[0105] The processor 11 assigns all orders to each of the multiple stations located in the automated warehouse (step St200).

[0106] In step St200, the processor 11 assigns the multiple tasks included in the multiple orders assigned to each station to each of the multiple transport devices, and generates a travel route for each order and each transport device (step St201). At this time, the tasks included in the orders assigned to a station are assigned to the transport device assigned to that station.

[0107] The processor 11 generates the overall travel path for each transport device by concatenating the travel paths generated in step St201 in the order of processing (step St202).

[0108] The processor 11 divides the overall travel path into M sections at arbitrary time intervals for each transport device and generates M divided travel paths (step St203).

[0109] The processor 11 sets the value i, which is used to determine the end of the simulation process that started in step St205, to 1 (step St204).

[0110] The processor 11 starts the simulation process and executes the processes from step St206 to step St209 while i is less than or equal to M (step St205). At this time, the processor 11 focuses on the i-th section of the divided travel paths for each transport device, which are generated in step St203 and consist of sections from the first section to the M-th section.

[0111] The processor 11 converts each of the i-th section division travel paths generated in step St203 for each transport device into input data to be used for simulation, i.e., division path data (step St206). Since the value of i is set to 1 in step St204, the division travel path of the first section out of the M sections of division travel paths for each transport device is converted into division path data.

[0112] The processor 11 calculates the estimated travel time for each transport device in each of the i-th section of the i-th section of the i-th section of the i-th section of the transport device's divided travel path, generated in step St203, by simulating using the divided travel path for each transport device, which was converted into divided travel path data in step St206 (step St207).

[0113] The processor 11 records learning data (step St208) in which the divided travel routes for each transport device converted into divided route data in step St206, the automated warehouse map M1, and the time intervals used when dividing the overall travel route in step St203 are input data, and the estimated travel time calculated by simulation in step St207 is used as the ground truth data. The learning data may be stored in memory 12, for example. Since the value of i is set to 1 in step St204, the processor 11 can record learning data including the estimated travel time in the divided travel route for the first section for each transport device.

[0114] The processor 11 adds 1 to the value of i (step St209). For example, if the value of i is set to 1, step St209 makes the value of i 2.

[0115] The processor 11 repeatedly executes the processes from step St206 to step St209 as long as i is less than or equal to M. This allows the processor 11 to generate multiple training data, including the estimated travel time in each of the M sections of the divided travel path for each transport device.

[0116] The processor 11 generates a trained model 60 based on multiple training data sets generated by repeating the processes from step St206 to step St209 (step St210). Then, the processor 11 terminates this processing flow.

[0117] In this way, the processor 11 can generate a trained model 60 that can estimate the travel time of transport equipment, taking into account congestion of the transport equipment.

[0118] (Summary of Embodiment 1) The following technology is disclosed based on the description of Embodiment 1 above. The components etc. in the above Embodiment 1 are examples, but are not limited to these.

[0119] (Technology 1) The processing time estimation method involves assigning multiple tasks (e.g., task TSK1, task TSK2) that constitute a work instruction (e.g., order OD1) in an automated warehouse to each of multiple transport devices (e.g., transport device V1, transport device V2), and dividing the overall travel path (e.g., overall travel path VR1) when processing the tasks for each transport device into N (N: a natural number of 2 or more) sections at a predetermined first time interval (e.g., time interval TS1), thereby generating multiple divided travel paths (e.g., divided travel paths DR1) for each transport device, and the divided travel paths in a predetermined section for each transport device, In response to inputs of an automated warehouse map (e.g., map M1) and predetermined time intervals corresponding to predetermined sections, a trained model (e.g., trained model 60) capable of outputting estimated travel times on segmented travel paths in predetermined sections for each transport device is input for each of the N sections, along with the segmented travel paths, map, and first time interval for each of the N sections. The estimated travel times on segmented travel paths in each of the N sections output by the trained model for each transport device are then summed up for each transport device to estimate the processing time of work instructions.

