Traveling vehicle system

Abstract

A transport vehicle system is provided with a plurality of transport vehicles that travel along a transport path, and a transport vehicle controller that controls the travel of the transport vehicles. The transport vehicle controller includes: a path determination unit that determines a travel path of the transport vehicle on the basis of a connection path cost for each of the plurality of connection paths; a selection unit that selects, from among the plurality of connection paths, a target connection path for which the future utilization degree is predicted to exceed a predetermined reference; and a cost changing unit that calculates, for each of a plurality of upstream connection paths that exist within a prescribed upstream range with an upstream convergence section closest to the target connection path as a starting point, an index value indicating the degree of concentration ease of the transport vehicle, preferentially selects the upstream connection path having a high index value as one or more bottleneck connection paths, and changes the cost of the bottleneck connection paths. And the connection path cost of the one or more bottleneck connection paths is increased.

Description

Technical Field

[0001] This disclosure relates to vehicle driving systems. Background Technology

[0002] Previously, a vehicle system was known for controlling the movement of vehicles transporting items such as FOUPs (Front Opening Unified Pods) containing semiconductor substrates, for example in semiconductor manufacturing plants (see Patent Document 1). Patent Document 1 describes a method for setting the vehicle's travel path based on the connection path costs set for each connecting path on a predetermined drivable path.

[0003] Patent Document 1: Japanese Patent No. 7059999

[0004] In the methods described above, there is a situation where, by increasing the cost of the connection path predicted to have high utilization (hereinafter referred to as the "object connection path"), it is easier to choose a path that bypasses the object connection path, thereby suppressing traffic congestion on the object connection path. However, simply increasing the cost of the object connection path as described above is sometimes insufficient to adequately suppress traffic congestion in the vehicle system. Summary of the Invention

[0005] Therefore, the purpose of this disclosure is to provide a vehicle system that can effectively suppress traffic congestion.

[0006] This disclosure includes vehicle systems of [1] to [5].

[0007] [1] A vehicle system comprising: a plurality of vehicles traveling along a pre-defined travel route; and a controller that controls the travel of the vehicles by assigning travel instructions indicating a destination on the travel route to vehicles selected from the plurality of vehicles. In the vehicle system, the travel route includes: a plurality of nodes including branches and merging points, and a plurality of connecting paths connecting the nodes. The controller includes: a storage unit that stores the layout of the travel route and the connecting path costs associated with the plurality of connecting paths; and a path determination unit that, based on the connecting path costs of the plurality of connecting paths, selects from the plurality of vehicles to execute the travel instructions. The candidate path in the selected path has the smaller sum of the connecting path costs of the multiple connecting paths included in the path selection and is preferentially determined as the driving path of the vehicle; the selection unit selects the target connecting path from the multiple connecting paths that is expected to have future utilization exceeding a predetermined benchmark; and the cost adjustment unit obtains an index value representing the concentration ease of the vehicle for each of the multiple upstream connecting paths that exist within a predetermined upstream range with the upstream confluence closest to the target connecting path among the multiple connecting paths as the starting point, preferentially selects the upstream connecting path with the higher index value as one or more bottleneck connecting paths, and increases the connecting path cost of the one or more bottleneck connecting paths.

[0008] According to the vehicle system described above [1], it not only increases the connection path cost of the target connection path, but also increases the connection path cost of bottleneck connection paths where vehicles tend to converge in upstream connection paths within a defined upstream range, starting from the upstream merging point closest to the target connection path. This prevents traffic congestion in bottleneck connection paths from occurring in advance. In other words, it can suppress the possibility of a significant increase in the number of vehicles using bottleneck connection paths to reach the target connection path in the future (i.e., the occurrence of traffic congestion in bottleneck connection paths). As a result, it can effectively suppress traffic congestion in the vehicle system.

[0009] [2] In the vehicle system of [1], the aforementioned cost change unit calculates the aforementioned index values ​​for each of the aforementioned multiple upstream connection paths based on graph theory.

[0010] Based on the structure described above [2], the bottleneck connection path in the cost change department can be efficiently selected through relatively simple calculation processing.

[0011] [3] In the vehicle system of [1] or [2], the selection unit repeatedly performs the process of selecting the connection path of the object at a predetermined time interval, and the cost change unit performs the process of selecting one or more bottleneck connection paths and increasing the connection path cost of the one or more bottleneck connection paths whenever the selection unit selects the connection path of the object.

[0012] Based on the structure described above [3], the combination of object connection paths and bottleneck connection paths can be updated appropriately at specified time intervals. Therefore, traffic congestion in the vehicle system can be appropriately suppressed according to changes in the overall system status.

[0013] [4] In any of the vehicle systems in [1] to [3], for the selection unit, for each of the plurality of connection paths, a maximum allowable number representing the maximum number of vehicles that can exist simultaneously in the connection path is obtained, and the connection path in which the number of vehicles existing in the connection path is a predetermined proportion or more of the maximum allowable number of the connection path is selected as the target connection path.

[0014] Based on the structure described above [4], object connection paths can be selected relatively easily based on the index of the maximum allowed number of each connection path.

[0015] [5] In any of the driving vehicle systems of [1] to [4], the driving instruction is associated with a destination contained in any of the plurality of connecting paths, the controller controls the driving of the vehicle in such a way that the vehicle assigned the driving instruction travels to the destination, and the selection unit calculates the number of driving instructions associated with each of the plurality of connecting paths that are predicted to be generated in the future, based on the number of past driving instructions that set the location contained in the connecting path as the destination, and selects the target connecting path based on the number of driving instructions calculated for each of the plurality of connecting paths.

[0016] Based on the structure described above [5], it is possible to select object connection paths with high probability of future driving vehicle concentration based on the number of future driving instructions generated for each connection path, which is predicted based on the number of past driving instructions associated with each connection path.

