Transportation route determination method, transportation route determination device, and computer program

Abstract

This invention provides a method for determining transportation routes, a device for determining transportation routes, and a computer program. [Solution] The method for determining a transport route involves a computer defining a transport network in which nodes are assigned to the originating points of multiple goods, points that transport equipment can pass through, and destination points; assigning a virtual potential including positive or negative signs and magnitude to the surrounding edges according to the quantity of goods to be shipped from the point corresponding to each node; assigning a virtual potential including the opposite positive or negative signs and magnitude to the surrounding edges according to the quantity of goods to be transported to the point corresponding to each node; assigning a virtual charge including positive or negative signs and magnitude to the transport equipment according to the empty state or loading rate of the cargo bed; performing a simulation of the movement of the transport equipment based on the attractive or repulsive force of the transport equipment's charge on the potential of each edge; and extracting a travel route from the travel routes between points obtained from the simulation based on a predetermined evaluation quantity.

Description

Technical Field

[0006]

[0001] The present invention relates to a transport route determination method, a transport route determination apparatus, and a computer program that contribute to solving the transport problems in logistics.

Background Art

[0002] There is a demand throughout the logistics industry to optimize routes in a logistics system. In Patent Document 1, a method has been proposed for optimizing logistics based on geographical information by adding labels to transport routes such as trunk transport and regional transport.

[0003] Regarding methods for solving transport problems, such as Patent Document 1, various methods have been proposed. However, the operation of loading as many goods as possible on the vehicle from the departure point to the destination and returning with an empty vehicle remains common even today.

[0004] In trunk transport, it is common practice to improve efficiency by aggregating and transporting goods destined for the same or neighboring bases as transit points or destinations. However, considering labor conditions, it is necessary to return the driver of a transport vehicle such as a truck. It is difficult for a transport vehicle such as a truck to conveniently load goods at the delivery destination of the goods and return, and it still remains common to return with an empty vehicle.

[0005] Therefore, as proposed in Patent Documents 2 and 3, a method of sequentially determining a transport route on the premise of transshipping goods at an intermediate base and instructing the transport vehicle of the destination has begun to be adopted.

Prior Art Documents

Patent Documents

[0006]

Patent Document 1

Patent Document 2

Patent Document 3

[0007] Calculating the optimal route from the origin to the final destination of a package, while allowing transshipment at any number of locations, requires enormous computational resources, and furthermore, recalculations are necessary as the situation changes sequentially. Numerous packages are delivered from various origins to various destinations, and determining the optimal transport route considering not only the combination of origins and destinations but also any one or more intermediate stops along the way can take a considerable amount of time to find a solution.

[0008] The present invention aims to provide a transportation route determination method, a transportation route determination device, and a computer program that contribute to calculating transportation routes for transporting goods more quickly and appropriately. [Means for solving the problem]

[0009] A transport route determination method according to one embodiment of the present disclosure defines a transport network in which a computer performs a simulation using agent-based modeling in which the transport equipment is an agent moving between the nodes, assigning to nodes the point where goods are dispatched, to points that transport equipment transporting the goods can pass through, and to points where goods are delivered, and assigning to edges between the nodes the routes connecting adjacent points, and the computer performs a simulation using agent-based modeling in which the transport equipment is an agent moving between the nodes, wherein the computer assigns a virtual potential including positive or negative signs and magnitude to the surrounding nodes or edges centered on the node, according to the quantity of goods to be dispatched from the point corresponding to each node, and assigns a virtual potential including the opposite positive or negative sign and magnitude to the same point according to the quantity of goods to be transported to the point corresponding to each node A simulation is performed in which an electric potential is applied to surrounding nodes or edges centered on the node, a virtual charge including positive or negative signs and magnitude is applied to the transport equipment according to the empty state or loading rate of the cargo bed, the probability of selecting an edge to the other node is determined based on the attractive or repulsive force of the transport equipment to the virtual charge with respect to the virtual electric potential applied to each edge in order for the transport equipment to move from one node to another adjacent node, an edge is selected based on the probability, and the process of applying the virtual electric potential and virtual charge, determining the probability, and selecting the edge is repeated until the delivery of the goods is completed, and a process is performed to extract the movement path of the transport equipment from the movement paths between points obtained in the simulation based on a predetermined evaluation quantity.

[0010] In the transportation route determination method of this disclosure, numerical values ​​(virtual potentials) with opposite positive and negative signs are assigned to the nodes or edges surrounding the node corresponding to the origin (dispatch point) of the goods and to the nodes or edges surrounding the node corresponding to the destination (arrival point) of the goods. When the cargo bed is empty, transportation equipment is assigned numerical values ​​(virtual charges) that make it easier to move towards the origin by attractive forces from the node corresponding to the origin or the edges leading to that node, and conversely, that make it easier to avoid the destination by repulsive forces from the node corresponding to the destination or the edges leading to that node.

[0011] Virtual charge corresponds to each moving load or its quantity, and virtual potential is a field induced at the node corresponding to each stationary point due to the load present at that point. Virtual potential is a field that generates attractive / repulsive forces with respect to virtual charge. The attractive / repulsive forces here are determined such that the larger the value of the virtual potential or virtual charge, the stronger the attractive force from surrounding nodes or edges of other nodes, the closer the selection probability of the edge leading to that node approaches 1, and the stronger the repulsive force from surrounding nodes or edges of other nodes, the closer the selection probability of the edge leading to that node approaches 0.

[0012] By setting virtual potentials and virtual charges in this way, it is possible to represent that empty transport equipment is more likely to head towards the origin of goods (dispatch point in the network), and transport equipment loaded with goods is more likely to head towards the destination of goods (arrival point in the network), in order to determine the optimal route for the movement of transport equipment. By setting virtual potentials and virtual charges, an attractive force acts between each point and the transport equipment passing through each point, causing negative charges to gather in areas with high potential and a repulsive force acts to avoid negative charges, enabling the determination of the optimal route in the simulation. [Effects of the Invention]

[0013] According to this disclosure, it will be possible to calculate the optimal transportation route more quickly and accurately. [Brief explanation of the drawing]

[0014] [Figure 1] This is an overview diagram of the logistics system disclosed herein. [Figure 2] This is a block diagram showing the configuration of a transport route determination device. [Figure 3] This block diagram shows the configuration of the terminal device used by the operator. [Figure 4] This is an explanatory diagram for converting map data to a transportation network. [Figure 5] This flowchart shows an example of the process for determining transportation routes for each region. [Figure 6]It is a flowchart showing an example of a process for determining a transportation route for each region. [Figure 7] It is a diagram showing an example of the content of a method for deriving an optimal route. [Figure 8] It is a diagram showing an example of the content of a method for deriving an optimal route. [Figure 9] It is a diagram showing an example of the content of a method for deriving an optimal route. [Figure 10] It is a diagram showing an example of the content of a method for deriving an optimal route. [Figure 11] It is a diagram showing an example of the content of a method for deriving an optimal route. [Figure 12] It is a diagram showing an example of the content of a method for deriving an optimal route. [Figure 13] It is a flowchart showing an example of a transportation route determination processing procedure by a transportation route determination device. [Figure 14] It is a flowchart showing an example of a transportation route determination processing procedure by a transportation route determination device. [Figure 15] It shows the relationship between the overall pallet route and the segmented transportation network. [Figure 16] It shows an example of delivery conditions in Variation 1. [Figure 17] It is a flowchart showing an example of a process for determining a transportation route for each region in Variation 1. [Figure 18] It is a flowchart showing an example of a process for determining a transportation route for each region in Variation 1. [Figure 19] It is a diagram showing an example of the content of a method for deriving an optimal route in Variation 1. [Figure 20] It is a diagram showing an example of the content of a method for deriving an optimal route in Variation 1. <00001​​​​​​​​​​​​This disclosure will be described in detail with reference to drawings illustrating its embodiments. The following embodiments will describe a logistics system including a transportation route determination method and a transportation route determination device according to this disclosure.

[0016] Figure 1 is an overview diagram of the logistics system 100 of this disclosure. The logistics system 100 includes a transport device 1 for transporting goods such as agricultural products and machinery, a base 2 which is a transfer center that the transport device 1 stops at, a collection and delivery center 3 which is a distribution center that collects and delivers goods, and a control center 4.

[0017] The logistics system 100 is operated based on route information determined by the transport route determination device 400 in the control center 4. The transport route determination device 400 can communicate with transport equipment 1, base controllers 20 installed at base 2, and collection / delivery controllers 30 installed at collection / delivery center 3. The transport route determination device 400 can communicate with terminal devices 5 used by shippers of goods and operators at various locations such as base 2 and collection / delivery center 3.

[0018] Transportation equipment 1 is a vehicle such as a transport truck, train, airplane, or ship. Transportation equipment 1 may also be a transport robot that operates autonomously.