[0120] This allows the processing time estimation method to estimate the travel time for task processing included in the work instructions for each conveying device, taking into account congestion of conveying devices in the automated warehouse. As a result, the processing time estimation method can estimate the processing time of work instructions in the automated warehouse with greater accuracy compared to cases where congestion of conveying devices is not taken into account.

[0121] (Technology 2) The processing time estimation method described in Technology 1 assigns all work instructions in the automated warehouse to each of the multiple stations (e.g., station S1) provided in the automated warehouse for processing work instructions, and each of the multiple transport equipment is assigned to one of the multiple stations. For each station, the estimated travel time in the divided travel path in each of the N sections for each transport equipment assigned to the station is summed up to estimate the processing time for each of the multiple stations assigned to the work instructions, and the total processing time for all work instructions is estimated by summing the processing times for each of the multiple stations assigned to the work instructions.

[0122] This allows the processing time estimation method to estimate the processing time for work instructions assigned to each station in the automated warehouse, and by summing the processing times for work instructions assigned to each station, the total processing time for all work instructions can be estimated.

[0123] (Technology 3) The processing time estimation method described in Technology 1 or 2 divides the overall travel path when processing tasks for each transport device into M (M: a natural number of 2 or more) segments at predetermined second time intervals (e.g., time interval TS2), thereby generating multiple divided travel paths for each transport device. Based on the generated multiple divided travel paths for each transport device, the estimated travel time for each divided travel path in each of the M segments is calculated by simulating operations in an automated warehouse. A trained model is generated based on training data that uses multiple pairs of divided travel paths, maps, and second time intervals in the m (m: a natural number from 1 to M) segment for each transport device, and the estimated travel time for the divided travel path in the m segment calculated by simulation.

[0124] As a result, the processing time estimation method can generate a trained model that outputs the estimated travel time for each of the multiple transport devices' divided travel paths, taking into account congestion of the multiple transport devices, based on inputs such as time intervals, each divided travel path of multiple transport devices, and a map of the automated warehouse. Furthermore, the processing time estimation method can generate a large amount of training data at high speed and generate a highly accurate trained model by, for example, using discrete simulation technology for the simulation.

[0125] (Technology 4) The processing time estimation program causes the computing unit to assign multiple tasks constituting work instructions in the automated warehouse to each of the multiple transport devices, divides the overall travel path when processing tasks for each transport device into N (N: a natural number of 2 or more) sections at predetermined first time intervals, and generates multiple divided travel paths for each transport device, and in response to inputs of the divided travel paths in predetermined sections for each transport device, the automated warehouse map, and predetermined time intervals corresponding to the predetermined sections, the program causes a trained model capable of outputting estimated travel times in the divided travel paths in predetermined sections for each transport device to execute the inputs of the divided travel paths, the map, and the first time interval for the nth (n: a natural number from 1 to N) section for each transport device for each of the N sections, and estimates the processing time of work instructions by summing the estimated travel times in the divided travel paths in each of the N sections for each transport device output by the trained model.

[0126] As a result, the processing time estimation program can achieve the same effect as in Technique 1.

[0127] (Embodiment 2) Embodiment 2 describes an example in which the computing device 10 uses a trained model 61 (see Figure 13) to generate a time-series congestion information map within an automated warehouse when multiple conveying devices process orders in the automated warehouse. The congestion information map is a map showing the congestion status of conveying devices within the automated warehouse. In the description of Embodiment 2, explanations that are the same as those in Embodiment 1 will be simplified or omitted, and the explanation will focus on the differences.

[0128] [Traffic Congestion Information Map Generation Process] First, a method for generating a traffic congestion information map using the trained model 61 will be described with reference to Figures 13 and 14. Figure 13 is a schematic diagram illustrating the flow of traffic congestion information map generation according to Embodiment 2.

[0129] In the example shown in Figure 13, it is assumed that the overall travel path of each of the six transport devices shown in Figure 5 is divided at time intervals TS1, generating multiple divided travel paths. Here, as an example for explanation, it is assumed that the overall travel path of each transport device is divided into N sections, generating N divided travel paths.

