JP6389976B1 - Vehicle facility monitoring system - Google Patents

Abstract

An object of the present invention is to accurately determine the state of a predetermined parking / stopping area in a vehicle facility and efficiently operate the vehicle facility. A vehicle facility monitoring system according to an embodiment includes a camera, a generation unit, and a determination unit. The camera is provided for each predetermined parking / parking area in the vehicle facility, and images each parking / parking area. The generation unit performs machine learning using a positive captured image captured by the camera, thereby determining the state of the parking / stopping region including a case where the allowable number of vehicles for each parking / stopping region dynamically changes. Generate a model. The discriminating unit discriminates the state of the parking / stopping region based on an output value obtained by inputting a captured image, which is a portion to be discriminated captured by the camera, to the state discrimination model. [Selection] Figure 2

Description

  The disclosed embodiment relates to a vehicle facility monitoring system.

  2. Description of the Related Art Conventionally, a management system that manages entry / exit of vehicles in a vehicle facility having a predetermined parking / stopping area such as a parking lot is known.

  In such a management system, for example, by counting the number of vehicles entering the entrance gate and the number of vehicles entering the entrance gate, and subtracting the total number of vehicles from the total number of vehicles entered, Know the number of parked cars. When the number of parked vehicles reaches a predetermined maximum number of parked vehicles, entry of the vehicle is restricted, and when the number of parked vehicles is below the maximum number of parked vehicles, entry of the vehicle is permitted (see, for example, Patent Document 1). ).

JP 2003-141581 A

  However, the above-described prior art has room for further improvement in accurately determining the state of a predetermined parking / stopping area in the vehicle facility and operating the vehicle facility efficiently.

  Specifically, since the above-described conventional technology detects a participating vehicle at a participating gate machine, for example, even if the participating vehicle departs and the parking / parking area is vacant, the participating vehicle reaches the participating gate machine. In the meantime, the parking / stopping area is determined to be full. For this reason, the time for which entry of the vehicle is restricted becomes long, and the operation rate of the vehicle facility is lowered.

  In particular, if the vehicle facility is a distribution center that serves as a truck transportation distribution center, the distribution center will increase in size in response to an increase in distribution demand accompanying the rapid spread of electronic commerce using the Internet these days. In addition, the frequency of transportation opportunities is increasing. In such a distribution center, it is strongly desired to improve the operational efficiency by accurately determining the state of the parking and stopping area of the truck as well as the cargo handling area.

  One aspect of the embodiment is made in view of the above, and provides a vehicle facility monitoring system capable of accurately determining the state of a predetermined parking / stopping area in the vehicle facility and operating the vehicle facility efficiently. The purpose is to do.

A vehicle facility monitoring system according to an aspect of an embodiment includes a camera, a generation unit, and a determination unit. The camera is provided for each predetermined parking / stopping area in the vehicle facility, and images each of the parking / stopping areas. The parking unit includes a case where the allowable number of vehicles for each parking / parking region dynamically changes by executing machine learning using a positive captured image captured by the camera. Generate a state discrimination model for the region. The discriminating unit discriminates the state of the parking / stopping region based on an output value obtained by inputting the picked-up image as a discrimination target imaged by the camera to the state discrimination model. The vehicle facility is a logistics facility, and the parking and stopping area is a span that is a predetermined section in the logistics facility or a waiting area for a truck in the site of the logistics facility.

  According to one aspect of the embodiment, it is possible to accurately determine the state of a predetermined parking / stopping area in the vehicle facility and efficiently operate the vehicle facility.

FIG. 1 is a schematic explanatory diagram of a vehicle facility monitoring system according to an embodiment. FIG. 2 is a block diagram of the vehicle facility monitoring system according to the embodiment. FIG. 3A is a diagram (part 1) illustrating an example of a state determination model. FIG. 3B is a second diagram illustrating an example of a state determination model. FIG. 4A is a diagram (part 1) illustrating a specific example of a span state pattern. FIG. 4B is a second diagram illustrating a specific example of a span state pattern. FIG. 4C is a third diagram illustrating a specific example of a span state pattern. FIG. 4D is a diagram (part 4) illustrating a specific example of a span state pattern. FIG. 4E is a fifth diagram illustrating a specific example of a span state pattern. FIG. 4F is a sixth diagram illustrating a specific example of a span state pattern. FIG. 4G is a diagram (No. 7) illustrating a specific example of a span state pattern. FIG. 4H is a diagram (No. 8) illustrating a specific example of a span state pattern. FIG. 4I is a diagram (part 9) illustrating a specific example of a span state pattern. FIG. 4J is a diagram (No. 10) illustrating a specific example of a span state pattern. FIG. 4K is a diagram (No. 11) illustrating a specific example of a span state pattern. FIG. 5A is a diagram illustrating a specific example of the overall monitoring screen. FIG. 5B is a diagram illustrating a specific example of the individual monitoring screen. FIG. 5C is a diagram illustrating a specific example of the usage status monitoring screen. FIG. 5D is a diagram illustrating a specific example in the case of security management of a distribution center.