[0019] Base 2 is a transit center located on a so-called trunk line in logistics, and is installed at locations that serve as bases for the transport equipment 1, such as ports, airports, freight stations, and interchanges (ICs) in the road network. Base 2 is provided, for example, at predetermined intervals. Base 2 is equipped with a group of devices that accept the arrival of transport equipment 1 at Base 2 and load and unload pallets P, which are logistics components, from transport equipment 1, and a base controller 20 that controls the group of devices. The base controller 20 can communicate with the transport route determination device 400 and controls the group of devices based on instructions from the transport route determination device 400.

[0020] In the logistics system 100 of this disclosure, the goods to be transported are loaded onto logistics components and transported. Specifically, the logistics components are pallets P and small containers C. The following description will focus on pallets P, but small containers C may be treated similarly. Logistics components may also be flexible containers, so-called FIBCs, steel containers, or cardboard boxes. In addition, bags, boards, and boxes used for loading or storing goods in logistics are also included in logistics components.

[0021] The collection and delivery center 3 is a location for collecting and delivering goods to shippers (including consignors and consignees). The collection and delivery center 3 is also a base for receiving and storing goods from consignors and transporting them to consignees. The collection and delivery center 3 also functions as a wholesale or distribution center. Multiple collection and delivery centers 3 may be established for each base 2. The collection and delivery center 3 may be managed by a consignee, which is a retailer. When the collection and delivery center 3 collects goods to be transported to base 2, the goods may be loaded onto pallets P at the collection and delivery center 3. The collection and delivery center 3 is equipped with a collection and delivery controller 30 that receives instructions from the transport route determination device 400. The collection and delivery controller 30 receives input from operators at the collection and delivery center 3 regarding outgoing goods and pallets P, as well as input from arriving goods. The collection and delivery controller 30 stores the correspondence between the pallet identification data of the pallet P on which the goods to be shipped are stacked and the identification data of the goods, and the correspondence between the identification data of the arrived goods and the pallet identification data of the pallet P on which the goods are stacked, and transmits this information to the transport route determination device 400.

[0022] A region has one or more bases 2, and each base 2 has multiple collection and distribution centers 3. Regions are not limited to administrative divisions and can be defined in any unit, and are stored using latitude and longitude information, identification data for bases 2 and collection and distribution centers 3, and identification data for the region to which they belong. Regions may overlap. Collection and distribution centers 3 may belong to different regions. Bases 2 and collection and distribution centers 3 are both collection points and distribution points in transportation within a region (corresponding to "collection and distribution points"). Multiple collection and distribution centers 3 within a region are each both collection points and distribution points. Any collection and distribution center 3 may function only as a collection point, and similarly, any other collection and distribution center 3 may function only as a distribution point.

[0023] The transport route determination device 400 sequentially acquires the position of the transport equipment 1 and stores the position data indicating the position of the transport equipment 1 in association with the transport equipment identification data of the transport equipment 1. The transport route determination device 400 also stores the pallet identification data of the pallet P contained in the transport equipment 1 in association with the storage section identification data that identifies the storage location in the transport equipment 1. The correspondence between the pallet identification data of the pallet P and the identification data of the goods loaded on the pallet P is transmitted from the collection and delivery center 3 and stored. Therefore, the transport route determination device 400 can sequentially identify the position of the transport equipment 1 containing the goods.

[0024] In the logistics system 100 of this disclosure, the transport equipment 1 transports goods in pallet units P. In the logistics system 100, the goods to be transported are not limited to being transported by a single transport equipment 1 from the consignor's collection and distribution center 3 to the consignee's collection and distribution center 3, but are permitted to be transshipped in pallet units P at intermediate hubs 2. The goods are transported by changing between one or more transport equipment 1. The route along which the goods are transported is hereinafter referred to as the pallet P movement route. The pallet P movement route is the history of the storage compartments of one or more transport equipment 1 in which the pallet P is stored until it is delivered to the consignee. In this way, the transport equipment 1 also transports goods by transshipping pallets P stored at intermediate hubs 2. The route along which the transport equipment 1 transports goods, for example, from the starting point "A" hub 2, the intermediate hub "B" hub 2, the destination "C" hub 2, and back to the final destination hub "E" via the destination hub "D" hub 2, is called the transport route. The transportation route of transportation equipment 1 includes data connecting the intermediate points that the equipment visits sequentially for collection or loading / unloading from the departure point (dispatch point) base 2 or collection / delivery center 3 to the destination base 2 or collection / delivery center 3.

[0025] The transport route determination device 400 sequentially updates the position of transport equipment 1 at each point in time and the identification data of the goods contained in transport equipment 1 at that time, and determines the next destination base 2 or collection / delivery center 3 (transport route) for each of the multiple transport equipment 1. The transport route determination device 400 obtains pallet identification data of pallets P already loaded with goods that are expected to be present at base 2 and collection / delivery center 3 at each point in time from pallets P being transported by transport equipment 1 and the scheduled shipment of goods. At each base 2 or collection / delivery center 3 that each transport equipment 1 will visit, the transport route determination device 400 determines the pallet identification data of pallets P of goods to be shipped out of transport equipment 1 and the pallet identification data of pallets P of goods to be loaded into transport equipment 1. The transport route determination device 400 instructs transport equipment 1 of the determined transport route and instructs base controllers 20 and collection / delivery controllers 30 of the determined pallet identification data of pallets P to be shipped out and loaded. The transport route determination device 400 sequentially updates the correspondence between pallet identification data of pallets P housed in transport equipment 1, which changes with loading and unloading, and storage unit identification data that identifies the storage location in transport equipment 1.

[0026] Figure 2 is a block diagram showing the configuration of the transportation route determination device 400. The transportation route determination device 400 is a server computer and comprises a processing unit 40, a storage unit 41, and a communication unit 42. The transportation route determination device 400 may not only be configured to use a single server computer (hardware), but may also be configured to distribute processing among multiple server computers, or it may be one of multiple server computers (instances) virtually generated on a large computer. Except for the information update processing of the logistics DB 410 in the storage unit 41, the transportation route determination device 400 may perform calculations using a quantum computer.

[0027] The processing unit 40 is a processor using a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit). The processing unit 40 executes processing using built-in memory such as ROM (Read Only Memory) and RAM (Random Access Memory). The processing unit 40 can sequentially acquire time information using a built-in timer.

[0028] The storage unit 41 includes a non-volatile storage medium such as a hard disk or an SSD (Solid State Drive). The storage unit 41 stores the information processing program P4. The processing unit 40 performs a process to derive the optimal path based on the information processing program P4 stored in the storage unit 41, using calculations described later.

[0029] The memory unit 41 stores a web server program, and the processing unit 40 performs web server functions. Through these web server functions, it may receive requests from terminal devices 5 for sharing the cargo area of ​​the transport equipment 1, etc.

[0030] A logistics database (DB) 410 is constructed in the storage unit 41 or an external storage device. The processing unit 40 can read from and write to the logistics database 410 using a database operation module. The logistics database 410 stores, for example, a user database 411, a pallet database 412, and a transport equipment database 413, as will be described later. The logistics database 410 also stores a transport route database 414, which is derived by calculations described later.

[0031] The communication unit 42 enables communication over the public communication network N1 or the carrier network N2. The processing unit 40 can send and receive information with the terminal device 5 via the public communication network N1 or the carrier network N2 using the communication unit 42. The processing unit 40 can also send and receive information with the in-vehicle unit installed on the transportation equipment 1 via the carrier network N2 using the communication unit 42. The processing unit 40 can communicate with the base controller 20 at base 2 and the collection and delivery controller 30 at collection and delivery center 3 via the public communication network N1, the carrier network N2, or a dedicated line using the communication unit 42.

[0032] Figure 3 is a block diagram showing the configuration of the terminal device 5 used by the operator. The terminal device 5 uses a desktop, laptop, or tablet type personal computer. The terminal device 5 may also be a smartphone. The terminal device 5 comprises a processing unit 50, a storage unit 51, a communication unit 52, a display unit 53, and an operation unit 54.

[0033] The processing unit 50 is a processor using a CPU or GPU. The processing unit 50 executes processing based on a program stored in the storage unit 51 using built-in memory such as ROM and RAM.

[0034] The storage unit 51 includes a non-volatile storage medium such as a hard disk or SSD. The storage unit 51 stores the terminal program 50P. The processing unit 50 receives operator input based on the terminal program 50P stored in the storage unit 51 and executes a process to display data.

[0035] The communication unit 52 enables communication over the public communication network N1 or the carrier network N2. The processing unit 50 can send and receive information with the transport route determination device 400 via the public communication network N1 or the carrier network N2 using the communication unit 52.

[0036] The display unit 53 is a display such as a liquid crystal display or an organic EL (Electro-Luminescence) display. The display unit 53 displays a screen based on the data stored in the storage unit 51.

[0037] The operation unit 54 is a user interface such as a keyboard and pointing device that can input and output to and from the processing unit 50. The operation unit 54 may also be an audio input unit. The operation unit 54 may also be a touch panel of the display unit 53.