[0130] Of the N divided travel paths of each transport device, we focus on the divided travel paths of the nth section: divided travel path DR1, divided travel path DR2, divided travel path DR3, divided travel path DR4, divided travel path DR5, and divided travel path DR6. The processor 11 of the arithmetic unit 10 converts each divided travel path into divided travel path data DRD1, divided travel path data DRD2, divided travel path DR3, divided travel path DR4, divided travel path DR5, and divided travel path DR6, respectively, in order to use them as input data for the trained model 61.

[0131] Furthermore, the processor 11 uses the time interval TS1 and the automated warehouse map M1 as input data to the trained model 60. In other words, the processor 11 inputs the data of the divided travel path in the nth section for each transport device, the map M1, and the time interval TS1 to the trained model 60.

[0132] The trained model 61 is generated in such a way that it can output a congestion information map of the automated warehouse for a predetermined time interval, in response to inputs of a divided travel route in a predetermined section for each transport device, a map M1 of the automated warehouse, and a predetermined time interval corresponding to the predetermined section. As explained in Embodiment 1, the predetermined time interval corresponding to the predetermined section is the time interval used when dividing the overall travel route in order to generate the divided travel routes, and in this case, it is the time interval TS1. For example, suppose the time interval TS1 is 1 hour. In this case, the overall travel route of the transport device is divided into, for example, a travel route between 1:00 and 2:00, a travel route between 2:00 and 3:00, and a travel route between 3:00 and 4:00. For example, if the travel route for each transport device between 1:00 and 2:00 is input to the trained model 61, the congestion information map of the automated warehouse for that predetermined time interval means the congestion information map between 1:00 and 2:00. In other words, the congestion information map output from the trained model 61 shows the congestion status of the transport devices in the automated warehouse as time changes.

[0133] The trained model 61 may also be capable of generating a traffic congestion information map that shows the congestion situation at a specific point in time. A specific point in time could be, for example, 2:00, 3:00, or 4:00 in an example where the time interval TS1 is 1 hour, and the overall travel route of the transport equipment is divided into a travel route between 1:00 and 2:00, a travel route between 2:00 and 3:00, and a travel route between 3:00 and 4:00. In other words, a specific point in time could be the end of the time interval, or an intermediate point in the time interval (in the above example, 1:30, 2:30, 3:30), etc.

[0134] The method for generating the trained model 61 will be described later with reference to Figures 15, 16, and 17.

[0135] In the example shown in Figure 13, as described above, the processor 11 inputs the data of the divided travel route in the nth section for each transport device, the map M1, and the time interval TS1 into the trained model 61. The processor 11 then obtains the congestion information map JM1 for the time interval TS1 corresponding to the nth section for each transport device, which is output from the trained model 61.

[0136] By repeatedly inputting data into the trained model 61 in this manner, the processor 11 can obtain congestion information maps for each of the N sections, from the first section to the Nth section, corresponding to the time interval TS1 for each transport device. In other words, the processor 11 can obtain N congestion information maps. Then, by concatenating the N congestion information maps in a time series, the processor 11 can obtain a congestion information map for the entire time. More specifically, the entire time refers to the total time during which orders are processed in the automated warehouse.

[0137] Next, with reference to Figure 14, the traffic congestion information map generation process will be explained according to the flowchart. Figure 14 is a flowchart of the traffic congestion information map generation process according to Embodiment 2. Note that each process in the flowchart shown in Figure 14 is performed by the processor 11 of the arithmetic unit 10. Furthermore, it is assumed that there are multiple orders in the automated warehouse, multiple stations are provided in the automated warehouse, and multiple transport devices are installed. In addition, it is assumed that the processor 11 assigns multiple transport devices to each of the multiple stations.

[0138] The processor 11 assigns all orders to each of the multiple stations located in the automated warehouse (step St300).