  Hereinafter, an embodiment of a vehicle facility monitoring system disclosed in the present application will be described in detail with reference to the accompanying drawings. In addition, this invention is not limited by embodiment shown below.

  In the following, the case where the vehicle facility is a distribution center serving as a distribution base for truck transportation will be described as an example. Vehicles that can be loaded and transported to and from the distribution center are widely included.

  First, the outline | summary of the vehicle facility monitoring system which concerns on embodiment is demonstrated with reference to FIG. FIG. 1 is a schematic explanatory diagram of a vehicle facility monitoring system 1 according to the embodiment.

  As shown in FIG. 1, the vehicle facility monitoring system 1 according to the embodiment includes a distribution center 2 and a server device 10. The distribution center 2 performs various operations related to distribution such as arrival, storage, picking, distribution processing, inspection, and shipment of transported goods.

  For example, the distribution center 2 has a plurality of floors into which trucks can enter, and a span 21 that is a predetermined section is provided on each floor. The span 21 is an example of a predetermined parking and stopping area in the distribution center 2 and is an area that serves as a truck unloading area. In the present embodiment, the distribution center 2 has a plurality of floors, but the distribution center 2 may be a single floor, that is, a flat.

  As shown in FIG. 1, the span 21 is formed between pillars in the distribution center 2, for example. The span 21 is formed to have a dimension capable of stopping a plurality of tracks. The stop point of the truck in the span 21 is called “berth”.

  As shown in FIG. 1, the span 21 can stop a plurality of trucks so that the “allowable number dynamically changes”. In other words, the number of berths can be changed dynamically.

  In the present embodiment, it is assumed that three small trucks T <b> 2 can be stopped in one span 21 as indicated by a span 21 of “402” in the drawing. Further, for example, as shown by a span 21 of “403” in the figure, it is assumed that two large trucks T1 can be stopped in one span 21.

  “Large” and “small” as used herein refer to the relative size, and are classified according to vehicle dimensions, maximum loading capacity, total vehicle weight, etc. It does not refer to the standards of “truck” and “large truck”.

  Further, as indicated by a span 21 of “404” in the drawing, the span 21 can include a shutter 22 between the aforementioned pillars. By closing the shutter 22, the span 21 is locked. The shutter 22 may be stored in any manner such as “winding type”, “folding type”, and “over slider type”.

  In addition, the distribution center 2 has a waiting area 23 for trucks T1 and T2, for example, in a site. The waiting area 23 is an example of a predetermined parking / stopping area in the distribution center 2, and is a place where the trucks T1 and T2 before entering the distribution center 2 are kept waiting.

  The vehicle facility monitoring system 1 includes a camera 24. The camera 24 is provided for each span 21 and waiting area 23 in the distribution center 2 and images each span 21 and waiting area 23.

  The camera 24 is attached to, for example, a pillar in the distribution center 2 in such a direction that a specific span 21 can be imaged from, for example, an oblique front. Further, the camera 24 is attached to a pillar in the site of the distribution center 2, for example, so that the entire waiting area 23 can be imaged from obliquely above. However, in any case, the direction in which the camera 24 is attached is not limited.

  The camera 24 may be a wireless camera or a wired camera. In the case of a wireless camera, there is no need to route the wiring of the camera 24, and there is an advantage that expansion / change is easy. The camera 24 may be a still camera or a movie camera. In the present embodiment, it is assumed that the camera 24 is a wireless still camera.

  In the present embodiment, by executing machine learning using a positive captured image captured by the camera 24, the allowable number dynamically changes for each predetermined parking / stopping area in the distribution center 2. It was decided to generate a state discrimination model for parking and stopping areas including cases. Then, based on an output value obtained by inputting a captured image captured by the camera 24 as a determination target to the state determination model, the state of each parking / parking area, for example, the span 21 is determined.

  Specifically, as shown in FIG. 1, the vehicle facility monitoring system 1 periodically transmits captured images captured by the cameras 24 via the network N such as the Internet from the distribution center 2 ( Step S1).

  And the server apparatus 10 collects the captured image of each camera 24 transmitted from the distribution center 2 (step S2). The server device 10 is provided as a cloud server in cloud computing, for example.