[0017] According to this disclosure, a vehicle system capable of effectively suppressing traffic congestion can be provided. Attached Figure Description

[0018] Figure 1 This is a diagram showing an example of the layout of a transport vehicle system.

[0019] Figure 2 This is a block diagram representing the functional structure of the transport vehicle system.

[0020] Figure 3 This is a block diagram illustrating an example of the hardware configuration of a transport vehicle controller.

[0021] Figure 4 (a) and Figure 4 (b) is a schematic diagram illustrating an example of the processing of the selection section.

[0022] Figure 5 (a) and Figure 5 (b) is a schematic diagram illustrating an example of the processing of the cost change department.

[0023] Figure 6 This is a diagram representing an example of a delivery log.

[0024] Figure 7 This is a diagram showing an example of summary data from the delivery logs for each period.

[0025] Figure 8 This is a diagram representing an example of the learning data and validation data used in the learning of a predictive model.

[0026] Figure 9 This is a diagram that schematically illustrates an example of predictive processing performed by a predictive model.

[0027] Figure 10 This is a flowchart illustrating an example of the processing of a transport vehicle system. Detailed Implementation

[0028] Hereinafter, an embodiment of the present disclosure will be described with reference to the accompanying drawings. Furthermore, in the description of the drawings, the same or equivalent elements are labeled with the same reference numerals, and sometimes repeated descriptions are omitted.

[0029] like Figure 1 As shown, the transport vehicle system 1 (travel vehicle system) according to this embodiment includes: a transport road 4 (a predetermined travel road) and multiple transport vehicles 2 (travel vehicles) traveling along the transport road 4. The transport road 4 is, for example, a guide rail (track) laid in a factory. The transport vehicles 2 are, for example, unmanned transport vehicles that transport items. The transport vehicles 2 are, for example, elevated transport vehicles, railcars, etc. As an example, the transport vehicle 2 is an elevated transport vehicle that can travel along the transport road 4. For example, the transport vehicle 2 is an overhead hoist transport vehicle (OHT). As an example, the items transported by the transport vehicle 2 are boxes containing multiple semiconductor wafers (so-called FOUP (Front Opening Unified Pod)).

[0030] Conveyor path 4 is divided into multiple ( Figure 1 In this example, there are three blocks (partitions) B. The transport path 4 includes routes within block B (i.e., intra-block routes 5) and routes connecting different blocks B (i.e., inter-block routes 6). Along the transport path 4 are processing units 7 and storage vaults (not shown). Processing unit 7 is the device that performs processing on the semiconductor wafers. The storage vault is a location where the transport vehicle 2 can temporarily store items, serving as a buffer zone.

[0031] The transport path 4 includes multiple nodes 10 and multiple connecting paths 11 connecting adjacent nodes 10. The multiple nodes 10 represent specific locations such as branch points 8 and confluence points 9. In this embodiment, each connecting path 11 is a unidirectional connecting path. That is, the transport vehicle 2 can travel in a predetermined direction for each connecting path 11 (…). Figure 1 (The direction of the arrow shown) travels. Branch 8 is a node 10 connected to an upstream connection path (the upstream connection path 11 of branch 8) and multiple downstream connection paths (the downstream connection path 11 of branch 8). Merging point 9 is a node 10 connected to multiple upstream connection paths (the upstream connection path 11 of merging point 9) and a downstream connection path (the downstream connection path 11 of merging point 9).

[0032] The processing unit 7 is provided with an inlet port for loading items (i.e., the location for unloading items from the conveyor 2) and an outlet port for unloading items (i.e., the location for the conveyor 2 to pick up (load) items). The inlet port and the outlet port are located below the conveyor path 4. The inlet port can also serve as the outlet port.

[0033] like Figure 2 As shown, the conveyor system 1 includes multiple conveyor vehicles 2 and a conveyor vehicle controller 20 (controller). The conveyor vehicle controller 20 receives conveying commands (driving commands) from the host controller 12.

[0034] The transport vehicle controller 20 controls the movement of the transport vehicle 2 in a manner that causes the transport vehicle 2, which has been assigned a transport instruction, to proceed to its destination. Specifically, the transport vehicle controller 20 receives a transport instruction from the host controller 12 instructing the transport vehicle 2 to proceed to its destination (designated location) on the transport route 4. The transport vehicle controller 20 controls the movement of the transport vehicle 2 by assigning the transport instruction received from the host controller 12 to a transport vehicle 2 selected from a plurality of transport vehicles 2.

[0035] The transport instruction is associated with a destination contained in any one of the multiple connection paths 11. Examples of destinations include the aforementioned processing device 7 (inbound port, outbound port), storage facility, etc. In this embodiment, the transport instruction is associated with a "From location" indicating the location where the transported item is picked up (i.e., the transport source of the transported item) and a "To location" indicating the location where the transported item is unloaded (i.e., the transport destination of the transported item). That is, a transport instruction is associated with a From location as a first destination and a To location as a second destination. The transport vehicle 2, which is assigned such a transport instruction, first travels toward the From location as the first destination. Then, after picking up the transported item at the From location, the transport vehicle 2 transports the item toward the To location as the second destination and unloads the item at the To location.

[0036] The upper-level controller 12 includes, for example, a Manufacturing Execution System (MES) and a Material Control System (MCS). The MES is managed by the manufacturer, etc. The MES can communicate with the processing unit 7. For example, the processing unit 7 sends a transport request (grab request, unload request) for processed items to the MES. The MES sends the transport request received from the processing unit 7 to the MCS. If the MCS receives a transport request from the MES, it converts the transport request into a transport instruction as described above and sends the transport instruction to the transport vehicle controller 20. Thus, the transport instruction is assigned to a specific transport vehicle 2 via the transport vehicle controller 20.