[0139] In step St300, the processor 11 assigns the multiple tasks included in the multiple orders assigned to each station to each of the multiple transport devices, and generates a travel route for each order and each transport device (step St301). At this time, the tasks included in the orders assigned to a station are assigned to the transport device assigned to that station.

[0140] The processor 11 generates the overall travel path for each transport device by concatenating the travel paths generated in step St301 in the order of processing (step St302).

[0141] The processor 11 divides the overall travel path into N segments at arbitrary time intervals for each transport device and generates N segmented travel paths (step St303).

[0142] The processor 11 sets the value i, which is used to determine the end of the generation process that started in step St305, to 1 (step S304).

[0143] The processor 11 starts the estimation process and executes the processes from step St306 to step St308 while i is less than or equal to N (step St305). At this time, the processor 11 focuses on the i-th section of the divided travel paths for each transport device, which are generated in step St303 and consist of sections from the first section to the N-th section.

[0144] The processor 11 converts each of the i-th section division travel paths of each transport device, generated in step St 303, into input data for the trained model 61 (step St 306). This generates division path data. Since the value of i is set to 1 in step St 304, the division travel path of the first section out of the N sections of division travel paths for each transport device is converted into division path data.

[0145] The processor 11 generates a congestion information map (step St307) by inputting the divided travel routes for each transport device, which were converted into divided route data in step St306, the automated warehouse map M1, and the time intervals used when dividing the overall travel route in step St303, into the trained model 61. Since the value of i is set to 1 in step St304, the processor 11 can generate a congestion information map for the time intervals corresponding to the divided travel routes in the first section.

[0146] The processor 11 adds 1 to the value of i (step St308). For example, if the value of i is set to 1, step St308 makes the value of i 2.

[0147] The processor 11 repeatedly executes the processes from step St306 to step St308 as long as i is less than or equal to N. This allows the processor 11 to have the trained model 60 output and acquire N traffic congestion information maps.

[0148] Processor 11 concatenates each of the N traffic congestion information maps in chronological order (step St 109). Then, processor 11 terminates this processing flow. In this way, processor 11 can obtain a traffic congestion information map for the entire time period. By checking the traffic congestion information map, the user can understand when and where traffic congestion occurs in the automated warehouse. The user can then consider, for example, planning operations in the automated warehouse.

[0149] [Trained Model] Next, a method for generating a trained model 61 will be described with reference to Figures 15, 16, and 17. Figure 15 is a schematic diagram illustrating the training data according to Embodiment 2.

[0150] The processor 11 generates a trained model 61 by training the model using multiple pairs of input data and ground truth data as training data. The input data is the same as described in Embodiment 1, so its description is omitted here. The training algorithm for generating the trained model 61 is not particularly limited, but machine learning including neural network technologies such as CNN and U-Net, and deep learning technologies may be used, for example.

[0151] The processor 11 uses a traffic congestion information map generated by simulation as the ground truth data. In the example in Figure 15, the processor 11 divides the overall travel route VR5 of the transport device V5 into time intervals TS2 to generate divided travel routes DR53. Although not shown in the figure, the processor 11 also divides the overall travel routes of each transport device other than transport device V5 into time intervals TS2 to generate divided travel routes. The divided travel routes of each transport device may be used in the simulation to generate the traffic congestion information map.

[0152] To generate training data (ground truth data), the processor 11 generates a congestion information map for time intervals TS2 corresponding to a segmented travel route (in the case of transport equipment V5, segmented travel route DR53) in an automated warehouse by performing a simulation of operations in the automated warehouse. The processor 11 may convert the segmented travel route into segmented route data for use in the simulation. In this case, the processor 11 may use, for example, discrete simulation technology.

[0153] The congestion information map generated by the simulation, corresponding to the segmented driving routes of a certain section at time intervals TS2, is used as ground truth data. Next, the flow of generating the trained model 61 will be explained with reference to Figure 16. Figure 16 is a schematic diagram illustrating the flow of generating the trained model according to Embodiment 2.