[0038] This disclosure describes the process by which the transport route determination device 400, configured as described above, determines the movement path of the pallet P and the transport route of the transport equipment 1 based on information about the goods to be transported, the position data of the transport equipment 1, etc.

[0039] The transport route determination device 400, in determining the transport route of transport equipment 1 and the movement route of pallet P, divides the area to be transported into "regions," determines the optimal route for each region based on the delivery conditions of goods within that region, and connects the optimal routes for each region to determine the transport route and movement route. When calculating the optimal route for each region, the transport route determination device 400 creates either a grid-like transport network that holds actual distance information as separate data, or a graph-like transport network with no regularity. For each region, the transport route determination device 400 performs route calculation processing by running agent-based modeling simulations using the transport network. The transport route determination device 400 may also calculate the route as a graph connecting the optimal nodes using a GNN (Graph Neural Network).

[0040] Figure 4 is an explanatory diagram of the conversion from map data to a transportation network. In Figure 4, image A shows an example of the original map data, image B shows the transportation network created from the map data, and image C shows the segmented transportation network. As shown in A and B, the transportation network is created by assigning intersections, junctions, or passable points (hubs 2 or distribution centers 3) to vertices (nodes) arranged in a grid pattern, starting from a road network that is by no means a grid. The geographical length between each point does not correspond to the length in the transportation network, and it is better to simply store it as distance between nodes as separate data. The transportation network in B is segmented into a size that can be computed (Figure 4(C)). The size that can be computed should be appropriately set based on the processing capacity of the transportation route determination device 400, based on trial computation times as described later. As shown in images B and C, edges are removed when there are no corresponding roads on the map. By converting to such a transportation network, the road traffic network can be standardized as data.

[0041] The transportation network shown in Figure 4 is automatically created by the processing unit 40 of the transportation route determination device 400 based on map data acquired from the communication unit 42, and is not limited to a grid; it may also be a graph-like transportation network represented by nodes and edges. Alternatively, the transportation network may be created by the processing unit 40 of the transportation route determination device 400 or another computing unit, displaying the temporarily created transportation network on the display unit 53 of the terminal device 5, and accepting modifications such as edge deletion or edge addition from the operator of the terminal device 5.

[0042] Next, we will explain the calculation of routes using agent-based modeling simulation with a transportation network by the transportation route determination device 400. Figures 5 and 6 are flowcharts showing an example of the transportation route determination process for each region. The transportation route determination device 400 (or other calculation device) pre-executes the following processes for each transportation network corresponding to each region and stores them in the transportation route DB 414.

[0043] The processing unit 40 sets delivery conditions for the target transport network, including the initial position and number of transport devices 1 (agents), the initial position and number of pallets P, and the destination (destination) of each pallet P within the network (S101). In S101, the processing unit 40 associates the number of pallets P initially loaded onto transport devices 1 (e.g., zero) with the agents. The number of transport devices 1 may be one or more.

[0044] Delivery conditions may include the location of transport equipment 1 (agent) at a specific node in the transport network at a specific time to load and unload pallets P. For example, 100 pallets P are located at a specific node (hub 2 or distribution center 3, which is the originating point within the region), and these pallets P are to be transported to three other nodes (destination points within the region) in quantities of 40, 30, and 20 respectively. The delivery condition may also stipulate that the initial location of transport equipment 1 is not limited to hub 2 but is at a different node. There may be one or more types of pallets P.

[0045] The processing unit 40 assigns a virtual potential (signed numerical value) to the edges of each node in the transport network that are directly and indirectly heading to a specific node corresponding to the origin point, based on the delivery conditions (S102). The magnitude of the numerical value assigned in S102 is determined according to the amount of pallets P to be shipped from the specific node. In S102, the processing unit 40 assigns a signed numerical value not only to the edges that are directly heading to the specific node, but also to the surrounding edges, decreasing the absolute value of the numerical value according to the distance from the specific node. In S102, the processing unit 40 may also assign a combination of numerical values ​​representing multiple features to the edges heading to the specific node, such as (number of pallets P present, signed numerical value (virtual potential described later), ...). In the processing described below, the processing unit 40 assigns the virtual potential to the edges and performs calculations, but even if the target of the assignment is a node, the specific method of calculation for evaluating the attractive or repulsive force acting between the node to which the virtual potential is attached and the edge is within the scope of design.

[0046] Based on the delivery conditions, the processing unit 40 assigns a signed number (S103) to the edges of each node in the transport network that lead directly and indirectly to the node corresponding to the destination point, with the inverse of the number assigned in S102. The magnitude of the number assigned in S103 is determined according to the quantity of pallets P to be transported to the destination point. In S103, the processing unit 40 assigns a signed number not only to the edges directly connected to the node corresponding to the destination point, but also to the surrounding edges, reducing the absolute value of the number according to the distance from the destination point node.

[0047] In S103, the processing unit 40 may also assign a signed number, the opposite of the number assigned in S102, to the edge leading to a node that can be a transit point where transshipment is possible (a node corresponding to a location where a warehouse may be located). In this case, the magnitude of the number should be relatively small compared to the number assigned to the edge leading to the destination.

[0048] The values ​​assigned in S102 and S103 can be interpreted as virtual potentials. This can be expressed as assigning a positive virtual potential with a valency corresponding to the amount of pallet P to the edge heading towards the loading point, and a negative virtual potential with a valency corresponding to the amount of pallet P to the edge heading towards the loading point. The positive and negative signs of the potentials may be reversed.

[0049] The processing unit 40 assigns a signed numerical value (virtual charge) to the transport equipment 1 (agent) in its initial state, based on the delivery conditions, according to the number of pallets to be loaded on the loading platform (S104). In S104, for example, if the number of associated pallets is zero and the loading platform is empty, the processing unit 40 assigns a numerical value of a predetermined size with the opposite sign to that of S102. If the number of pallets associated with the agent is the same as the number of fully loaded storage compartments, the processing unit 40 assigns a numerical value of a predetermined size with the same sign as that of S102.

[0050] In S104, the processing unit 40 may assign a virtual charge according to the empty state or loading rate of the cargo bed. If the cargo bed is empty, the virtual charge may be assigned in such a way that it is easier to move towards the node of the originating point, and if the cargo bed is fully loaded, it is more difficult to move towards the node of the originating point. The numerical value assigned in S104 can also be interpreted as a virtual charge, and can be expressed as assigning a negative virtual charge of a predetermined valence to an empty cargo bed to make it easier to move towards the node of the originating point, and assigning a positive virtual charge of a predetermined valence to a fully loaded cargo bed to make it easier to move towards the node of the arriving point.

[0051] The processing unit 40 may store the settings and assigned values ​​from S101-S104, treating the state at each node (number of palettes P) or the values ​​assigned to the edges leading to each node as a combination of multiple values, such as a vector or matrix.

[0052] The processing unit 40 sets the number of executions to an initial value (S105). The processing unit 40 executes the following process multiple times and, among the multiple executions, derives the path with the fewest number of executions (the number of steps the agent takes from being at one node to moving to the next node) until the pallet P, which was at a specific node that is the originating point, is distributed to the destination point within the transport network, or the path with the shortest distance traveled by the agent.

[0053] The processing unit 40 sets the number of steps to an initial value (S106). The processing unit 40 performs agent-based modeling with the transport equipment 1 as the agent, as shown below, until the pallet P that was located at the specific node that is the originating point has been distributed to each destination point in the network.

[0054] The processing unit 40 initially assigns a reward to each node in the transport network according to the delivery conditions (S107). If there are zero pallets P loaded on the agent's cargo bed, the node corresponding to the originating point receives a reward that increases with the number of pallets P present at the node. Conversely, if the agent's cargo bed is set to be nearly full, the node corresponding to the destination point receives a reward that increases with the number of missing pallets P.

[0055] In that step, the processing unit 40 updates the virtual potential of each edge based on the number of pallets P associated with the agent (the number of pallets P loaded on the transport equipment 1) and the reward given to the node (S108).

[0056] In S108, the processing unit 40, similar to S103 and S104, assigns a signed numerical value (virtual potential) to the edges directly or indirectly connected to the node where the agent resides, with a magnitude corresponding to the number of pallets P remaining at that node, and assigns the agent a signed numerical value (virtual charge) with a magnitude corresponding to the number of pallets P associated with it. It is also preferable to update the signed numerical values ​​on the edges leading to surrounding nodes of the node whose signed numerical value has been updated, so as to propagate them according to the number of edges and distance connected to the node at the discharge point.

[0057] The processing unit 40 determines the selection probability of an edge to an adjacent node according to the attractive or repulsive force based on the signed numerical value (virtual potential) assigned to the edge toward the surrounding nodes of the agent and the signed numerical value (virtual charge) assigned to the target agent (S109).