[0037] The transport vehicle controller 20 determines the transport vehicle 2 to be the destination of the transport instruction based on a pre-determined selection criterion. For example, the transport vehicle controller 20 determines the transport vehicle 2 (empty transport vehicle) that is located closest to the From location and has not been assigned any other transport instructions as the destination of the transport instruction. Furthermore, the transport vehicle controller 20 executes a pre-determined path search algorithm (e.g., the Dijkstra algorithm or other shortest path retrieval algorithm) based on the connection path costs associated with each connection path 11, thereby determining the transport path for executing the transport instruction (i.e., the travel path from the current position of the transport vehicle 2 through the From location to the To location), and notifies the transport vehicle 2 of this transport path. Thus, the transport vehicle 2 travels on the transport road 4 based on this transport path.

[0038] The conveyor controller 20 can also be configured to perform the above-described processing distributedly by multiple sub-controllers. For example, the conveyor path 4 can be divided into multiple regions (modules), with one sub-controller for each region. In this case, the conveyor path within each region can also be determined by the sub-controller corresponding to that region. For example, if the conveyor path used to execute the conveyor command is a path that sequentially passes through three regions a, b, and c, the sub-controller managing region a can determine the sub-conveying path within region a, the sub-controller managing region b can determine the sub-conveying path within region b, and the sub-controller managing region c can determine the sub-conveying path within region c. Furthermore, if the conveyor path is divided into multiple sub-conveying paths for each region, the time point for determining the sub-conveying path for each region (i.e., the time point for calculating the optimal sub-conveying path using the path search algorithm described above) can be different for each region. For example, the sub-controller managing region c can determine the sub-conveying path within region c and notify the conveyor 2 immediately before the conveyor 2 enters region c. Alternatively, after temporarily determining the transport path for a particular transport vehicle 2, the transport vehicle controller 20 may update the transport path by recalculating it while the transport vehicle 2 travels along that path. In any case, the transport path of the transport vehicle 2 is determined based on the connection path cost of each connecting path 11.

[0039] The connection path cost associated with connection path 11 is, for example, a value related to the time required for the transport vehicle 2 to traverse connection path 11 (hereinafter referred to as "travel time"). A first example of connection path cost is the length of connection path 11 (the distance from the upstream node to the downstream node of connection path 11). A second example of connection path cost is the value obtained by dividing the length of connection path 11 by the average speed of the transport vehicle 2 (i.e., the theoretical average travel time). A third example of connection path cost is the average travel time of connection path 11 (e.g., the average travel time of a predetermined number of transport vehicles 2 that have just traversed connection path 11). However, connection path cost can also be in ways other than the first to third examples described above. In this embodiment, the connection path cost of each connection path 11 is the third example described above. In this case, the transport vehicle controller 20 may update the connection path cost of a connection path 11 each time a transport vehicle 2 passes through a connection path 11, or it may update the connection path cost of each connection path 11 together at a predetermined time step interval (15 minutes in this embodiment).

[0040] The transport vehicle controller 20 (the path determination unit 22 described later) prioritizes the transport path of the transport vehicle 2 as the transport path containing the candidate path with the smallest sum of the connecting path costs of the multiple connecting paths 11 contained in the candidate path used to enable the transport vehicle 2 to perform transport (transport corresponding to the specified transport command). For example, when using a shortest path search algorithm such as the Dijkstra algorithm described above, the candidate path with the smallest sum of the connecting path costs of the multiple connecting paths 11 contained in the candidate path is determined as the transport path. In addition, the transport vehicle controller 20 may not necessarily determine the transport path as the shortest path in a strict sense. For example, when the transport route 4 is divided into multiple regions as described above and path search is performed on each region, the path formed by connecting the shortest paths (sub-transport paths) found on each region may not be the shortest path when viewed as a whole of the transport route 4. Furthermore, when the transport vehicle controller 20 performs path search within a limited computation time in order to improve the response speed to the notification of the transport route of the transport vehicle 2, the determined transport path may sometimes be inconsistent with the shortest path when viewed as a whole of the transport route 4.

[0041] like Figure 3 As shown, the conveyor controller 20 can be configured as a computer system, which includes one or more processors such as CPUs (Central Processing Units) 201, one or more RAMs (Random Access Memory) 202 and one or more ROMs (Read Only Memory) 203 as main storage devices, an input device such as a keyboard for operator input 204, an output device such as a display for operator prompts 205, a communication module 206 for communicating with the conveyor 2, and auxiliary storage devices such as HDDs and SSDs 207. The conveyor controller 20 can also be configured as a single computer device or as multiple computer devices (e.g., multiple sub-controllers as described above).

[0042] The various functions of the transport vehicle controller 20 are achieved, for example, by reading a prescribed program into a memory such as RAM 202, and by activating the input device 204 and the output device 205 under the control of the processor 201, as well as by activating the communication module 206 and reading and writing data from RAM 202 and auxiliary storage device 207.

[0043] Next, refer to Figure 2 The functions of the conveyor controller 20 will be explained below. The conveyor controller 20 includes a storage unit 21, a route determination unit 22, a selection unit 23, and a cost change unit 24.

[0044] Storage unit 21 is a database that stores various information related to the transport vehicle system 1. Storage unit 21 may be a single database device or multiple database devices. In this embodiment, storage unit 21 stores... Figure 1 The layout of the transport path 4 as shown and the connection path costs associated with each of the multiple connection paths 11. Furthermore, if the cost change unit 24 described later temporarily increases the connection path cost of a certain connection path 11 (the connection path selected as a bottleneck connection path) through processing, the storage unit 21 temporarily stores the revised connection path cost of that connection path 11.