[0154] In the example shown in Figure 16, it is assumed that the overall travel path of each of the six transport devices shown in Figure 5 is divided at time intervals TS2, generating multiple divided travel paths. Here, as an example for explanation, let's assume that the overall travel path of each transport device is divided into M sections, and M divided travel paths are generated.

[0155] We focus on the m-th section of the M divided travel paths of each transport device: divided travel paths DR7, DR8, DR9, DR10, DR11, and DR12. Here, m is a natural number from 1 to M.

[0156] The divided travel paths DR7, DR8, DR9, DR10, DR11, and DR12 are the travel paths of transport equipment V1, V2, V3, V4, V5, and V6, respectively. The processor 11 of the arithmetic unit 10 converts each divided travel path into divided path data DRD7, DRD8, DRD9, DRD10, DRD11, and DRD12, respectively, in order to use them as input data for model 60a. Model 61a is the model in the training phase of the trained model 61. In other words, once the training of model 61a is complete, the trained model 61 is generated.

[0157] Furthermore, the processor 11 uses the time interval TS2 and the automated warehouse map M1 as input data to the model 61a. In other words, the processor 11 inputs the data of the divided travel path in the mth section for each transport device, the map M1, and the time interval TS2 to the model 61a.

[0158] Model 61a outputs a congestion information map of the automated warehouse for a predetermined time interval corresponding to a predetermined section, in response to inputs of data on the divided travel route in a predetermined section for each transport device, a map M1 of the automated warehouse, and a predetermined time interval corresponding to the predetermined section. Therefore, in response to inputs of data on the divided travel route in the m-th section for each transport device, a map M1, and a time interval TS2, Model 61a outputs a congestion information map JM2 for the time interval TS2 corresponding to the m-th section.

[0159] The processor 11 generates a traffic congestion information map JM3 for the time interval corresponding to the mth section through simulation. The processor 11 uses the traffic congestion information map JM3 generated by simulation as ground truth data.

[0160] The processor 11 adjusts the parameters of model 61a so that the difference between the traffic congestion information map JM2 output from model 61a and the traffic congestion information map JM3 generated by the simulation becomes smaller. The processor 11 generates a trained model 61 by repeatedly adjusting the parameters of model 61a in this way.

[0161] Next, the trained model generation process will be explained following the flowchart with reference to Figure 17. Figure 17 is a flowchart of the trained model generation process according to Embodiment 2. Note that each process in the flowchart shown in Figure 17 is performed by the processor 11 of the arithmetic unit 10. Furthermore, it is assumed that there are multiple orders in the automated warehouse, multiple stations are provided in the automated warehouse, and multiple transport devices are installed. In addition, it is assumed that the processor 11 assigns multiple transport devices to each of the multiple stations.

[0162] The processor 11 assigns all orders to each of the multiple stations located in the automated warehouse (step St400).

[0163] In step St400, the processor 11 assigns the multiple tasks included in the multiple orders assigned to each station to each of the multiple transport devices, and generates a travel route for each order and each transport device (step St401). At this time, the tasks included in the orders assigned to a station are assigned to the transport device assigned to that station.

[0164] The processor 11 generates the overall travel path for each transport device by concatenating the travel paths generated in step St401 in the order of processing (step St402).

[0165] The processor 11 divides the overall travel path into M sections at arbitrary time intervals for each transport device and generates M divided travel paths (step St403).

[0166] The processor 11 sets the value i, which is used to determine the end of the simulation process that started in step St405, to 1 (step St404).

[0167] The processor 11 starts the simulation process and executes the processes from step St406 to step St409 while i is less than or equal to M (step St405). At this time, the processor 11 focuses on the i-th section of the divided travel paths for each transport device, which are generated in step St403 and consist of sections from the first section to the M-th section.

[0168] The processor 11 converts each of the i-th section division travel paths generated in step St403 for each transport device into input data to be used for simulation, i.e., division path data (step St406). Since the value of i is set to 1 in step St204, the division travel path of the first section out of the M sections of division travel paths for each transport device is converted into division path data.