  And the server apparatus 10 produces | generates a state discrimination | determination model by the machine learning using the part for learning used as a positive example among the collected captured images (step S3). A known algorithm such as deep learning can be used for machine learning.

  Then, the server device 10 performs, for example, state determination of each span 21 using the generated state determination model (step S4). Specifically, the server device 10 inputs the discrimination target part of the collected captured images to the state discrimination model, and based on the output value from the state discrimination model obtained thereby, The state of each span 21 including the case of “changing” is determined in real time.

  For example, the server device 10 determines an object existing in the span 21. The object is, for example, a vehicle having a different size. Further, the server device 10 determines whether the span 21 is full, for example. Further, the server device 10 determines, for example, the open / closed state of the shutter 22. These discrimination examples of parking / stopping areas that can be discriminated by the vehicle facility monitoring system 1 according to the present embodiment will be described later with reference to FIGS. 4A to 4K.

  Then, the server device 10 displays the determined state of each span 21 on, for example, a monitoring screen such as a display provided in the server device 10. The operator grasps the state of each span 21 by confirming such a monitoring screen. For example, if a certain span 21 is vacant, the operator guides the tracks T1 and T2 waiting in the waiting area 23 to the span 21.

  A specific example of the monitoring screen will be described later with reference to FIGS. 5A and 5B. Moreover, the display destination of the monitoring screen may be, for example, a portable information terminal carried by the operator.

  As described above, since the vehicle facility monitoring system 1 according to the present embodiment determines each state of the predetermined parking / stopping area in real time based on the state determination model generated based on the captured image of the camera 24. It is possible to accurately determine the state of the parking / stopping area. Each state to be discriminated includes the case where the “allowable number changes dynamically” as described above. For example, it is possible to guide the tracks T1 and T2 having a size corresponding to the size of the berth vacated in the span 21. It becomes possible, and it becomes possible to operate the distribution center 2 efficiently.

  In addition, based on the determination result by the state determination model, it is possible to derive information on the usage status of each span 21 and display it on the monitoring screen. A specific example of this point will be described later with reference to FIG. 5C. Further, the security of the distribution center 2 can be managed based on the open / closed state of the shutter 22 determined by the state determination model. A specific example of this point will be described later with reference to FIG. 5D.

  Hereinafter, a configuration example of the vehicle facility monitoring system 1 according to the above-described embodiment will be described more specifically.

  FIG. 2 is a block diagram of the vehicle facility monitoring system 1 according to the embodiment. In FIG. 2, constituent elements necessary for describing the features of the present embodiment are represented by functional blocks, and descriptions of general constituent elements are omitted.

  In other words, each component illustrated in FIG. 2 is functionally conceptual and does not necessarily need to be physically configured as illustrated. For example, the specific form of distribution / integration of each functional block is not limited to the one shown in the figure, and all or a part thereof is functionally or physically distributed in an arbitrary unit according to various loads or usage conditions.・ It can be integrated and configured.

  In the description with reference to FIG. 2, the description of the components already described so far may be simplified or omitted.

  As shown in FIG. 2, the vehicle facility monitoring system 1 includes a distribution center 2 and a server device 10. The distribution center 2 includes a plurality of spans 21, a standby area 23, a relay device 25, and a gateway device 26.

  The relay device 25 is provided, for example, for each floor of the distribution center 2 and aggregates the captured images of the cameras 24 of each span 21 and transmits the collected images to, for example, the gateway device 26 that is also provided for each floor.

  In addition, the relay device 25 is provided between the camera 24 in the standby area 23 and the gateway device 26, for example, in the site of the distribution center 2, and installs the captured image of the camera 24 in the standby area 23 to the gateway device 26. By providing the relay device 25, even if the distribution center 2 is a large scale, it is possible to collect and periodically collect the captured images of the spans 21 and the standby areas 23.

  The gateway device 26 is a relay device from the distribution center 2 to the network N. The gateway device 26 is connected to the relay device 25 through a fieldbus network, for example, and transmits a captured image received from the relay device 25 to the network N.

  The server device 10 includes a communication unit 11, a control unit 12, and a storage unit 13. The control unit 12 includes a collection unit 12a, an extraction unit 12b, a generation unit 12c, a determination unit 12d, and a monitoring unit 12e.

  The communication part 11 is implement | achieved by NIC (Network Interface Card) etc., for example. The communication unit 11 is connected to the network N in a wired or wireless manner, receives images captured by the cameras 24 from the distribution center 2 via the network N, and passes them to the control unit 12.

  The control unit 12 is a controller, and various programs stored in a storage device inside the server device 10 are stored in a RAM (Random Access Memory) by, for example, a CPU (Central Processing Unit) or an MPU (Micro Processing Unit). ) Is executed as a work area. Moreover, the control part 12 is implement | achieved by integrated circuits, such as ASIC (Application Specific Integrated Circuit) and FPGA (Field Programmable Gate Array), for example.