[0045] The path determination unit 22 determines the transport path of the transport vehicle 2 based on the connection path costs of each of the multiple connection paths 11 stored in the storage unit 21. For example, when a transport instruction is assigned to a transport vehicle 2, the path determination unit 22 determines the transport path for the transport vehicle 2 to execute the transport instruction (i.e., the path from the current location of the transport vehicle 2 through the From location of the transport instruction to the To location of the transport instruction). As described above, the path determination unit 22 prioritizes the candidate path with the smaller sum of the connection path costs of the multiple connection paths 11 contained in the candidate path among the multiple candidate paths of the transport vehicle 2 capable of executing the transport instruction as the transport path of the transport vehicle 2. For example, the path determination unit 22 determines the transport path of the transport vehicle 2 assigned the transport instruction by using a shortest path search algorithm such as the Dijkstra algorithm. The transport vehicle controller 20 notifies the corresponding transport vehicle 2 of the transport path determined by the path determination unit 22. As a result, the transport vehicle 2 can travel on the transport road 4 according to the transport path determined by the path determination unit 22.

[0046] Selection unit 23 selects from multiple link paths 11 the object link path 11a whose future utilization is expected to exceed a specified benchmark (see reference). Figure 4 (b)

[0047] Cost change department 24 targets the upstream confluence 9a (refer to) that is closest to the object connection path 11a in the multiple connection paths 11. Figure 5 Multiple upstream connection paths 11b1~11b12 (refer to) exist within the specified upstream range as the starting point. Figure 5 The index values ​​v1 to v12, representing the ease of concentration of the transport vehicle 2, are calculated respectively. Furthermore, the cost change department 24 prioritizes upstream connection paths with higher index values ​​as one or more bottleneck connection paths 11c (see...). Figure 5 This increases the cost of the connection path for more than one bottleneck connection path 11c.

[0048] In this embodiment, the selection unit 23 and the cost change unit 24 repeatedly execute the processing at predetermined time intervals (e.g., the time step interval described above). That is, the selection unit 23 repeatedly executes the processing of selecting the target connection path 11a at predetermined time intervals. Whenever the selection unit 23 selects the target connection path 11a, the cost change unit 24 selects (updates) one or more bottleneck connection paths 11c and increases the connection path cost of one or more bottleneck connection paths 11c.

[0049] As an example, the connection path costs (modified (upgraded) connection path costs) of one or more bottleneck connection paths 11c determined by the selection unit 23 and the cost change unit 24 at a certain time t are temporarily stored in the storage unit 21 and applied during the period from time t to the next time t+1 (i.e., the time after a predetermined time interval from time t). That is, when the path determination unit 22 performs the processing of determining the transport path of the transport vehicle 2 (e.g., shortest path search) during the period from time t to the next time t+1, it uses the connection path costs of one or more bottleneck connection paths 11c determined at time t. In addition, the connection path costs of one or more bottleneck connection paths 11c determined at time t are reset and deleted from the storage unit 21 after the above-mentioned period has elapsed (i.e., after the time t+1 has elapsed).

[0050] Reference Figure 4 as well as Figure 5 Specific examples of the processing of the above-mentioned selection section 23 and cost change section 24 will be explained in turn. Figure 4 This is a diagram used to illustrate an example of the processing of the selection unit 23. Figure 5 This is a diagram illustrating an example of the processing of the cost change department 24.

[0051] First, refer to Figure 4 Specific examples (the first and second examples) of the processing of the selection section 23 will be explained. Figure 4 (a) is a diagram showing a portion of the schematic transport path 4 used in this description. Figure 4 (b) is a diagram representing the object connection path 11a selected by the selection unit 23.

[0052] (Example 1)

[0053] In the first example, firstly, the selection unit 23 obtains, for each of the multiple connecting paths 11, the maximum permissible number representing the maximum number of transport vehicles 2 that can simultaneously exist within the connecting path 11. The maximum permissible number n for each connecting path 11 is obtained, for example, based on the length x of each connecting path 11, the total length y of a transport vehicle 2, and the minimum inter-vehicle distance z (i.e., a distance pre-set as the minimum inter-vehicle distance to avoid collisions between adjacent transport vehicles 2), by calculating the largest n that satisfies "n×y+(n-1)×z≤x". Alternatively, the selection unit 23 may calculate (obtain) the maximum permissible number n for each connecting path 11 by performing the calculations described above. Alternatively, the maximum permissible number n for each connecting path 11 may be pre-stored in the storage unit 21. In this case, the selection unit 23 may also obtain the maximum permissible number n for each connecting path 11 by referring to the storage unit 21.

[0054] Next, the selection unit 23 selects the connection paths 11 from among the multiple connection paths 11 where the number of transport vehicles simultaneously present in the connection path 11 is at or above a predetermined proportion (e.g., 70%) of the maximum allowable number of transport vehicles in that connection path 11 as target connection paths 11a. As an example, the transport vehicle controller 20 is configured to continuously monitor whether the number of transport vehicles in each connection path 11 is at or above the predetermined proportion, and when it detects that the number of transport vehicles in a certain connection path 11 is at or above the predetermined proportion, it sends a predetermined signal GL1 (Grid Lock Level 1) to that connection path 11. Furthermore, the transport vehicle controller 20 is configured to send a signal GL2 (Grid Lock Level 2) to that connection path 11 when it detects that the number of transport vehicles in that connection path 11 has reached the maximum allowable number of transport vehicles in that connection path 11. In this case, the connection path 11 for which the aforementioned signal GL1 is detected can be considered a connection path 11 for which the number of transport vehicles 2 using the connection path 11 may continue to exceed a predetermined benchmark (in this example, the benchmark for transmitting signal GL1) during a future period (i.e., a predetermined period after the time when signal GL1 is detected). Therefore, the selection unit 23 can also select the connection path 11 for which such signal GL1 is transmitted as an object connection path 11a whose future utilization is predicted to exceed the predetermined benchmark.