[0169] The processor 11 generates a congestion information map of the automated warehouse for time intervals corresponding to the divided travel paths of the i-th section for each transport device, which were converted into divided route data in step St406, through simulation (step St407).

[0170] The processor 11 records learning data (step St 408) using the divided travel routes for each transport device converted into divided route data in step St 406, the automated warehouse map M1, and the time intervals used when dividing the overall travel route in step St 403 as input data, and the congestion information map generated by simulation in step St 407 as ground truth data. The learning data may be stored in memory 12, for example. Since the value of i is set to 1 in step St 404, the processor 11 can record learning data including congestion information maps for the time intervals corresponding to the first section for each transport device.

[0171] The processor 11 adds 1 to the value of i (step St409). For example, if the value of i is set to 1, step St409 makes the value of i 2.

[0172] The processor 11 repeatedly executes the processes from step St406 to step St409 as long as i is less than or equal to M. This allows the processor 11 to generate congestion information maps for each time interval used to divide the overall travel route in step St403. In other words, the processor 11 can generate M congestion information maps. The processor 11 takes the divided travel route, map M1, and time interval for the mth section for each transport device as input data, and uses the congestion information map generated based on the divided travel route for the mth section from the M congestion information maps generated by the simulation as the correct answer data. The processor 11 can generate and record multiple pairs of input data and correct answer data as training data.

[0173] The processor 11 generates a trained model 61 based on multiple training data sets generated by repeating the processes from step St 406 to step St 409 (step St 410). Then, the processor 11 terminates this processing flow.

[0174] In this way, the processor 11 can generate a trained model 61 capable of generating a congestion information map showing the congestion status of transport equipment in an automated warehouse.

[0175] (Summary of Embodiment 2) The following technology is disclosed based on the description of Embodiment 2 above. The components etc. in Embodiment 2 above are examples, but are not limited to these.

[0176] (Technology 5) The method for generating a traffic congestion information map involves assigning multiple tasks that constitute work instructions in an automated warehouse to each of multiple transport devices, dividing the overall travel route when processing tasks for each transport device into N (N: a natural number of 2 or more) segments at predetermined first time intervals, thereby generating multiple divided travel routes for each transport device, and then, in response to inputs of the divided travel route in a predetermined segment for each transport device, the automated warehouse map, and a predetermined time interval corresponding to the predetermined segment, a trained model capable of outputting a traffic congestion information map of the automated warehouse at a predetermined time interval is executed for each of the N segments, inputs of the divided travel route in the nth (n: a natural number from 1 to N) segment for each transport device, the map, and the first time interval, and finally concatenating the N traffic congestion information maps output from the trained model in a time series.

[0177] This allows the congestion information map generation method to generate a congestion information map that shows the congestion status of conveying equipment in an automated warehouse as it changes over time. As a result, users such as automated warehouse managers can easily understand when and where congestion of conveying equipment occurs by checking the congestion information map.

[0178] (Technology 6) The congestion information map generation method described in Technology 5 generates multiple divided travel routes for each transport device by dividing the overall travel route when processing tasks for each transport device into M (M: a natural number of 2 or more) sections at predetermined second time intervals, thereby dividing it into M sections. Based on the multiple divided travel routes for each transport device generated, M congestion information maps are generated for each second time interval by simulating work in an automated warehouse. A trained model is generated based on training data that uses multiple pairs of divided travel routes, maps, and second time intervals for the m (m: a natural number from 1 to M) section for each transport device, and congestion information maps generated from the M congestion information maps generated by the simulation based on the divided travel route in the m section.

[0179] As a result, the traffic congestion map generation method can generate a trained model that outputs a traffic congestion map for the time interval corresponding to the section of the divided travel route, based on inputs of time intervals, the divided travel routes of multiple transport devices, and the map of the automated warehouse. Furthermore, the traffic congestion map generation method can generate a large amount of training data at high speed and generate a highly accurate trained model by, for example, using discrete simulation technology for the simulation.