  The storage unit 13 is realized by, for example, a semiconductor memory device such as a RAM or a flash memory, or a storage device such as a hard disk or an optical disk. In the example of FIG. 2, a captured image DB (database) 13a and a learning device are used. A target data set 13b, a first discrimination model 13c and a second discrimination model 13d as state discrimination models, a discrimination target data set 13e, and a discrimination result DB 13f are stored.

  The control unit 12 performs overall control of the server device 10. The collection unit 12a collects captured images of each camera 24 from the distribution center 2 via the communication unit 11, and stores them in the captured image DB 13a.

  The extraction unit 12b extracts, from the captured image DB 13a, a captured image that is a positive example used for machine learning, that is, a learning target, and stores it in the learning target data set 13b. Further, a captured image that is a discrimination target during operation of the distribution center 2, that is, a discrimination target is extracted from the captured image DB 13 a and stored in the discrimination target data set 13 e.

  The generation unit 12c acquires the learning target data set 13b, performs machine learning using the learning target data set 13b, and generates a state determination model.

  Here, an example of the state determination model will be described with reference to FIGS. 3A and 3B. 3A and 3B are diagrams (part 1) and (part 2) illustrating an example of the state determination model.

  As shown in FIG. 3A, for example, the generation unit 12c performs machine learning using a captured image of a completely empty state of the span 21 as a learning target (step S11), and generates a first discrimination model 13c.

  On the other hand, as illustrated in FIG. 3B, for example, the generation unit 12c performs machine learning using captured images of each state pattern of the span 21 as a learning target (step S12), and generates a second discrimination model 13d. .

  Here, the “state pattern” refers to each state that can occur with respect to the span 21 in the operation of the distribution center 2. For example, “two large trucks”, “three or more small trucks”, “large and small trucks” Clustering is performed for each pattern such as “mixed” and “shutter closed”. With the second discrimination model 13d generated based on the learning target of each state pattern, the state pattern when the span 21 is not in a completely empty state can be modeled.

  Returning to FIG. 3A, the determination unit 12d described later first inputs a captured image for the determination target to the first determination model 13c, and as a result, has already learned based on the output value obtained from the first determination model 13c. The degree of deviation from the class is calculated (step S13). If the degree of divergence is equal to or less than a predetermined threshold, for example, the state of the span 21 indicated by the input captured image indicates a completely empty state. Further, if the degree of divergence exceeds the threshold value, it indicates that the state of the span 21 indicated by the input captured image is not a complete empty state.

  Then, as illustrated in FIG. 3B, the determination unit 12d inputs the captured image for the determination target having a large degree of deviation to the second determination model 13d, and the output value (for example, FIG. 3D) of the second determination model 13d obtained as a result. Based on the “category ID”), the state pattern of each span 21 is determined (step S14).

  It should be noted that the state determination model shown in FIGS. 3A and 3B and the determination method using the model are merely examples, and the number of models constituting the state determination model, clustering in each model, and the like are shown in FIG. 3A and FIG. It is not limited to the example shown in 3B.

  Returning to the description of FIG. The determination unit 12d acquires the determination target data set 13e, inputs this to the first determination model 13c, and receives an output value from the first determination model 13c. Then, the determination unit 12d calculates the above-described divergence based on the received output value. If the divergence is equal to or less than a predetermined threshold, the determination unit determines that the span 21 is in a completely empty state and determines the determination result. Store in the result DB 13f.

  In addition, if the divergence degree exceeds the threshold value, the determination unit 12d inputs the captured image for the corresponding determination target to the second determination model 13d. Then, the determination unit 12d determines the state pattern of each span 21 based on the output value of the second determination model 13d obtained as a result, and stores the determination result in the determination result DB 13f.

  Here, each example of the state pattern which the discrimination | determination part 12d discriminate | determines is demonstrated more concretely using FIG. 4A-FIG. 4K. 4A to 4K are diagrams (No. 1) to (No. 11) showing specific examples of the state pattern of the span 21. FIG.

  FIG. 4A shows a state pattern in which the span 21 is in a completely empty state. As shown in FIG. 4A, even in the same completely empty state, even if there is a difference in brightness due to time zone or the like by executing machine learning using captured images with different illuminance, the state of span 21 Can be generated with high accuracy. The same applies to the state patterns shown in FIGS. 4B to 4K.

  Next, FIG. 4B shows a state where two large trucks T <b> 1 are stopped in the span 21. In such a case, the span 21 is full when two tracks T1 are allowed in the span 21 by executing machine learning including the case where one of the tracks T1 is present / not present. Or whether one berth is vacant.