[0058] In S109, the processing unit 40 evaluates the difference between the signed numbers as an attractive force if the numerical value (virtual charge) assigned to the agent and the numerical value (virtual potential) assigned to the edge toward the adjacent node have opposite signs. If the numerical value (virtual charge) assigned to the agent and the numerical value (virtual potential) assigned to the edge toward the adjacent node have the same sign, the processing unit 40 evaluates the sum of the absolute values ​​of these signed numbers as a repulsive force. The processing unit 40 may also represent the signed numbers as vectors having the direction assigned to that node, and calculate the attractive or repulsive force as the product of the dot product of the unit vector of the edge direction and the virtual charge assigned to the agent. The method of calculating the attractive or repulsive force is not limited to the method described above. The processing unit 40 may calculate the product of the virtual potential or virtual charge assigned to the edge, and if the calculated product is a positive value, it is a repulsive force, and if it is a negative value, it is an attractive force, and the absolute value can be taken as the magnitude of the force. The processing unit 40 determines the selection probability of the edge with the greatest attractive force or the edge with the least repulsive force with the greatest selection probability. The processing unit 40 determines the selection probability based on the magnitude of the attractive or repulsive force, i.e., the product of the virtual potential and virtual charge, and the relationship between them, such as whether their signs are opposite or not (see Figure 7-13).

[0059] Based on the selection probability determined in S109, the processing unit 40 probabilistically selects an edge to the next adjacent node to proceed to (S110). In S110, the processing unit 40 uses the selection probability as the weight in a weighted draw to select edges. The higher the selection probability, the higher the probability that the edge will be selected, but there is still a possibility that an edge with a low selection probability may be selected.

[0060] The processing unit 40 moves the agent via the selected edge (S111). The processing unit 40 determines whether the node to which the agent has moved via the selected edge corresponds to the arrival point of the pallet P loaded on the agent (S112).

[0061] If it is determined in S112 that this is a destination (S112: YES), the processing unit 40 consumes (unloads) the pallets P currently loaded on the cargo bed of the transport device 1, which is the agent, and consumes the reward assigned to the node corresponding to the destination to which the user has moved (S113). In S113, the processing unit 40 subtracts the number of pallets P associated with the agent and subtracts the number of pallets P that are missing at the destination.

[0062] If it is determined in S112 that the destination is not the destination (S112: NO), the processing unit 40 skips S113 and proceeds to S114.

[0063] In S114, the processing unit 40 determines whether the node to which the agent has moved via the selected edge corresponds to the loading point and whether there is sufficient space on the agent's loading platform (S114). The processing unit 40 may also make this determination by determining whether the absolute value of the charge corresponding to the number of pallets P on the agent's loading platform is small or not.

[0064] If in S114 it is determined that the destination is a shipping point and that there is sufficient space on the agent's cargo bed (S114: YES), the processing unit 40 increases the number of pallets P loaded on the cargo bed of the transport device 1, which is the agent, and consumes the reward assigned to the node corresponding to the destination shipping point (S115). In S115, the processing unit 40 adds the number of pallets P associated with the agent and subtracts the number of pallets P to be transported from the shipping point.

[0065] If it is determined in S114 that the location is not a shipping point (S114: NO), the processing unit 40 skips S115 and proceeds to S116. In S116, the processing unit 40 adds the number of steps and stores it (S116). In S116, if distance data is associated with the moved edge, the processing unit 40 adds that distance to the cumulative distance traveled.

[0066] Next, the processing unit 40 determines in S113 and S115 whether or not a delivery that meets the delivery conditions within the transport network has been completed (S117). In S117, the processing unit 40 may determine that the delivery is complete if all signed numerical values ​​(virtual potentials) assigned to each node become zero, or it may determine that the delivery is complete if the correct number of pallets P that meet the delivery conditions have been unloaded at the destination node.

[0067] If it is determined that delivery has not been completed (S117:NO), the process returns to S108.

[0068] If it is determined that delivery is complete (S117:YES), the processing unit 40 stores the history of identification data of the nodes the agent passed through (transport route) and the number of steps (S118).

[0069] The processing unit 40 determines whether the processes in S108-S118 have been executed a predetermined number of times or more (S119). If it is determined that the number of executions is less than the predetermined number (S119: NO), the processing unit 40 adds to the execution count (S120) and returns the process to S106.

[0070] In S119, the processing unit 40 may determine whether it has performed the operation more than a predetermined number of times, or whether it has determined that the optimal solution (minimum number of steps, shortest cumulative distance, or an evaluation quantity converted into an indicator such as carbon dioxide emissions, cost, or driver's working hours) has been obtained through the executions performed so far, as described later, or it may determine whether it can perform the operation further to derive the optimal solution.

[0071] If it is determined in S119 that the process has been executed more than the predetermined number of times (S119: YES), the processing unit 40 extracts the transport route with the minimum number of steps or the shortest cumulative travel distance within the predetermined number of times (S121). The processing unit 40 stores the extracted transport route as the optimal solution under the delivery conditions (S122) and terminates the process.

[0072] The transport route determination device 400 may execute the processing procedures shown in the flowcharts of Figures 5 and 6 according to the actual delivery conditions, or it may execute them according to a plurality of predetermined delivery conditions.

[0073] Figures 7-12 illustrate examples of methods for deriving the optimal route. Figure 7 is a schematic diagram of a grid-like transportation network. Each node, indicated by a circle, is assigned a matrix number for identification. Thick lines connect nodes, but not all nodes are connected by thick lines. As shown in Figure 4, edges are represented by thick lines only between nodes where a road actually exists and transportation equipment 1 can move. These edges are directed edges. Edge data is represented by the matrix numbers of the starting and ending nodes.

[0074] In a transport network, a transport route can be represented as a column of node identification data (rows and columns). For example, the transport route of transport device 1, which departs from the central point in Figure 7, can be expressed as node (i,j) → node (i,j+1) → node (i-1,j+1) →… The transport route of the agent, node (i,j) → node (i,j+1) → node (i-1,j+1), is shown by the white arrow in Figure 7.

[0075] The agent (transportation device 1) initially located at node (i,j) (indicated as "A" in Figure 7) probabilistically selects an edge leading to the adjacent node, according to the selection probability determined by attractive or repulsive forces.

[0076] In the initial state based on the delivery conditions, the specific node that is the origin is node (i+1, j+1) and there are 100 pallets P. As a result, in Figure 7, the edge leading to this node (i+1, j+1) is assigned the signed value (virtual potential) "+100" in S102. The virtual potential corresponds to the number of pallets P to be shipped.

[0077] In the initial state based on the delivery conditions, the destination nodes are node (i-2,j), node (i-1,j-2), and node (i+2,j-3). Node (i-2,j) is assigned the condition to transport 40 pallets P. In Figure 7, the edge leading to this node (i-2,j) is assigned a virtual potential of, for example, "-40" corresponding to the number of pallets P to be transported in S103. Node (i-1,j-1) is assigned the condition to transport 20 pallets P. In Figure 7, the edge leading to this node (i-1,j-1) is assigned, for example, "-20". Similarly, node (i+2,j-3) is assigned the condition to transport 30 pallets P. In Figure 7, the edge leading to this node (i+2,j-3) is assigned, for example, "-30".

[0078] The transport network shown in Figure 8 is an example of the transport network shown in Figure 7 after the processing unit 40 in S102 has assigned virtual potentials to nodes surrounding a specific node according to their distance from that specific node. In Figure 8, for the node (i+1,j+1) which is assigned "+100" corresponding to the originating point, edges going to adjacent nodes via edges are assigned signed numbers with a gradient in magnitude according to the number of edges passed through or the sum of the distances of the corresponding edges. The nodes (i,j+1) connected to the originating point node (i+1,j+1) via one edge, and the edges going to the node (i+1,j), are each assigned a virtual potential of "+75". In addition, the edges going to the nodes (i,j+2), (i-1,j+1), and (i+2,j), which are connected to the originating point node (i+1,j+1) via two edges, are each assigned a virtual potential of "+50". The virtual potentials assigned to the edges to these surrounding nodes are used solely for calculating attractive and repulsive forces between adjacent nodes. If each node is assigned features (number of palettes P present, virtual potential, ...), then the feature of node (i+1,j+1) assigned "+100" is (100,+100), and the feature of the adjacent nodes (i,j+1) and (i+1,j) is (0,+75). The feature of nodes (i,j+1) and (i+1,j) separated by two edges is (0,+50).

[0079] In the transport network shown in Figure 8, virtual potentials are also assigned to edges leading to surrounding nodes of a specific node corresponding to a destination. In the example in Figure 8, the edges leading to the adjacent nodes (i-3,j), (i-1,j), and (i-2,j-1) of node (i-2,j-1), which are assigned "-40" to node (i-2,j) corresponding to the destination, are assigned a virtual potential of "-20". For the destination as well, the virtual potentials assigned to the edges leading to surrounding nodes are used only for calculating attractive and repulsive forces between adjacent nodes. When feature quantities (number of pallets P, virtual potential, ...) are assigned to each node, the feature quantity of node (i-2,j), which is assigned "-40", is initially (0,-40), and the feature quantities of the adjacent nodes (i-1,j), (i-3,j), and (i-2,j-1) are (0,-20).