[0180] (Technology 7) The traffic congestion map generation program causes the computing unit to assign multiple tasks constituting work instructions in the automated warehouse to each of the multiple transport devices, and divides the overall travel route when processing the tasks for each transport device into N (N: a natural number of 2 or more) sections at predetermined first time intervals, thereby generating multiple divided travel routes for each transport device. In response to the input of the divided travel route in a predetermined section for each transport device, the map of the automated warehouse, and a predetermined time interval corresponding to the predetermined section, the program causes a trained model capable of outputting a traffic congestion map of the automated warehouse at a predetermined time interval to execute the input of the divided travel route, map, and first time interval for the nth (n: a natural number from 1 to N) section for each transport device for each of the N sections, and concatenates the N traffic congestion maps output from the trained model in a time series.

[0181] As a result, the traffic congestion information map generation program can achieve the same effect as technology 5.

[0182] The functions of the various embodiments described above can also be realized by supplying programs and applications for realizing the functions of the various embodiments described above to a system or device using a network or storage medium, and by having one or more processors in the computer of that system or device read and execute the programs.

[0183] Furthermore, the functions of the various embodiments described above may be realized by circuits that implement one or more functions (for example, an Application Specific Integrated Circuit (hereinafter referred to as "ASIC") or an FPGA).

[0184] Although various embodiments have been described above with reference to the attached drawings, this disclosure is not limited to such examples. It will be obvious to those skilled in the art that various modifications, alterations, substitutions, additions, deletions, and equivalents can be conceived within the scope of the claims, and these will also be understood to fall within the technical scope of this disclosure. Furthermore, the components of the various embodiments described above can be arbitrarily combined without departing from the spirit of the invention.

[0185] This application is based on a Japanese patent application (JP 2025-024992) filed on February 19, 2025, the contents of which are incorporated by reference within this application.

[0186] The technology disclosed herein is useful as a processing time estimation method, a processing time estimation program, a traffic congestion information map generation method, and a traffic congestion information map generation program.

[0187] 10 Arithmetic unit 11 Processor 12 Memory 13 Communication device 14 Input device 15 External interface device 16 Display device 17 Internal interface device 60, 61 Trained models 60a, 61a Model M1 Map OD1, OD2, OD3, OD4 Order OS, BR1, BR2, BR3, PK1, PK2, PK3, RT1, RT2, RT3, SP, TSK1, TSK2 Task S1, S2, S3 Station a, b, V1, V2, V3, V4, V5, V6 Transport equipment VR1, VR2, VR3, VR4, VR5, VR6 Overall travel route DR, DR1, DR2, DR3, DR4, DR5, DR6, DR7, DR8, DR9, DR10, DR11, DR12, DR50, DR51, DR52, DR53 Divided Driving Routes DRD, DRD1, DRD2, DRD3, DRD4, DRD5, DRD6, DRD7, DRD8, DRD9, DRD10, DRD11, DRD12 Divided Route Data TS1, TS2 Time Intervals JM1, JM2, JM3 Traffic Congestion Information Map

Claims

1. A method for estimating processing time, comprising: assigning multiple tasks constituting work instructions in an automated warehouse to each of multiple transport devices; dividing the overall travel path when processing the tasks for each transport device into N (N: a natural number of 2 or more) sections at predetermined first time intervals to generate multiple divided travel paths for each transport device; and, in response to inputs of the divided travel paths in predetermined sections for each transport device, a map of the automated warehouse, and predetermined time intervals corresponding to the predetermined sections, a trained model capable of outputting estimated travel times in the divided travel paths in predetermined sections for each transport device, receiving inputs of the divided travel paths in the nth (n: a natural number from 1 to N) section for each transport device, the map, and the first time interval for each of the N sections; and summing the estimated travel times in the divided travel paths in each of the N sections for each transport device output from the trained model to estimate the processing time of the work instructions.