  Next, FIG. 4C shows a state where three small trucks T <b> 2 are stopped in the span 21. In such a case, when one of the two tracks T2 is allowed to be contained in the span 21 by executing machine learning including each of the cases where the track T2 is present / not present, the span 21 is It is possible to determine whether the vehicle is full or whether there is an empty berth.

  Next, FIG. 4D shows a state in which the large truck T1 and the small truck T2 are mixed in the span 21. In such a case, by executing machine learning including either “when present / not present”, when the tracks T1 and T2 can be mixed in the span 21, the span 21 is full, or It is possible to determine whether there is a vacancy in the berth.

  In the case of such mixed patterns, depending on the dimensions of the span 21, a state pattern in which small trucks T2 are stopped in a column in the depth direction can be assumed. In order to cope with such a case, as shown in FIG. 4E represented as a schematic plan view, for example, a machine 24 is attached to the ceiling of the span 21 and machine learning is performed based on a captured image obtained by capturing the span 21 from above. May be executed to generate a state determination model.

  Then, as shown in FIG. 4F, which is also represented as a schematic plan view, for example, a case where a large truck T1 stops and one or both of the small trucks T2 which stop in a column next to each other are “present / not present” respectively. By executing the included machine learning, it is possible to determine whether the truck 21 is full or the berth has an empty space in the span 21 in which the parallel stop of the truck T2 can be allowed.

  In addition, the description has been given mainly by taking the relative size of the trucks T1 and T2 in the span 21 as an example, but it is of course classified according to vehicle dimensions, maximum loading capacity, total vehicle weight, and the like. According to the standard, “small trucks”, “medium trucks”, “large trucks”, etc. may be clustered and machine learning may be executed. Needless to say, machine learning may be performed by clustering captured images based on differences between manufacturers and vehicle types.

  Next, FIG. 4G shows a closed state of the shutter 22 included in the span 21. By executing machine learning including the closed state of the shutter 22, it is possible to determine whether the span 21 is in the closed state.

  As shown in FIG. 4H, the determination result of the captured image of the span 21 is time-series using the state determination model generated by executing the machine learning including the case where the degree of the closed state of the shutter 22 is different. It is also possible to determine whether the shutter 22 has changed from the closed state to the open state, whether the shutter 22 has changed from the open state to the closed state, whether it is in the middle, or whether it is stopped in the middle. It is.

  Next, FIG. 4I shows a state where an object other than a vehicle, for example, a container C is placed. In such a case, an object other than a vehicle (in this case, container C) exists in the span 21 by executing machine learning including the case where either one of the containers C is present / not present. Can be determined. In such a case, it is possible to determine whether the span 21 is in a state where the tracks T1 and T2 cannot enter or whether one of the berths is vacant.

  Next, FIG. 4J shows a state where an object other than the vehicle, for example, a human M such as an operator exists. In such a case, it is possible to determine that the person M exists in the span 21 by executing machine learning including the case where the person M exists / does not exist. In such a case, it is possible to determine whether the span 21 is in a state where the tracks T1 and T2 cannot enter, or a state where the span 21 can enter but requires attention because there is an operator.

  As shown in FIG. 4I or FIG. 4J, by performing machine learning including a case where there is a work vehicle such as a forklift that may exist in the span 21, there is a work vehicle. It is also possible to determine the state relating to the presence of

  Next, FIG. 4K shows an example in which the numbers of the trucks T1 and T2 stopped in the span 21 are determined. For example, as shown in FIG. 4K, the span 21 is assumed to be full. In such a state, machine learning is performed based on the captured image obtained by capturing the entire one span 21, and the number is determined by machine learning of the range designation portions of the number positions R 1 and R 2 in the captured image. Reading and discrimination are possible.

  As a machine learning algorithm in such a case, for example, “CNN (Convolutional Neural Network)” which is a kind of deep learning, “Faster R-CNN (Region-based Convolutional Neural Networks)” or the like can be used. .

  Returning to the description of FIG. 2, the monitoring unit 12e will be described. The monitoring unit 12e acquires the state of the span 21 determined by the determination unit 12d from the determination result DB 13f and displays it on a predetermined monitoring screen. The monitoring screen is displayed on the display unit 30 which is a display or the like connected to the server device 10, for example.

  In addition, the monitoring unit 12e displays at least the full state of each span 21 and the open / closed state of the shutter 22 of the span 21 on the monitoring screen. In addition, the monitoring unit 12e displays the latest captured image of the span 21 or the standby area 23 by the camera 24 on the monitoring screen. The captured image is stored in the discrimination result DB 13f in association with the discrimination result, for example.