[0080] Similarly, the edges leading to the adjacent nodes (i,j-2) and (i-2,j-2) via an edge to the node (i-1,j-2) that has been assigned "-20" corresponding to the arrival point are assigned a virtual potential of "-10". When features (number of existing pallets P, virtual potential, ...) are assigned to each node, the features of the node (i-1,j-2) that has been assigned "-20" are initially (0,-20), and the features of the adjacent nodes (i,j-2) and (i-2,j-2) are (0,-10).

[0081] Node (i+2,j-3), which is assigned "-30" corresponding to the arrival point, has edges leading to its adjacent nodes (i+2,j-2), (i+3,j-3), and (i+1,j-3), each assigned a virtual potential of "-15". When features (number of existing pallets P, virtual potential, ...) are assigned to each node, the features of node (i+2,j-3), which is assigned "-30", are initially (0,-30), and the features of its adjacent nodes (i+2,j-2), (i+3,j-3), and (i+1,j-3) are (0,-15).

[0082] In the examples shown in Figures 7 and 8, the cargo bed of the transport device 1 agent located at node (i,j) is empty, meaning the number of pallets P associated with the agent is zero. In the example in Figure 8, at S104, the initial agent is assigned a virtual charge of "-50". In the example in Figure 7, the virtual charge is assigned as "-50" if the cargo bed is empty, "+50" if it is fully loaded, and "±0" if the ratio of the number of storage compartments containing pallets P to the total number of compartments (e.g., "20") is 50%. The absolute value of the virtual charge of transport device 1 may be based on the load capacity of each transport device 1. For example, if a transport device 1 has "20" compartments and its cargo bed is empty, the charge may be "-20" and if it is fully loaded, the charge may be "+20". If another transport device 1 with a larger load capacity has "40" compartments and its cargo bed is empty, the charge may be "-40" and if it is fully loaded, the charge may be "+40". This can provide clear information on how many more pallets P can be loaded.

[0083] The transport device 1, which is an agent located at node (i,j), calculates the attractive or repulsive force from the edge to the adjacent node (i+1,j) and the edge to node (i,j+1). The virtual charge attached to the agent is "-50", and the virtual potential attached to the edge to node (i+1,j) and the edge to node (i,j+1) is "+75". Therefore, in S110, the processing unit 40 calculates an attractive force of, for example, "125 (= +75 - (-50))" from each edge to the virtual charge. The processing unit 40 probabilistically selects an edge to one of the nodes (S110). Since the attractive forces are the same, the selection probability of each edge is the same.

[0084] Figure 9 shows the state of the transport network one step forward from the state shown in Figure 8. In Figure 9, the agent is moving by selecting an edge between node (i,j) and node (i+1,j). Node (i+1,j) is neither the origin nor the destination.

[0085] In the state shown in Figure 9, the transport device 1, which is an agent located at node (i+1,j), calculates the attractive or repulsive forces from the edges to the adjacent node (i+2,j), the edge to node (i+1,j+1), and the edge to node (i,j). The virtual charge attached to the agent is "-50", the signed numerical values ​​attached to the edges to the adjacent node (i+2,j) and the edge to node (i,j) are "+50", and the signed numerical value attached to the edge to node (i+1,j+1) is "+100". Therefore, in S109, the processing unit 40 calculates an attractive force of "100 (=+50-(-50))" from the edges to the adjacent node (i+2,j) and the edge to node (i,j), and an attractive force of "150 (=+100-(-50))" from the edge to node (i+1,j+1). The processing unit 40 is more likely to select an edge with a stronger gravitational force (S109).

[0086] Figure 10 shows the state of the transport network after one step has been taken from the state shown in Figure 9. Figure 10A shows the state immediately after the step has been taken, and Figure 10B shows the state after processing in S111. The agent has moved from node (i+1,j) to node (i+1,j+1) by selecting an edge leading to node (i+1,j+1). Node (i+1,j+1) is the departure point. At the time of Figure 10A, the number of pallets associated with the agent is "0 (zero)" because the loading platform is empty. At the time of Figure 10A, the number of pallets P at node (i+1,j+1) where the agent is located is "100". In S115, the processing unit 40 decides to stack "20" pallets P so as to fill all of the empty loading platforms. As a result, the number of pallets P associated with the agent after stacking is "20", and the virtual charge assigned is "+50", which corresponds to a full load (S115).

[0087] As shown in Figure 10B, the S115 process reduces the number of pallets P at node (i+1,j+1) from "100" to "80". The reward for node (i+1,j+1) is consumed in proportion to the decrease in the number of pallets P, and the virtual potential applied to the edge leading to node (i+1,j+1) decreases to "+80".

[0088] In the state shown in Figure 10B, the transport device 1, which is an agent located at node (i+1,j+1), calculates the attractive or repulsive force from the edge toward the adjacent node (i+1,j) and the edge toward node (i,j+1), respectively. As the agent's pallet P has increased, the virtual charge has been updated to "+50", and the signed numerical values ​​assigned to the edges toward the adjacent node (i+1,j) and the edge toward node (i,j+1) are both "+60". Therefore, in S109, the processing unit 40 calculates a repulsive force of "110 (=60+50)" for the edge toward the adjacent node (i+1,j) and the edge toward node (i,j+1), respectively. If the repulsive forces are the same, the processing unit 40 probabilistically selects an edge toward one of the nodes with the same probability (S110).

[0089] Figure 11 shows the state of the transport network two steps after the state shown in Figure 10. In Figure 11, the agent is moving along a transport route that can be represented as (i+1,j) → node (i+1,j+1) → node (i,j+1) → node (i-1,j+1) (indicated by the white arrows). Since node (i-1,j+1) is neither a departure point nor a destination point, there is no change in the number of pallets associated with the nodes and agents in S108.

[0090] In the state shown in Figure 11, the transport device 1, which is an agent located at node (i-1, j+1), calculates the attractive or repulsive force from the edge to the adjacent node (i-1, j) and the edge to node (i-1, j+2). The signed numerical value assigned to the agent is "+50", the virtual potential assigned to the edge to the adjacent node (i-1, j) is "-20", and no virtual potential is assigned to the edge to node (i-1, j+2). Therefore, in S109, the processing unit 40 calculates an attractive force of "70 (=-20-(+50))" from the edge to the adjacent node (i-1, j), and does not calculate an attractive or repulsive force from the edge to node (i-1, j+2). The processing unit 40 probabilistically selects an edge (S110).

[0091] Figure 12 shows the state of the transport network two steps further from the state shown in Figure 11. Figure 12A shows the state immediately after the step has been advanced, and Figure 12B shows the state after the update process in S110. In Figure 12, the agent is moving along a transport route that can be expressed as (i+1,j) → node (i+1,j+1) → node (i,j+1) → node (i-1,j+1) → node (i-1,j) → node (i-2,j) from node (i,j), which is the origin point. Node (i-2,j) is the destination point. At the time of Figure 12A, the number of pallets associated with the agent is "20" because the cargo bed is full. At the time of Figure 12A, the number of pallets at node (i-2,j), where the agent is located, is "0". In S113, the processing unit 40 decides to unload all 20 pallets P that are loaded at node (i-2,j) in such a way that it reduces the number of pallets P required by 40 to 20. As a result, the number of pallets P associated with the agent after unloading becomes 0, and the virtual charge becomes -50, which corresponds to an empty state.

[0092] As shown in Figure 12B, the processing in S113 reduces the required number of pallets P at node (i-2,j) from "40" to "20", and the virtual potential of the edge leading to node (i-2,j) decreases to "-20".

[0093] In the state shown in Figure 12B, the transport device 1, which is an agent located at node (i-2,j), calculates the attractive or repulsive forces from the edges to the adjacent nodes (i-1,j), (i-3,j), (i-2,j-1), and (i-2,j+1). The virtual charge attached to the agent has been updated to "-50", and the virtual potentials attached to the edges to the adjacent nodes (i-1,j), (i-3,j), and (i-2,j-1) are all "-10". Therefore, in S109, the processing unit 40 calculates a repulsive force of "60 (=10+50)" from the edges to the adjacent nodes (i-1,j), (i-3,j), and (i-2,j-1). The processing unit 40 randomly selects an edge because all edges exert similar repulsive forces (S110). The processing unit 40 adds virtual potentials to the surrounding nodes, and for the edge leading to node (i-1,j), it is influenced by, for example, "+20" from the discharge point node (i+1,j+1) through nodes (i,j+1) and (i-1,j+1), updating it to "+20+(-10)" to make it easier to select the edge leading to node (i-1,j). The virtual potentials applied to the edges leading to each node are applied continuously, like the actual potentials, and the processing unit 40 may select edges to move closer to or further away from them based on this relationship between virtual charge and virtual potential.

[0094] The process, based on step-by-step agent-based modeling as shown in Figures 8-12, should ideally calculate the optimal path that minimizes the number of pallets P at the origin and destination points as quickly as possible, or maximizes the total reward given to the selected edges as quickly as possible. Reinforcement learning can be performed by treating the revenue as the sum of rewards, thereby enabling reinforcement learning for the selection of each edge. When performing reinforcement learning, an action-value function based on Equation 1 below can be calculated for each edge, and reinforcement learning can be performed based on this to quickly derive the transport path of the transport device 1, which is the agent.