2. A method for estimating processing time according to claim 1, comprising: assigning all work instructions in the automated warehouse to each of a plurality of stations provided in the automated warehouse for processing the work instructions; assigning each of the plurality of transport equipment to one of the plurality of stations; estimating the processing time for each of the plurality of stations by summing the estimated travel times in the divided travel paths in each of the N sections for each transport equipment assigned to the station; and estimating the processing time for all work instructions by summing the processing times for each of the plurality of stations.

3. A method for estimating processing time according to claim 1, comprising: dividing the overall travel route for processing the task for each transport device into M (M: a natural number of 2 or more) sections at predetermined second time intervals to generate a plurality of divided travel routes for each transport device; calculating the estimated travel time for each of the M sections of the divided travel route for each transport device by simulating the work in the automated warehouse based on the plurality of divided travel routes for each transport device generated; and generating the trained model based on training data that uses a plurality of pairs of the divided travel route in the m (m: a natural number from 1 to M) section for each transport device, the map and the second time interval, and the estimated travel time for the divided travel route in the m section calculated by the simulation.

4. A processing time estimation program that causes a computing device to assign multiple tasks constituting work instructions in an automated warehouse to each of multiple transport devices, divides the overall travel path when processing the tasks for each transport device into N (N: a natural number of 2 or more) sections at predetermined first time intervals, thereby generating multiple divided travel paths for each transport device, and, in response to inputs of the divided travel paths in predetermined sections for each transport device, a map of the automated warehouse, and predetermined time intervals corresponding to the predetermined sections, causes a trained model capable of outputting estimated travel times in the divided travel paths in predetermined sections for each transport device to execute the inputs of the divided travel paths in the nth (n: a natural number from 1 to N) section for each transport device, the map, and the first time interval for each of the N sections, and estimates the processing time of the work instructions by summing the estimated travel times in the divided travel paths in each of the N sections for each transport device output from the trained model.

5. A method for generating a traffic congestion map, comprising: assigning multiple tasks constituting work instructions in an automated warehouse to each of multiple transport devices; dividing the overall travel route when processing the tasks for each transport device into N (N: a natural number of 2 or more) sections at predetermined first time intervals to generate multiple divided travel routes for each transport device; and, in response to inputs of the divided travel routes in predetermined sections for each transport device, a map of the automated warehouse, and predetermined time intervals corresponding to the predetermined sections, executing the inputs of the divided travel routes in the nth (n: a natural number from 1 to N) section for each transport device, the map, and the first time interval for each of the N sections; and concatenating the N traffic congestion maps output from the trained model in a time series.

6. A method for generating a traffic congestion map according to claim 5, comprising: dividing the overall travel route for processing the task for each transport device into M (M: a natural number of 2 or more) sections at predetermined second time intervals to generate a plurality of divided travel routes for each transport device; generating M traffic congestion information maps for each second time interval by simulating operations in the automated warehouse based on the plurality of divided travel routes for each transport device generated; and generating a trained model based on training data that uses a plurality of pairs of the divided travel route, the map, and the second time interval in the m (m: a natural number from 1 to M) section for each transport device, and the traffic congestion information map generated from the M traffic congestion information maps generated by the simulation based on the divided travel route in the m section.

7. A traffic congestion information map generation program that causes a computing device to assign multiple tasks constituting work instructions in an automated warehouse to each of multiple transport devices, divides the overall travel route when processing the tasks for each transport device into N (N: a natural number of 2 or more) sections at predetermined first time intervals, thereby generating multiple divided travel routes for each transport device, and, in response to inputs of the divided travel routes in predetermined sections for each transport device, a map of the automated warehouse, and predetermined time intervals corresponding to the predetermined sections, causes a trained model capable of outputting a traffic congestion information map of the automated warehouse at predetermined time intervals to execute the inputs of the divided travel routes in the nth (n: a natural number from 1 to N) section for each transport device, the map, and the first time interval for each of the N sections, and concatenates the N traffic congestion information maps output from the trained model in a time series.

Images