  Here, a specific example of the monitoring screen will be described with reference to FIGS. 5A and 5B. FIG. 5A is a diagram illustrating a specific example of the overall monitoring screen. FIG. 5B is a diagram illustrating a specific example of the individual monitoring screen.

  As shown in FIG. 5A, the overall monitoring screen for monitoring the entire distribution center 2 displays, for example, at least the full state of each span 21 and the open / closed state of the shutter 22 of the span 21. 5A, “101” to “105” on 1F, “201” to “205” on 2F, “301” to “305” on 3F, “401” to “405” on 4F, Although the case where the state of the span 21 is displayed is shown, the number of floors and the number of spans are not limited.

  Specifically, the overall monitoring screen displays, for example, icons for each span 21 of each floor. Each span is displayed by displaying a word indicating the state such as “full”, “0/2” or “1/2”, “closed”, “NG”, etc. The state of 21 is visually recognized by a viewer such as an operator. For example, the latest status can be displayed by pressing an “update” button. The “update” button may include a switching function between “manual update” and “automatic update”.

  For example, “full” indicates that the span 21 is full, and is displayed when the determination unit 12d determines that the span 21 is full.

  For example, fractions such as “0/2”, “1/2”, and “2/3” indicate that there is an empty berth in the span 21. For example, “0/2” is displayed when the determination unit 12d determines that the span 21 is completely empty when the corresponding span 21 can accept two large trucks T1. If not only the track T1 but also a small track T2 can be allowed, “0/3” may be displayed together with “0/2”.

  For example, “1/2” is displayed when the determination unit 12d determines that there is one empty berth when the corresponding span 21 allows two large trucks T1. .

  For example, “2/3” is displayed when the determination unit 12d determines that there is one empty berth when the corresponding span 21 can accept three small trucks T2. .

  For example, “closed” indicates that the shutter 22 of the span 21 is in the closed state, and is displayed when the determination unit 12d determines that the shutter 22 of the corresponding span 21 is in the closed state. In the example shown in FIG. 5A, all the spans 21 are in the “closed” state in 4F, and in such a case, the monitoring unit 12e can perform security management on the floor of 4F, for example. Such an example will be described later with reference to FIG. 5D.

  For example, “NG” indicates that the shutter 22 of the span 21 is in an abnormal state, and is displayed when, for example, the determination unit 12d determines that the shutter 22 of the corresponding span 21 is out of order. The The determination unit 12d can determine that the shutter 22 is in failure, for example, when a predetermined time elapses with the shutter 22 stopped halfway.

  Moreover, as shown to FIG. 5A, the whole monitoring screen can display the number of empty berths of each floor, for example. At this time, even if the display is such as “2F” “5 (of which 1 is small)”, the size of the empty berth, in other words, the size of the trucks T1 and T2 that can be stopped is displayed. Good.

  Further, as shown in FIG. 5A, the entire monitoring screen can display the latest captured image of the waiting area 23. For example, the latest captured image can be displayed by pressing the “shoot” button.

  Further, the overall monitoring screen can be changed to an individual monitoring screen of the span 21 by performing an operation for designating a specific span 21, for example. Here, it is assumed that the span 21 indicated by “102” of 1F is designated.

  Then, as shown in FIG. 5B, the display can be shifted to the individual monitoring screen of the span 102 of “102”. On the individual monitoring screen, the full information on the berth in the designated span 21 and the parking time, for example, are displayed. Similar to the overall monitoring screen, for example, the latest status can be displayed by pressing an “update” button.

  5B, the individual monitoring screen can display the latest captured image of the corresponding span 21. Further, for example, the latest captured image can be displayed by pressing a “shoot” button.

  The operator of the distribution center 2 visually recognizes the individual monitoring screen, and the container C is placed in the detailed status of the corresponding span 21, for example, the span 21 of “102” from the displayed parking time. It can be immediately confirmed that the trucks T1 and T2 cannot be entered.

  Then, the operator identifies the span 21 with an empty berth by visually recognizing such an overall monitoring screen and the individual monitoring screen, and displays the latest images of the waiting areas 23 on the tracks T1 and T2 that can stop at the span 21. And can be guided, for example, by voice or a guide light. In addition, you may guide | induct by notifying the information of empty berth to portable information terminals which the driver of truck T1, T2 carries, or vehicle-mounted apparatuses, such as a navigation mounted in truck T1, T2.

  Returning to the description of FIG. In addition, the monitoring unit 12e accumulates and analyzes the state of the span 21 determined by the determination unit 12d, derives information on the usage status of the span 21, and presents information on the usage status. Further, the monitoring unit 12e performs security management of the distribution center 2 based on the open / closed state of the shutter 22.