[0055] (Example 2)

[0056] In the second example, firstly, the selection unit 23 calculates, for each of the multiple connection paths 11, the number of past transport instructions that set a location (e.g., processing device 7, vault, etc.) contained in the connection path 11 as the destination (in this embodiment, From location or To location), which is predicted to occur in the future (in this embodiment, from the current time t to the next time t+1). Here, a transport instruction associated with a certain connection path (hereinafter, it is also referred to simply as "transport instruction of a certain connection path") refers to a transport instruction that sets a location contained in that connection path as the destination (From location or To location).

[0057] Next, the selection unit 23 selects the target connection path 11a based on the number of transmission instructions calculated for each of the multiple connection paths 11 (i.e., the predicted number of future transmission instructions). In this embodiment, the selection unit 23 obtains the predicted number of future transmission instructions for each connection path 11 by using the prediction model M, which will be described later. The prediction model M is a learned model configured to take the past number of transmission instructions for each connection path 11 as input (for example, the number of transmission instructions generated in each of the past three time steps), and output the predicted number of transmission instructions for each connection path 11 for the next time step (i.e., the period from time t to time t+1).

[0058] Reference Figures 6-9 Let's illustrate this with an example of the prediction model M. Figure 6 This example shows the delivery logs of multiple deliveries arriving at the From location during a certain period (two time steps TS1, TS2). Figure 6 A single record (line) represents the transport log corresponding to a transport instruction. As an example, the transport log includes information such as: From arrival date and time, transport completion date and time, From connection path, and To connection path. "From arrival date and time" indicates the date and time when transport vehicle 2 arrives at the From location. "Transport completion date and time" indicates the date and time when transport vehicle 2 arrives at the To location and completes the unloading of the item (FOUP). "From connection path" is information identifying the connection path 11 containing the From location (connection path ID). "To connection path" is information identifying the connection path 11 containing the To location (connection path ID).

[0059] Figure 7 The summary data for each period shown is summarized by each period (time step) and each link path. Figure 6The data is obtained from multiple transport logs shown. More specifically, by summarizing the number of transport commands set as From links (From transport command count) and the number of transport commands set as To links (To transport command count) for each link path, data is obtained that correlates the past specified period, the link path (link path ID), and the total number of transport commands (From transport command count + To transport command count; hereinafter referred to as "transport command count").

[0060] The number of From transmission commands for the connection path ID "1001" at time step TS1 is determined by the transmission log (see reference) for the period (here, the period up to "January 1, 00:00 to 00:14:59") within which the From arrival date and time are included in that time step TS1. Figure 6 The number of transport logs with the From link path "1001" in the time step TS1 is obtained by counting the number of transport logs with the From link path "1001" in the transport logs included in the time step TS1 on the transport completion date.

[0061] Here, it is assumed that there is a certain relationship (pattern) between the number of transmission commands for each connection path in the most recent past period and the number of transmission commands for each connection path in the future period (the next time step). Therefore, in this embodiment, a prediction model M is used, which takes the number of transmission commands for each connection path in the past period as input and outputs the prediction results of the number of transmission commands for each connection path in the future period corresponding to the next time step.

[0062] In this embodiment, each time step is included ( Figure 7 In the example, multiple connection paths (time steps TS1, TS2) Figure 7 In the example, the number of transmission instructions for each of the connection paths (IDs "1001" to "9990") is used as a vector of elements. Figure 7 In the example, data V1 and V2 are used as data for learning the prediction model M (hereinafter referred to as learning data or validation data). The prediction model M is constructed, for example, using Gradient Boosting Decision Tree (GBDT) as a form of supervised learning. However, the learning method used to construct the prediction model M is not limited to the GBDT described above.

[0063] Figure 8 This is a diagram illustrating an example of the learning and validation data used in the learning of the predictive model M. Figure 8 In the example, the prediction model M is constructed based on data of the number of transmission instructions for each link path over a past period (45 minutes in three time steps). Figure 7 The data (V1, V2) are used to output the predicted number of transport instructions for each linked path in the immediate future period (a time step of 15 minutes) following the past period. Figure 8 The example displays past periods corresponding to time steps with IDs "350" to "364". For example... Figure 8 As shown, the data on the number of transmission instructions over multiple past time steps is divided into learning data and verification data. Furthermore, the preferred period for the learning data (within...) Figure 8 In the example, the time step before ID "355" and the period of the data used for verification (in) Figure 8 In the example, the time step after ID "358" is set to be discontinuous in time. Figure 8 In the example, a 30-minute interval is set between the period of learning the data and the period of validating the data.

[0064] Figure 8 In this model, each row of data corresponds to either a training data point or a validation data point. Data during periods marked with shading (the period of three time steps) corresponds to the input data of the prediction model M, while data during periods completely shaded (the period of one time step) corresponds to the output of the prediction model M (positive result marker). In this way, by sliding the time steps sequentially from past periods containing multiple time steps, data for four consecutive time steps is obtained, thereby efficiently acquiring multiple training and validation data points. The prediction model M, for example, learns using multiple training data points to minimize the difference between the output obtained by inputting the data for the first three time steps of each validation data point into the prediction model M and the data for the last time step of each validation data point. Furthermore, the length of the input data period for the prediction model M (three time steps in this embodiment), the number of training data points, and the number of validation data points are appropriately set based on the accuracy evaluation results of the prediction model M.