[0095]

number

[0096] In this way, the optimal transportation route information calculated for each region is stored in the transportation route DB 414, and the transportation route determination device 400 determines the overall transportation route connecting the regions and the movement route of the pallet P at the same time, based on the actual transportation request.

[0097] Figures 13 and 14 are flowcharts showing an example of the transportation route determination process by the transportation route determination device 400.

[0098] The processing unit 40 of the transport route determination device 400 obtains the arrangement of pallets P from pallet information 412 and transport equipment DB 413 (S201). Based on the destination and desired arrival time of each pallet P, the processing unit 40 obtains conditions indicating where each pallet P must be transported to after a predetermined time or at a specific point in time (S202). In S202, the processing unit 40 aggregates the number of pallets P for each base 2 or collection and distribution center 3 that should be located, for example, 3 hours after the processing in S201.

[0099] The processing unit 40 determines the transportation conditions in the transportation network based on the current position of pallet P and the position of transport equipment 1 obtained in the processing of S201, and the desired positions of pallet P and transport equipment 1 after a predetermined time obtained in the processing of S202 (S203).

[0100] In processing S203, the processing unit 40 determines the transportation conditions for each time period, specifying where pallet P is located in the transportation network and where it should have reached. Figure 15 shows the relationship between the overall route of pallet P and the divided transportation networks. The bases 2 are connected in a network-like manner. As shown in Figure 15, each pallet P travels through multiple divided transportation networks that span from the departure base 2 to the final arrival point, the collection and distribution center 3. In the example in Figure 15, one pallet P departs from base 2 in Nagano Prefecture and its final destination is the collection and distribution center 3 in Shiga Prefecture. By determining where pallet P should be transported from each point in time and superimposing the data for multiple pallets P, the transportation conditions can be determined for each divided transportation network, specifying how many pallets P should gather at base 2 and how many should be transported to other bases 2 or collection and distribution centers 3.

[0101] The processing unit 40 selects one transport network (S204), extracts the optimal route for delivery conditions similar to the transport conditions for that transport network from the data stored in the storage unit 41, and stores it (S205). In S205, the processing unit 40 performs a route calculation based on agent-based modeling (Figures 5 and 6) for the optimal route for similar delivery conditions, calculating only the difference between the delivery conditions and the actual transport conditions, and derives the optimal route for each condition.

[0102] The processing unit 40 determines whether all transport networks have been selected (S206), and if it determines that not all networks have been selected (S206: NO), it returns to processing S204.

[0103] If it is determined that all transportation networks have been selected (S206:YES), the processing unit 40 obtains and stores the number of transportation equipment 1 required for each transportation network, the optimal transportation route for each of the required number of transportation equipment, and the delivery conditions (type and number of pallets P to be transported, and the number of pallet storage units) (S207).

[0104] The processing unit 40 assigns and stores pallet identification information to the pallet storage compartments on the loading platforms of the transport equipment 1 at each location, based on information about pallets P already loaded and the information obtained in the processing of S207 (S208).

[0105] The processing unit 40 determines the overall transportation route by selecting a partial transportation route within the divided transportation network for each transportation device 1 (S209). In S209, the processing unit 40 may determine the route within the transportation network as the overall transportation route for the transportation device 1.

[0106] The processing unit 40 determines (updates) the movement path for each pallet P, that is, the pallet storage section of the transport equipment 1 that should be transferred to reach the destination (S210).

[0107] Based on the movement path of each pallet P, the processing unit 40 lists, for each transport equipment 1, the pallet identification information of the pallet P to be shipped out at the intermediate base 2, and the storage unit identification information of the storage unit in which the pallet P is stored (S211).

[0108] Based on the movement path of each pallet P, the processing unit 40 lists, for each transport equipment 1, the pallet identification information of the pallet P to be brought in at the transit point, base 2, and the storage unit identification information of the pallet storage unit where that pallet P should be stored (S212).

[0109] The processing unit 40 outputs the determined transport route for the transport equipment 1, the movement route for the pallet P, and a list of pallets P to be loaded and unloaded at each base 2 (S213). Based on the output, the processing unit 40 sends instructions to the on-board unit 10, the base controller 20, and the collection and delivery controller 30 (S214), and then terminates the process.

[0110] In this way, by referring to the calculated optimal route within the divided transport network, the transport route of transport equipment 1 and the movement route of pallet P can be determined, and the movement of the entire pallet P can be controlled based on this.

[0111] Furthermore, in the S205 process, it is preferable not to simply select the optimal route derived in advance, but to add calculations regarding the difference conditions based on an optimal route with similar conditions to the already derived route. The processing unit 40 may calculate and integrate the number of transport equipment and transport route corresponding only to the additional collection and delivery locations, destinations, and number of pallets, in addition to the information on the number of transport equipment and transport route for which calculation results have already been obtained. By accumulating the route calculation results in advance, the calculation time required is limited to the time to access the data, thus reducing the calculation time.

[0112] In the embodiment described above, a grid-like transport network defined in Figure 4 was used, and agent-based modeling simulations were performed using the transport network. However, the path calculation process may also be represented by a graph and optimized using a GNN.

[0113] (Variation 1) In the above embodiment, regardless of the type of goods transported by the transport equipment 1, the pallets P are assigned virtual potentials and virtual charges based on the number of pallets P that are to be shipped without classification. In Modification 1, the pallets P can be classified according to the type of goods loaded on them, their destination, etc.

[0114] Figure 16 shows an example of delivery conditions in Modification 1. In Figure 16, the initial positions and numbers of pallets P in a transportation network corresponding to a certain region are shown in a distinguishable manner for each type. In Modification 1, as shown in Figure 16, for example, the initial state is set such that there are 50 pallets P mainly loaded with item A (indicated by the symbol PA in Figure 16) at a specific node (indicated by D1), and 50 pallets P mainly loaded with item B (indicated by the symbol PB in Figure 16) at a specific node (indicated by D2). The delivery conditions are, for example, to transport 20, 18, and 12 pallets P mainly loaded with item A to three nodes in the transportation network (arrival points T1, T2, and T3 within the region), respectively. The delivery conditions are further to transport 30 and 20 pallets P mainly loaded with item B to two nodes in the transportation network (arrival points T4 and T3 within the region), respectively. In Modification 1, when the origin and destination differ depending on the type of goods stacked on pallet P, virtual potentials and virtual charges are assigned to each type as follows, and the superposition, attractive and repulsive forces are evaluated for each type to select an edge and perform an agent-based simulation.

[0115] Figures 17 and 18 are flowcharts illustrating an example of the process for determining transportation routes for each region in Modification Example 1. For the processing steps shown in Figures 17 and 18, steps that are common to those shown in Figures 5 and 6 are given the same step numbers, and detailed explanations are omitted.

[0116] In the modified example, the processing unit 40 sets delivery conditions for the target transport network, including the initial value and number of transport equipment 1 which is an agent, the initial position for each type of pallet P, and the destination (arrival point within the region) of each pallet P within the network (S301).

[0117] Based on the delivery conditions, the processing unit 40 assigns virtual potentials (signed numerical values) to specific nodes corresponding to the originating point, the surrounding nodes, and the edges leading to each of them, for each type of pallet P among the nodes in the transport network (S302). The assigned virtual potentials are assigned for each type by the same process as the process in S102 shown in Figures 5 and 6.

[0118] Based on the delivery conditions, the processing unit 40 assigns a virtual potential with the opposite sign to the numerical value assigned in S302 to each node in the transport network corresponding to the arrival point within the region of pallet P, the surrounding nodes, and the edges leading to each of them, according to the type of pallet P (S303). The virtual potential is assigned for each type by the same process as the process in S103 shown in Figures 5 and 6.

[0119] Based on the delivery conditions, the processing unit 40 assigns a virtual charge to the transport equipment 1 (agent) in its initial state, corresponding to the available space on the cargo bed, for each type of pallet P that the transport equipment 1 can deliver (S304). In S304, for example, if the cargo bed of the transport equipment 1 is empty and can accommodate 20 pallets P, the processing unit 40 assigns, for example, the number of pallets P that can be accommodated, "20," to each type. If the cargo bed of the same transport equipment 1 is loaded with "5" pallets P mainly containing goods A, the processing unit 40 assigns a virtual charge of "-10" to pallets P of type A, which is the sum of the number of pallets P already accommodated, "+5," and the number of pallets P that can be accommodated further, "-15." In this case, the processing unit 40 assigns a virtual charge of "-15" to pallets P of type B, which is the number of pallets P that can be accommodated, to the transport equipment 1.