  Here, an example in which the monitoring unit 12e presents information on the usage status of the span 21 will be described with reference to FIG. 5C. An example in which the monitoring unit 12e performs security management of the distribution center 2 based on the open / closed state of the shutter 22 will be described with reference to FIG. 5D.

  FIG. 5C is a diagram illustrating a specific example of the usage status monitoring screen. FIG. 5D is a diagram showing a specific example in the case of security management of the distribution center 2.

  The monitoring unit 12e can derive information on the usage status of each span 21 based on the determination result DB 13f and display the information on the usage status monitoring screen as illustrated in FIG. 5C.

  The usage status monitoring screen displays, for example, “operating state”, “parking start” date, “elapsed time”, “cumulative usage time”, “usage rate”, “message”, etc. for each span 21 or berth. be able to.

  “Operating state” corresponds to, for example, the wording that briefly indicates the state of each span 21 described above. “Parking start” and “elapsed time” are literally the time when parking was started and the time elapsed since parking was started. The “cumulative usage time” is the cumulative time that the corresponding span 21 or berth is used on the corresponding operation day. “Usage rate” is a usage rate (may be referred to as an operating rate) on the day of operation. The “message” displays special notes etc. that should be notified on the system.

  The operator of the distribution center 2 can guide the trucks T1 and T2 preferentially to a berth with a low usage rate, for example, in the span 21 where there is an empty berth by visually recognizing such a use state monitoring screen. Therefore, the distribution center 2 can be operated efficiently.

  As already described in the description with reference to FIG. 5A, for example, the monitoring unit 12e can perform security management on the floor in which all the spans 21 are in the “closed” state on the overall monitoring screen. If all the spans 21 on each floor are in the “closed” state, the monitoring unit 12e can perform security management for the entire building of the distribution center 2.

  Here, as in FIG. 5A, the case where all the spans 4F are in the “closed” state will be described as an example. As shown in FIG. 5D, it is assumed that all spans 21 of 4F are in the “closed” state.

  In such a case, the monitoring unit 12e determines that all spans 21 of 4F are in the door-tight state, and shifts 4F to the security management state. For example, as shown in FIG. 5D, the monitoring unit 12e displays that 4F is “security monitoring in progress”.

  In the distribution center 2, for example, a specific tenant may use one floor as a dedicated floor, so that a situation where security management is required for each floor may occur.

  The monitoring unit 12e also monitors the status of each span 21 based on the determination result of the determination unit 12d even in such a security management state. Here, as illustrated in FIG. 5D, it is assumed that the determination unit 12d determines “the shutter 22 has changed from the closed state” in the span 21 of “404”.

  Then, based on the determination result, for example, the monitoring unit 12e blinks the icon display of the span 21 of “404” or blinks the message display of “Security Abnormality” to indicate an abnormal situation to the operator. Inform. In addition, you may alert | report this situation with an audio | voice.

  In addition, the monitoring unit 12e can display the latest captured image of the span 21 in which an abnormality has occurred. As a result, the operator can immediately grasp the situation related to the span 21 where the abnormality has occurred. In addition, when it is outside the operation time of the distribution center 2 or the like, the monitoring unit 12e may notify an abnormal situation to a portable information terminal or the like carried by the operator.

  As described above, the vehicle facility monitoring system 1 can be used as a crime prevention equipment regardless of the operation time of the distribution center 2.

  As described above, the vehicle facility monitoring system 1 according to the embodiment includes the camera 24, the generation unit 12c, and the determination unit 12d. The camera 24 is provided for each span 21 (corresponding to an example of “predetermined parking area”) in the distribution center 2 (corresponding to an example of “vehicle facility”), and images the spans 21 respectively. The generation unit 12c performs machine learning using a positive captured image captured by the camera 24, thereby determining the state of the span 21 including a case where the allowable number of vehicles for each span 21 dynamically changes. Generate a model. The determination unit 12d determines the state of the span 21 based on an output value obtained by inputting a captured image, which is a determination target image captured by the camera 24, to the state determination model.

  Therefore, according to the vehicle facility monitoring system 1 according to the present embodiment, the state of the span 21 in the distribution center 2 can be accurately determined and the distribution center 2 can be operated efficiently.

  In the above-described embodiment, the case where the state of the span 21 is mainly determined by machine learning has been described as an example. However, the state of the standby area 23 may be determined as a matter of course. That is, machine learning is executed based on the captured image of the waiting area 23, and the generated state determination model, for example, indicates whether the parking spaces in the waiting area 23 are full or the size of the parked trucks T1 and T2. Alternatively, the vehicle type, number, etc. may be determined.