[0065] Figure 9 This is a diagram schematically illustrating an example of predictive processing performed by a predictive model M (a learned model) created through machine learning as described above. For example... Figure 9As shown, if the prediction model M is input with data (V1, V2, V3) for the past period (t-2, t-1, t) of the most recent three time steps, it outputs data X for the future period (t+1) of the next time step. Data X represents the number of transport instructions (the number of transport instructions predicted to be generated) for each connection path in that future period. Thus, in this embodiment, the selection unit 23 uses the prediction model M to obtain the predicted number of transport instructions generated in the future period (data X output from the prediction model M) based on the number of past transport instructions that set the destinations of each connection path as destinations (in this embodiment, the data V1, V2, V3 of the most recent three time steps input to the prediction model M) as information related to the future utilization of each connection path.

[0066] Next, the selection unit 23 selects, for example, connection paths 11 where the predicted number of future transport instructions is above a predetermined threshold as target connection paths 11a. The threshold can also be set individually for each connection path 11. Alternatively, the selection unit 23 can select connection paths 11 where the increase in the number of transport instructions is above a predetermined benchmark as target connection paths 11a. For example, the selection unit 23 can also select connection paths 11 where the predicted number of future transport instructions is above a predetermined threshold. P The average number of transport instructions n relative to the three most recent time steps AVE proportion (n) P / n AVE The link path 11 that is above a predetermined threshold (e.g., 1.2 (20%)) is selected as the target link path 11a. As described above, according to the second example, the selection unit 23 can select the target link path 11a whose predicted future utilization exceeds a predetermined benchmark based on the predicted number of delivery instructions generated in the future period.

[0067] Furthermore, the first and second examples described above can be used concurrently. For example, the selection unit 23 may periodically execute the processing of the second example at predetermined time step intervals (15 minutes in this embodiment), and periodically execute the processing of the first example when the connection path 11 that has been reported with signal GL1 is detected. In addition, for the object connection path 11a selected by the first example (and the modified connection path cost of the bottleneck connection path 11c selected based on the object connection path 11a), it may be reset at the moment when the reporting of signal GL1 of the object connection path 11a is no longer performed, or it may be reset after a predetermined time has elapsed since the time when the reporting of signal GL1 was detected.

[0068] Next, refer to Figure 5Here is a specific example of how the cost change department 24 processes the selected object connection path 11a. First, as... Figure 5 As shown in (a), the cost change unit 24 selects multiple upstream connection paths that exist within a defined upstream range, starting from the upstream junction 9a closest to the object connection path 11a. The upstream junction 9a is the junction 9 closest to the object connection path 11a among multiple junctions 9 located upstream of the object connection path 11a. In this embodiment, as an example, the aforementioned "defined upstream range" is a sub-map G consisting of the range reachable from the upstream junction 9a upstream via N (for example, "N=4") or fewer connection paths 11. That is, in this embodiment, the cost change unit 24 selects multiple (12 in this embodiment) upstream connection paths 11b1 to 11b12 contained in the sub-map G.

[0069] Next, the cost change unit 24 obtains index values ​​v1 to v12 representing the concentration ease of the transport vehicle 2 for each upstream connection path 11b1 to 11b12 selected as described above. For example, the cost change unit 24 may also calculate index values ​​v1 to v12 based on graph theory. For example, the cost change unit 24 may also adopt the index related to centrality in graph theory as index values ​​v1 to v12. As an example, the cost change unit 24 may also calculate the BC (Betweenness Centrality) value of each upstream connection path 11b1 to 11b12 as index values ​​v1 to v12. Here, the BC value of a certain connection path 11 represents the frequency with which the connection path 11 exists on the shortest path between all pairs of nodes in the graph of the entire transport path 4 (a graph composed of multiple nodes 10 and multiple connection paths 11). It can be said that the larger the BC value of a connecting path 11, the easier it is for the transport vehicles 2 to be concentrated in the layout of the transport road 4. Furthermore, since the BC value of each connecting path 11 depends on the layout of the transport road 4, it can be calculated in advance. For example, the BC value of each connecting path 11 can also be calculated in advance and stored in the storage unit 21. In this case, the cost change unit 24 can also obtain the index values ​​v1 to v12 (BC values) of each upstream connecting path 11b1 to 11b12 by referring to the storage unit 21.

[0070] Next, the cost change department 24 prioritizes upstream connection paths 11b with high indicator values ​​as one or more bottleneck connection paths 11c. For example, the cost change department 24 selects the upstream connection paths up to the m-th (in this example, the third-highest) from the upstream connection paths 11b1 to 11b12 with indicator values ​​v1 to v12 arranged in descending order as bottleneck connection paths 11c. Here, as an example, the indicator values ​​v1, v3, and v5 of the three upstream connection paths 11b1, 11b3, and 11b5 are ranked as the third-highest. In this case, as... Figure 5 As shown in (b), the three upstream connection paths 11b1, 11b3, and 11b5 with index values ​​v1, v3, and v5 up to the third one above are selected as bottleneck connection paths 11c.

[0071] Next, the cost change unit 24 increases the connection path cost of each bottleneck connection path 11c selected as described above. The increase in connection path cost can be arbitrarily determined. For example, the cost change unit 24 may increase the connection path cost of the bottleneck connection path 11c by multiplying a predetermined cost multiplier K (e.g., K = 1.2) by the original connection path cost of the bottleneck connection path 11c. Alternatively, the cost change unit 24 may display each bottleneck connection path 11c on the screen of the operator's terminal and determine the increase in connection path cost of each bottleneck connection path 11c based on instructions from the operator. As another example, the cost modification unit 24 can increase the connection path cost of a bottleneck connection path 11c by changing the connection path cost of a bottleneck connection path 11c, thereby changing the shortest path from the current position to the target connection path 11a for at least a predetermined number of transport vehicles 2 from the first path utilizing the bottleneck connection path 11c to a second path that does not utilize (bypasses) the bottleneck connection path 11c. According to this method, it is expected that the number of transport vehicles 2 using the bottleneck connection path 11c to reach the target connection path 11a can be reduced. Furthermore, by adjusting the aforementioned "predetermined number," the increase in connection path cost can be adjusted. Specifically, by increasing the aforementioned "predetermined number," the increase in connection path cost can be increased.