[0120] After the processing unit 40 sets the execution count and step count to initial values ​​(S105, S106), the processing unit 40 initially assigns rewards to each node in the transport network according to the delivery conditions and the type of pallet P (S307). If there are no pallets P loaded on the agent's cargo bed, the node corresponding to the originating point is given a reward that increases with the number of pallets P present at the node. Conversely, if the agent's cargo bed is set to be nearly full, the node corresponding to the destination is given a reward that increases with the number of missing pallets P.

[0121] The processing unit 40 updates the number of each type of pallet P associated with the agent, the number of each type of pallet P remaining at the node, the signed numerical value (virtual charge) of each type of pallet P assigned to the agent, and the signed numerical value (virtual potential) of each type of pallet P assigned to the node where the agent resides (S308). The details of the processing unit 40 in S308 are the same as the processing in S108 shown in Figures 5 and 6.

[0122] The processing unit 40 calculates attractive or repulsive forces for each type of pallet P based on the signed numerical values ​​(virtual potentials) assigned to the surrounding nodes of the agent and the signed numerical values ​​(virtual charges) assigned to the target agent (S309). The processing unit 40 superimposes the attractive or repulsive forces calculated for each type of pallet P (S310), determines the selection probability from the superimposed result (S311), and probabilistically selects the edge to the next adjacent node to proceed to based on the selection probability determined in S313 (S312). The processing unit 40 draws the edge using the selection probability as the weight in a weighted draw. The higher the selection probability, the higher the probability that the edge will be selected, but there is still a possibility that an edge with a low selection probability may be selected.

[0123] The processing unit 40 moves the agent via the selected edge (S313). The processing unit 40 determines whether the node to which the agent has moved via the selected edge corresponds to the arrival point of the type of pallet P loaded on the agent (S314).

[0124] If it is determined in S314 that this is a destination (S314: YES), the processing unit 40 consumes (unloads) the pallets P of the type corresponding to the destination from among the pallets P loaded on the cargo bed of the transport equipment 1, which is the agent, and consumes the reward corresponding to the type assigned to the node corresponding to the destination (S315). In S315, the processing unit 40 subtracts the number of pallets P of the corresponding type associated with the agent. It also subtracts the number of pallets P that are missing at the destination.

[0125] If it is determined in S314 that the destination is not the destination (S314: NO), the processing unit 40 skips S315 and proceeds to S316.

[0126] In S316, the processing unit 40 determines whether the node to which the agent has moved via the selected edge corresponds to the loading point and whether there is sufficient space on the agent's loading platform (S316). The processing unit 40 may also determine this by whether the absolute value of the charge corresponding to the number of pallets P on the agent's loading platform is small or not.

[0127] In S316, if it is determined that the destination is a shipping point and that there is sufficient space on the agent's cargo bed (S316: YES), the processing unit 40 increases the number of pallets P loaded on the cargo bed of the transport device 1, which is the agent, and consumes the rewards of the types of pallets P assigned to the node corresponding to the destination shipping point (S317). In S317, the processing unit 40 adds up the number of pallets P associated with the agent by type. It then subtracts the number of pallets P of the type to be transported from the shipping point.

[0128] If it is determined in S316 that it is not a shipping point (S316: NO), the processing unit 40 skips S317 and proceeds to S116. The processing unit 40 adds the number of steps and stores it (S116). In S116, if distance data is associated with the moved edge, the processing unit 40 adds that distance to the cumulative distance traveled.

[0129] Next, the processing unit 40 determines in S315 and S317 whether or not a delivery that meets the delivery conditions within the transport network has been completed (S117). In S117, the processing unit 40 may determine that the delivery is complete when the virtual potentials applied to the edges around each node all become zero, or it may determine that the delivery is complete when the correct number of pallets P that meet the delivery conditions have been unloaded at the destination node.

[0130] If it is determined that delivery has not been completed (S117:NO), the process is returned to S308.

[0131] If it is determined that delivery is complete (S117:YES), the processing unit 40 stores the history of identification data of the nodes the agent passed through (transport route) and the number of steps (S118).

[0132] The processing unit 40 determines whether the processes in S308-S118 have been executed a predetermined number of times or more (S119). If it is determined that the number of executions is less than the predetermined number (S119: NO), the processing unit 40 adds to the execution count (S120) and returns the process to S106.

[0133] If it is determined in S119 that the process has been executed more than the predetermined number of times (S119: YES), the processing unit 40 extracts the transport route with the minimum number of steps or the shortest cumulative travel distance within the predetermined number of times (S121). The processing unit 40 stores the extracted transport route as the optimal solution under the delivery conditions (S122) and terminates the process.

[0134] According to the processing procedures shown in Figures 17 and 18, virtual potentials and virtual charges are assigned according to the type of pallet P, and the attractive or repulsive force calculated for each type is superimposed to perform the processing.

[0135] Figures 19-21 show examples of the method for deriving the optimal route in Modification 1. Figure 19 corresponds to Figure 16. Figure 19 shows the state after the processing unit 40 has assigned virtual potentials to the transport network and delivery conditions shown in Figure 16. In Figure 16, an agent (indicated as "A" in Figures 19 and 20) is located at node (i,j). Figure 16 shows the state in which virtual potentials have been assigned to each edge according to the number of pallets P on which goods A and goods B are stacked. For example, similar to the example shown in Figure 8, the edge directly connected to a specific node (i+1,j+1) indicated by D1 is assigned a virtual potential of "+50" according to the number of pallets P of goods A to be shipped from node D1. The edges heading to node (i+1,j) and node (i,j+1), which are one node away from the specific node D1, are assigned a virtual potential of "+30" according to the number of pallets P of goods A to be shipped from node D1. The virtual potential attenuates according to the number of edges from the specific node (i+1,j+1) indicated by D1. Therefore, the edges leading to the nodes (i+2,j), (i,j), (i,j+2), and (i-1,j+1), which are two nodes away from the specific node D1, are assigned a virtual potential of "+15".

[0136] Similarly, for example, an edge directly connected to a specific node (i-2, j-2) indicated by D2 is assigned a virtual potential of "+50" depending on the number of pallets P of item B to be shipped from D2. Edges leading to nodes (i-1, j-2), (i-2, j-3), and (i-2, j-1), which are one node away from the specific node D2, are assigned a virtual potential of "+30" depending on the number of pallets P of item B to be shipped from D2. Of these, node (i-1, j-2) is the destination T2 for item A. An edge directly connected to destination T2 is assigned a negative virtual potential of "-18". Therefore, among the edges leading to the destination point T2, the edges that indirectly lead to a specific node D2 (edges from node (i,j-2) to node (i-1,j-2)) have a virtual potential of "+30" related to item B and a virtual potential of "-18" corresponding to the pallet P of the 18 items A that should be collected at the destination point T2 superimposed on them. Similarly, among the edges that directly connect to the specific node (i-2,j-2) indicated by D2, the edges that indirectly lead to the destination point T2 (node ​​(i-1,j-2)) (edges from node (i-2,j-1) to node (i-2,j-2)) have a virtual potential of "+30" related to item B and a virtual potential of "-9" related to item A superimposed on them.

[0137] The processing unit 40 selects an edge from among the edges where virtual potentials are distributed as shown in Figure 19, and proceeds with the simulation by having the transport device 1 (agent) calculate the attractive or repulsive force.

[0138] Figure 20 shows the state after six steps have passed since Figure 19. The agent, initially located at node (i,j), is assigned a virtual charge of "-50", a negative value, because the loading platform is empty. In the state shown in Figure 20, the agent has subsequently selected nodes (i,j) → (i+1,j) → (i+1,j+1) → (i,j+1) → (i-1,j+1) → (i-1,j) → (i-2,j), as indicated by the white arrows. The agent loads pallet P of item A at node (i+1,j+1), the loading point for pallet P. Therefore, in Figure 20, the virtual potential values ​​of the edges around node (i+1,j+1) have been updated. Before proceeding to node (i-2,j), the agent loaded pallet P to full capacity at node (i+1,j+1), thus being assigned a virtual charge of "+50," which makes it easier to select edges that have been assigned a negative virtual potential.

[0139] In the state shown in Figure 20, the node (i-2, j) where the agent is located is the loading point for pallet P of item A (S314: YES). Therefore, the transport equipment 1, which is the agent, will unload pallet P of item A that it is carrying.

[0140] Figure 21 shows the state after the virtual potentials assigned to each node and edge, and the virtual charge of the agent, have been updated, following the loading and unloading of pallet P of item A from transport equipment 1, as shown in Figure 20. In the state shown in Figure 21, since pallet P was unloaded at node (i-2,j) where the agent is located, the virtual potential of the edge leading to node (i-2,j) is zero. Also, the virtual charge assigned to the agent has been updated to "-50" again because it is now empty. As a result, the agent is more likely to select an edge with a positive sign, and will next select node (i-2,j-1).

[0141] In this way, it becomes possible to calculate the optimal travel route for transport equipment 1 that carries pallets P of different types of goods, each with a different origin and destination.