  In this way, by making it possible to determine the state of the waiting area 23, for example, when it is determined that a berth of a certain span 21 is vacant, the size is such that the berth can enter the berth and is waiting longer. It is possible to automatically select the tracks T1 and T2 and automatically guide the system by voice or guide light.

  In addition, for the guidance work of the trucks T1 and T2, when another vehicle guidance system is introduced and operated, the vehicle facility monitoring system 1 determines the determination result of the state of the span 21 and the waiting area 23, You may make it cooperate with this other vehicle guidance system.

  Thereby, even if a plurality of systems by a plurality of vendors are introduced and operated in the distribution center 2, the distribution center 2 is efficiently and smoothly operated based on the determination result of the vehicle facility monitoring system 1. It becomes possible.

  In the above-described embodiment, the case where the vehicle facility is the distribution center 2 has been described as an example. For example, it may be a parking lot of a large commercial facility.

  In the above-described embodiment, the server device 10 has been described as an example in which the server device 10 includes the collection unit 12a, the extraction unit 12b, the generation unit 12c, the determination unit 12d, and the monitoring unit 12e. 12a, the generation unit 12c, and the monitoring unit 12e may be configured as a dedicated device specialized for each function.

  In the above-described embodiment, deep learning is used as a machine learning algorithm, but the algorithm to be used is not limited. Therefore, machine learning may be executed by a regression analysis method such as support vector regression using a pattern discriminator such as SVM (Support Vector Machine) to generate a state discrimination model. Here, the pattern discriminator is not limited to SVM, and may be, for example, AdaBoost. A random forest or the like may be used.

  Further effects and modifications can be easily derived by those skilled in the art. Thus, the broader aspects of the present invention are not limited to the specific details and representative embodiments shown and described above. Accordingly, various modifications can be made without departing from the spirit or scope of the general inventive concept as defined by the appended claims and their equivalents.

DESCRIPTION OF SYMBOLS 1 Vehicle facility monitoring system 2 Distribution center 10 Server apparatus 12a Collection part 12c Generation part 12d Discriminating part 12e Monitoring part 13b Learning object part data set 13c 1st discrimination | determination model 13d 2nd discrimination | determination model 13e Discrimination object part data set 21 Span 22 Shutter 23 Waiting area 24 Camera

Claims (10)

A camera provided for each predetermined parking / stopping area in the vehicle facility, and imaging each of the parking / stopping areas;
The state determination model of the parking / stopping region including a case where the allowable number of vehicles for each parking / stopping region dynamically changes by executing machine learning using a positive captured image picked up by the camera A generating unit for generating
A discriminating unit for discriminating the state of the parking / stopping area based on an output value obtained by inputting the captured image that is the discrimination target imaged by the camera to the state discrimination model ;
The vehicle facility is a logistics facility,
The parking or stopping area, the vehicle facilities monitoring system, wherein the Ru holding area der track site of span or the distribution facilities is a predetermined section of the distribution facility. The discrimination unit
The vehicle facility monitoring system according to claim 1, wherein an object existing in the parking / stopping area is determined based on the output value. The discrimination unit
The vehicle facility monitoring system according to claim 2, wherein a full state of the vehicle in the parking / stopping area is determined based on the output value. The discrimination unit
The vehicle facility monitoring system according to claim 2 or 3, wherein an article other than the vehicle in the parking / stopping area is determined based on the output value. The discrimination unit
5. The vehicle facility monitoring system according to claim 2, wherein a plurality of the vehicles having different sizes parked and stopped in the parking / stopping area are determined based on the output value. A collection unit that periodically collects the captured images of the camera;
The discrimination unit
The vehicle facility monitoring system according to any one of claims 1 to 5, wherein a state of the parking / stopping area is determined based on the determination target collected by the collection unit. A monitoring unit for displaying a state of the parking / stopping area determined by the determining unit on a predetermined monitoring screen;
The monitoring unit
The vehicle facility monitoring system according to any one of claims 1 to 6, wherein at least an empty state for each span and an open / closed state of a shutter provided in the span are displayed on the monitoring screen. The monitoring unit
The vehicle facility monitoring system according to claim 7 , wherein the current captured image of the span or the waiting area by the camera is displayed on the monitoring screen. The monitoring unit
The vehicle facility monitoring system according to claim 7 or 8 , wherein security management is performed on the physical distribution facility based on an open / close state of the shutter. The monitoring unit
Claim 7, characterized in that the said determined by the determining unit to accumulate and analyze the state of the parking and stopping area to derive information about the usage of the parking and stopping area, presents information regarding said use situation, 8 Or the vehicle facility monitoring system according to 9 ;