[0072] Next, refer to Figure 10 An example of the processing of the conveyor system 1 (conveyor controller 20) is illustrated. For example... Figure 10 As shown, the processing of the path determination unit 22 is performed independently of the processing of the selection unit 23 and the cost change unit 24.

[0073] First, the processing flow of the selection unit 23 and the cost change unit 24 will be explained. If the calculation time point (trigger) arrives (step S11: Yes), the selection unit 23 executes the link path 11a for selecting objects (refer to...). Figure 4 The processing of (b) is described in step S12. In this embodiment, the trigger has a first time point and a second time point. The first time point is the time when a signal GL1 is transmitted for a certain connection path 11. In this case, in step S12, the selection unit 23 performs the processing of the first example described above. The second time point is a time point that is repeatedly generated at each predetermined time step (15 minutes in this embodiment). When the second time point arrives, the selection unit 23 performs the processing of the second example described above.

[0074] Next, as before... Figure 5 (a) and Figure 5 As explained in (b), the cost modification unit 24 selects one or more bottleneck connection paths 11c based on the object connection path 11a selected by the selection unit 23 (step S13). Next, the cost modification unit 24 increases the connection path cost of the selected bottleneck connection path 11c (step S14). The modified (increased) connection path cost of the bottleneck connection path 11c is stored in the storage unit 21.

[0075] Next, the processing flow of the path determination unit 22 will be explained. When the time for path searching (e.g., the time when the transport vehicle controller 20 receives the transport instruction from the upper controller 12) arrives (step S21: Yes), the path determination unit 22 retrieves the connection path costs of each of the multiple connection paths 11 from the storage unit 21 (step S22). At this time, the path determination unit 22 retrieves the connection path costs originally set for each connection path 11, and retrieves the corrected connection path cost of the bottleneck connection path 11c (i.e., the connection path cost of the bottleneck connection path 11c corrected by the latest step S14). Therefore, the path determination unit 22 uses the corrected connection path cost for the bottleneck connection path 11c, thereby enabling the path search described later. Next, the path determination unit 22 determines the transport path of the transport vehicle 2 by performing a path search based on the connection path costs of each of the multiple connection paths 11 (e.g., shortest path search using Dijkstra's algorithm, etc.) (step S23).

[0076] [Effects]

[0077] According to the above-described conveyor system 1, it does not only increase the cost of the connection path 11a, but also ensures that the upstream confluence 9a, which is closest to the object connection path 11a, exists within a specified upstream range. Figure 5The upstream connecting paths 11b1~11b12 of the subgraph G) are prone to congestion of transport vehicles, which increases the connection cost of the bottleneck connecting path 11c. This can prevent traffic congestion in the bottleneck connecting path 11c from occurring in advance. In other words, it can suppress the possibility of a significant increase in the number of transport vehicles using the bottleneck connecting path 11c to reach the target connecting path 11a in the future (i.e., the occurrence of traffic congestion in the bottleneck connecting path 11c). As a result, traffic congestion in the transport vehicle system 1 can be effectively suppressed.

[0078] To further address the aforementioned effects, according to the inventors, the transport vehicle 2 heading towards the target connection path 11a inevitably passes through the upstream connection path 11 of the target connection path 11a. Therefore, the upstream connection path 11 (upstream connection path) is more likely to become a bottleneck than the upstream confluence 9a, which is closest to the target connection path 11a. According to the transport vehicle system 1, by selecting the upstream connection path 11, particularly the one where transport vehicles 2 are considered to be easily concentrated, as the bottleneck connection path 11c, and increasing the connection path cost of the bottleneck connection path 11c, the system guides the vehicles to use the bottleneck connection path 11c in a way that reduces the number of transport vehicles using it, thereby achieving the aforementioned effects.

[0079] The cost change unit 24 obtains the index values ​​v1 to v12 for each of the multiple upstream connection paths 11b1 to 11b12 based on graph theory. According to the above structure, the bottleneck connection path 11c can be efficiently selected through relatively simple calculation processing. Furthermore, in this embodiment, by pre-calculating and storing the BC values ​​of each connection path 11 in the storage unit 21 as described above, the index values ​​v1 to v12 for each upstream connection path 11b1 to 11b12 can be easily and quickly obtained.

[0080] The selection unit 23 repeatedly performs the process of selecting the target connection path 11a at predetermined intervals. Each time the selection unit 23 selects the target connection path 11a, the cost change unit 24 performs the process of selecting one or more bottleneck connection paths 11c and increasing the connection path cost of those bottleneck connection paths 11c. Based on this structure, the combination of the target connection path 11a and the bottleneck connection path 11c can be appropriately updated at predetermined time intervals. Therefore, traffic congestion in the transport vehicle system 1 can be appropriately suppressed according to changes in the overall system condition.

[0081] For each of the multiple connection paths 11, the selection unit 23 obtains the maximum allowable number, representing the maximum number of transport vehicles 2 that can simultaneously exist in the connection path 11. It then selects the connection paths 11 whose number of transport vehicles present in the connection path 11 is at least a predetermined proportion of the maximum allowable number of that connection path 11 as target connection paths 11a. For example, the selection unit 23 performs the processing described in the first example above. Based on this structure, the target connection path 11a can be selected relatively easily based on the index of the maximum allowable number of each connection path 11.