[0142] (Modification 2) In the above embodiment, the probability of selecting an edge was determined based on the attractive or repulsive force between the virtual potential of each edge and the virtual charge attached to the transport device 1, using the virtual potential attached to the edge and the virtual charge attached to each transport device 1. In Modification 2, the processing unit 40 calculates the attractive or repulsive force with the virtual charge attached to other transport devices 1 for each type of pallet P and determines the selection probability. Figure 22 is an explanatory diagram of the relationship with the virtual charge of other transport devices 1. Figure 22, like Figure 7, shows a schematic diagram of the transport network. In Figure 22, the agent is located at node (i,j) indicated by "A" in the figure. In Figure 22, unlike Figure 7, another transport device 1, indicated by "V" in the figure, is located at node (i+2,j). A virtual charge is attached to this transport device 1 indicated by "V" in S104.

[0143] In the modified example 2, when the processing unit 40 executes the processes shown in Figures 5 and 6, or Figures 17 and 18, in the processes of S109 and S309, it calculates not only the attractive or repulsive force of the virtual potential of the edge to the surrounding nodes to the virtual charge of the agent located at node (i,j), but also the attractive or repulsive force between the virtual charges with the virtual charges of other transport devices 1 located at the surrounding nodes. In the example of Figure 22, the signed numerical value (virtual potential) of the agent located at node (i,j) is "-50". In contrast, there is a virtual charge of "-50" at node (i+2,j). Therefore, since the virtual charges have the same sign, a repulsive force is generated, and the selection probability of the edge to node (i+1,j) for the agent located at node (i,j) is determined to be small.

[0144] This makes it possible to avoid a situation where one transport device (agent) 1 follows another transport device (agent) 1 and concentrates on heading towards a single node.

[0145] In the above-described embodiment, an example of application to a logistics system 100 was explained, but it can be applied not only to logistics using transport equipment 1 that travels on roads, but also to unmanned transport robots in factories, transport systems in logistics warehouses, and so on.

[0146] The embodiments disclosed above are illustrative in all respects and not restrictive. The scope of the present invention is defined by the claims, and all modifications within the meaning and scope equivalent to the claims are included. [Explanation of symbols]

[0147] 100 Logistics Systems 2 locations 3. Collection and Delivery Center 400 Transportation Route Determination Device 40 Processing Unit 41 Storage section P4 Program

Claims

1. In a method for determining a transport route in which a computer performs a simulation using agent-based modeling, assigning a point where goods are dispatched, a point that transport equipment transporting the goods can pass through, and a point where the goods arrive to nodes, and assigning routes connecting adjacent points to the edges between the nodes, the transport route is defined and the transport equipment is an agent moving between the nodes, The aforementioned computer, A virtual potential, including positive or negative signs and magnitude, is assigned to the surrounding nodes or edges, centered on the node, according to the quantity of goods to be shipped from the point corresponding to each node. Depending on the quantity of goods to be transported to the point corresponding to each node, a virtual potential including a sign and magnitude opposite to the aforementioned sign is assigned to the surrounding nodes or edges centered on the node. A virtual charge, including positive and negative signs and magnitudes, is assigned to the aforementioned transport equipment according to the empty state or loading rate of the cargo bed. In order for the transport device to which the virtual charge is attached to move from one node to another adjacent node, the probability of selecting an edge to the other node is determined based on the attractive or repulsive force of the transport device to the virtual charge with respect to the virtual potential attached to each edge. Select an edge based on probability. The simulation is performed by repeating the processes of assigning the virtual potential and virtual charge, determining the probability, and selecting the edge until the delivery of the item is completed. From the movement paths between points obtained from the simulation, the movement path of the transport equipment is extracted based on a predetermined evaluation quantity. A method for determining the transport route to execute a process.

2. The computer extracts the transportation route of the transportation equipment where the predetermined evaluation quantity is the cumulative distance traveled by the transportation equipment and the cumulative distance traveled is the shortest, or where the predetermined evaluation quantity is the number of times the process is repeated until the delivery is completed and the number of times the process is the fewest. The method for determining a transport route according to claim 1.

3. The computer assigns to the surrounding nodes or edges of each node a positive number with an absolute value that increases as the quantity of goods to be shipped from the point corresponding to the node increases, as the virtual potential. The method for determining a transport route according to claim 1.

4. The computer assigns to the surrounding nodes or edges of each node a negative numerical value as a virtual potential, the larger the absolute value of the goods to be transported to the point corresponding to that node. The method for determining a transport route according to claim 2.

5. The computer assigns the virtual potential to the edges around a specific node that corresponds to the point of origin of the goods within the transport network, or the point of arrival of the goods within the transport network, in proportion to the distance from the point corresponding to the specific node. The method for determining a transport route according to claim 1.

6. The aforementioned computer, For the aforementioned transport equipment, if the cargo bed is empty, a negative number with the largest absolute value is assigned. When the cargo bed is fully loaded, assign the largest possible positive number. As the load factor increases, a value is assigned ranging from the negative number with the largest absolute value to the positive number with the largest absolute value. The method for determining a transport route according to claim 4.

7. The aforementioned computer, If the sign of the numerical value assigned to the transport device is opposite to the sign of the numerical value assigned to the edge connected to the node corresponding to the location where the transport device is located, it is determined that an attractive force is acting on the transport device from the edge. If the sign of the numerical value assigned to the transport device is the same as the sign of the numerical value assigned to the edge connected to the node corresponding to the location where the transport device is located, it is determined that a repulsive force acts on the transport device from the edge. Select an edge to an adjacent node according to the strength of the attractive or repulsive force. A method for determining a transport route according to any one of claims 3 to 6.

8. The aforementioned computer, If the goods to be transported by the aforementioned transport equipment are classified into multiple types, a virtual potential is assigned to each edge according to the type. Depending on the type of goods being transported by the transport equipment, a virtual charge is assigned to each of the transport equipment according to the type. The attractive or repulsive force of the transport device to the virtual charge with respect to the virtual potential applied to each edge is calculated and superimposed for each of the multiple types of the article. When determining the probability of selecting an edge to the aforementioned other node, the probability is determined based on superimposed attractive or repulsive forces. The method for determining a transport route according to claim 1.

9. The computer determines, in the simulation, the selection probability for the transport device to move from one node to another, based on the attractive or repulsive force of the transport device to the virtual charge with respect to the virtual potential assigned to each edge, and the attractive or repulsive force of the transport device to the virtual charge with respect to the virtual charge assigned to other transport devices located at each node. The method for determining a transport route according to claim 1.

10. In a transport route determination device that defines a transport network by assigning the departure point of an item, the points through which the transport equipment transporting the item passes, and the arrival point of the item to nodes, and assigning the routes connecting adjacent points to the edges between the nodes, and performs a simulation using agent-based modeling in which the transport equipment is an agent moving between the nodes, A virtual potential, including positive or negative signs and magnitude, is assigned to the surrounding edge centered on each node, according to the quantity of goods to be shipped from the point corresponding to each node. Depending on the quantity of goods to be transported to the point corresponding to each node, a virtual potential including a sign and magnitude opposite to the aforementioned sign is applied to the surrounding edge centered on the node. A virtual charge, including positive and negative signs and magnitudes, is assigned to the aforementioned transport equipment according to the empty state or loading rate of the cargo bed. In order for the transport device to which the virtual charge has been assigned to move from one node to another adjacent node, the probability of selecting an edge to the other node is determined based on the attractive or repulsive force of the transport device to the virtual charge with respect to the virtual potential assigned to each edge, and an edge is selected based on the probability. The simulation is performed by repeating the processes of assigning the virtual potential and virtual charge, determining the probability, and selecting the edge until the delivery of the item is completed. From the movement paths between points obtained from the simulation, the movement path of the transport equipment is extracted based on a predetermined evaluation quantity. A transport route determination device equipped with a processing unit that performs processing.

11. In a computer program that defines a transportation network by assigning the origin of an item, the points through which the transportation equipment transporting the item passes, and the destination of the item to nodes, and assigning the paths connecting adjacent points to the edges between the nodes, and then runs a simulation using agent-based modeling in which the transportation equipment is an agent moving between the nodes, To the aforementioned computer, A virtual potential, including positive or negative signs and magnitude, is assigned to the surrounding edge centered on each node, according to the quantity of goods to be shipped from the point corresponding to each node. Depending on the quantity of goods to be transported to the point corresponding to each node, a virtual potential including a sign and magnitude opposite to the aforementioned sign is applied to the surrounding edge centered on the node. A virtual charge, including positive and negative signs and magnitudes, is assigned to the aforementioned transport equipment according to the empty state or loading rate of the cargo bed. In order for the transport device to which the virtual charge is attached to move from one node to another adjacent node, the probability of selecting an edge to the other node is determined based on the attractive or repulsive force of the transport device to the virtual charge with respect to the virtual potential attached to each edge. Select an edge based on probability. The simulation is performed by repeating the processes of assigning the virtual potential and virtual charge, determining the probability, and selecting the edge until the delivery of the item is completed. From the movement paths between points obtained from the simulation, the movement path of the transport equipment is extracted based on a predetermined evaluation quantity. A computer program that executes a process.

Images