[0082] For each of the multiple connection paths 11, the selection unit 23 calculates the number of transport instructions associated with each of the multiple connection paths 11 that are predicted to be generated in the future, based on the number of past transport instructions that set the locations included in the connection path 11 as destinations (in this embodiment, From location or To location). Based on the number of transport instructions calculated for each of the multiple connection paths 11, the selection unit 23 selects the target connection path 11a. For example, the selection unit 23 performs the processing described in the second example above. According to the above structure, based on the number of future transport instructions generated for each connection path 11 predicted according to the number of past transport instructions associated with each connection path 11, the target connection path 11a with a high probability of being generated in the future transport vehicle 2 can be selected with high accuracy.

[0083] [Variation Example]

[0084] The embodiments of this disclosure have been described above, but this disclosure is not limited to the above embodiments and various changes can be made without departing from its spirit.

[0085] For example, the index values ​​v1 to v12 of each upstream connection path 11b1 to 11b12 obtained by the cost change department 24 can also be graph theory-based index values ​​other than BC values ​​(e.g., index values ​​other than BC values ​​that represent the centrality of connection paths within the graph). Alternatively, the aforementioned index values ​​v1 to v12 can also be index values ​​obtained through methods different from graph theory.

[0086] The selection unit 23 may repeatedly execute the process of selecting the object connection path 11a without a predetermined interval, or it may execute it at any point in time. For example, if the selection unit 23 is configured to only execute the process described in the first example above, the process of the selection unit 23 (and the subsequent process of the cost change unit 24) may be executed intermittently. Alternatively, the selection unit 23 may select the object connection path 11a through processes other than the first and second examples described above. Furthermore, in the above embodiment, the object connection path 11a and the bottleneck connection path 11c (including the increased connection path cost) selected at a certain time are reset after a constant period has elapsed since that time, but they may also be reset based on predetermined conditions other than after a constant period. Examples of predetermined conditions include the utilization rate of the object connection path 11a (the number of transport vehicles in the connection path, the number of transport instructions generated, etc.) being lower than a predetermined benchmark.

[0087] Furthermore, in the above embodiment, an example of an item (transported object) transported by the transport vehicle 2 is a FOUP containing multiple semiconductor wafers, but the item is not limited to this. For example, it could also be other containers or other items containing glass wafers, intermediate masks, etc. In addition, the location where the transport vehicle system 1 is installed is not limited to a semiconductor manufacturing plant, and the transport vehicle system 1 can also be installed in other facilities.

[0088] Furthermore, in the above embodiments, the form in which multiple transport vehicles 2 transport items on the transport route 4 has been described. However, the structure of this disclosure can also be applied to a vehicle system in which multiple vehicles that do not transport items travel along a pre-set travel route. That is, in the above embodiments, it is not necessary for the transport vehicles 2 to travel based on a transport command that presupposes the transport of items; the transport vehicles 2 can also be configured to travel based on a travel command that does not involve the transport of items.

[0089] Explanation of reference numerals in the attached figures

[0090] 1...Conveyor system (traveling vehicle system); 2...Conveyor (traveling vehicle); 4...Conveyor path (traveling path); 8...Branch; 9...Merging point; 9a...Upstream merging point; 10...Node; 11...Connection path; 11a...Object connection path; 11b1~11b12...Upstream connection path; 11c...Bottleneck connection path; 20...Conveyor controller (controller); 21...Storage unit; 22...Path determination unit; 23...Selection unit; 24...Cost change unit.

Claims

1. A vehicle system comprising: a plurality of vehicles traveling along a pre-defined route; And a controller that controls the movement of the vehicles by assigning driving instructions indicating a destination on the driving route to the vehicles selected from the plurality of vehicles. The vehicle driving system is characterized in that... The travel route includes: multiple nodes comprising branching sections and converging sections, and multiple connecting paths connecting the nodes. The controller has: The storage unit stores the layout of the travel route and the connection path costs associated with the plurality of connection paths, respectively. The route determination unit, based on the link path cost of each of the plurality of link paths, prioritizes the candidate path with the smaller sum of the link path costs of the plurality of link paths contained in the candidate path of the vehicle executing the driving command as the driving path of the vehicle. The selection unit selects from the plurality of connection paths the object connection paths whose future utilization is expected to exceed a predetermined benchmark; and The cost change department obtains an index value representing the concentration ease of the vehicle for each of the multiple upstream connection paths that exist within a specified upstream range, starting from the upstream junction closest to the target connection path among the multiple connection paths. It prioritizes the upstream connection path with the higher index value as one or more bottleneck connection paths and increases the connection path cost of the one or more bottleneck connection paths.

2. The vehicle driving system according to claim 1, characterized in that, The cost change department calculates the respective index values ​​of the multiple upstream connection paths based on graph theory.

3. The vehicle driving system according to claim 1, characterized in that, The selection unit repeatedly performs the process of selecting the connection path of the object at specified time intervals. Whenever the selection unit selects the object connection path, the cost change unit performs a process that selects one or more bottleneck connection paths and increases the connection path cost of the one or more bottleneck connection paths.

4. The vehicle driving system according to claim 1, characterized in that, For the selected part For each of the multiple connection paths, obtain the maximum allowed number representing the maximum number of vehicles that can simultaneously exist within the connection path. The connection paths in which the number of vehicles existing in the connection path is greater than or equal to a predetermined proportion of the maximum allowable number of vehicles in the connection path are selected as the target connection paths.

5. The vehicle driving system according to claim 1, characterized in that, The driving instruction is associated with a destination contained in any of the plurality of connecting paths. The controller controls the movement of the vehicle in a manner that causes the vehicle, which has been assigned the driving command, to proceed to the destination. For each of the plurality of connection paths, the selection unit calculates the number of driving instructions associated with each of the plurality of connection paths that are predicted to be generated in the future, based on the number of past driving instructions that set the locations contained in the connection path as the destination, and selects the target connection path based on the number of driving instructions calculated for each of the plurality of connection paths.

Images