JP7544580B2 - Passenger approach detection system and passenger approach detection method - Google Patents

Description

The present invention relates to a passenger approach detection system and a passenger approach detection method.

Conventionally, safety checks on station platforms when trains depart have generally been carried out by constructing a system that visually informs crew of the status of passengers on the platform using several cameras installed above the platform and a display device that shows images captured by these cameras.
In addition, an invention has been proposed for a system for detecting obstacles on railway vehicles, which detects obstacles that may impede operation from images obtained by an imaging unit possessed by the railway vehicle and its traveling position, and displays the detection on a display unit (see Patent Document 1).

In addition, an invention has been proposed for a system that performs object and obstacle detection from images received from at least two video devices to avoid collisions on the tracks, displays the tracks and objects on the driver's monitor, and issues an alarm to an alarm unit if an obstacle is detected (see Patent Document 2).

An invention has also been proposed that relates to a risk prediction method that uses a convolutional neural network to estimate risk areas in images captured by an on-board camera while a vehicle is moving, and outputs the risk level (see Patent Document 3).

JP 2020-1476 A Special table 2019-537534 publication JP 2017-162438 A

However, in the inventions described in Patent Documents 1, 2, and 3, detection is performed using images captured by a camera mounted on the vehicle, so when sounding an alarm using an alarm device in the station, a communication connection must be established between the device in the vehicle and the device in the station, making system construction complicated.
Furthermore, the inventions described in Patent Documents 1, 2, and 3 all aim to detect obstacles in general-purpose means of transportation, and there is a risk that approaching passengers may not be detected efficiently in a crowded station platform environment. In particular, when a train departs from a station platform while it is crowded with passengers getting on and off after arriving at the station platform, there is a risk that passengers may approach and come into contact with the departing train on a narrow platform without platform doors, and it is desirable to reliably detect passengers who are at risk of coming into contact with the departing train.

The present invention was made with a focus on the above-mentioned problems, and aims to provide a passenger approach detection system and a passenger approach detection method that are relatively easy to construct and can efficiently detect approaching passengers to a train when the train departs and the station platform is crowded.

In order to achieve the above object, the present invention provides
an imaging unit that is installed so as to be able to capture images in a direction substantially parallel to the direction of travel of the train and that continuously captures images of the train and a predetermined range of the edge of the platform;
a determination unit that analyzes the images of the train continuously acquired by the imaging unit and determines whether the train has departed from a platform and started moving;
a detection unit that detects the presence of an approaching passenger located within a predetermined range of the edge of the platform from images of the edge of the platform within the predetermined range acquired by the imaging unit after the judgment unit has judged that the train has departed from the platform and started moving;
Equipped with
The determination unit determines the movement of the train by analyzing images of the roof of the train .

With a passenger approach detection system configured as described above, when it is determined that a train has left the platform and started moving, the presence of approaching passengers within a specified range of the edge of the platform is detected. This makes it possible to eliminate detection in areas other than the specified range of the edge of the platform and at times other than when the train is departing, thereby enabling efficient detection of approaching passengers to the train.
In addition, because station platforms are less likely to become crowded at times other than when trains are departing, the occurrence of false positives can be further reduced by eliminating detections other than when unnecessary trains are departing, preventing disruptions to train operations and ensuring stable transportation.
In addition, since the passenger approach detection system analyzes images of the train roof acquired by the imaging unit to determine whether the train has left the platform and started moving, it does not require connection to equipment inside the train, making it possible to build the system relatively easily.

Here, preferably, the detection unit detects the presence of an approaching passenger located within a specified range of the edge of the platform from the time when the judgment unit determines that the train has left the platform and started moving until the train deviates from a specified range in the direction of train travel.
With this configuration, it is possible to eliminate detection during any time other than the period from the time the train departs until the train departs from a specified range in the direction of train travel, thereby making it possible to more efficiently detect the approach of passengers to the train.

Furthermore, preferably, the detection unit detects the presence of an approaching passenger located within a predetermined range of the edge of the platform by image analysis using machine learning.
With this configuration, it is possible to more accurately detect the approach of passengers to a train within a specified range at the edge of the platform when a train departs and the station platform is crowded.

Furthermore, it is preferable that the vehicle has a notification unit that notifies the passenger by controlling a predetermined device when the detection unit detects an approaching passenger.
With this configuration, a notified staff member or other person can stop the train from moving by pressing the emergency stop button, thereby preventing approaching passengers from coming into contact with the moving train.

Also, preferably, the specified device is any one of a special signal light emitting device, a departure signal, a display device carried by a train crew member or station staff, and a train emergency stop device.
With this configuration, a notified staff member or other person can stop the train from moving by pressing the emergency stop button, thereby preventing approaching passengers from coming into contact with the moving train.

Furthermore, preferably, the imaging unit is installed at a position at a predetermined height from the edge of the platform.
With this configuration, it is possible to clearly detect approaching passengers to the train within a specified range at the edge of the platform, thereby ensuring safe transportation.

The passenger approach detection system and passenger approach detection method of the present invention are relatively easy to construct, and can efficiently detect approaching passengers to a train when the train is departing from a crowded station platform.

1 is a block diagram showing an embodiment of a passenger approach detection system according to the present invention; 1 is a diagram showing an example of a method of installing an imaging unit for acquiring image data used in a passenger approach detection system according to the present invention. FIG. 13 is a diagram showing an example of setting of a departure detection area in an image acquired by an imaging unit of an embodiment, and an example of train departure judgment by a judgment unit (before the train arrives). FIG. 13 is a diagram showing an example of setting a departure detection area in an image acquired by an imaging unit of an embodiment, and a train departure judgment by a judgment unit (when a train arrives). FIG. 13 is a diagram showing the setting of a departure detection area in an image acquired by an imaging unit of the embodiment, and an example of train departure judgment by a judgment unit (when passengers board and disembark). FIG. 13 is a diagram showing an example of setting of a departure detection area in an image acquired by an imaging unit of an embodiment, and an example of train departure judgment by a judgment unit (when a train departs). FIG. 13 is a diagram showing an example of setting of a departure detection area in an image acquired by an imaging unit of an embodiment, and an example of train departure judgment by a judgment unit (after the train departs). FIG. 11 is a flowchart showing a departure determination process according to an embodiment. 11 is a flowchart showing an approaching passenger detection process according to an embodiment.

Below, a preferred embodiment of the passenger approach detection system 100 according to the present invention will be described in detail with reference to the drawings.

FIG. 1 shows a block diagram of one embodiment of a passenger approach detection system 100 according to the present invention.
1, the passenger approach detection system 100 includes an imaging device 10, a management device 20, and an alarm device 30. The management device 20 is capable of communicating with the imaging device 10 and the alarm device 30 via a communication network.

The imaging device 10 includes an imaging unit 11 and a communication unit 12 .
The imaging unit 11 is preferably a video camera capable of capturing moving images, but may also be a still camera capable of continuously capturing a plurality of still images in a short period of time.
The communication unit 12 is used for communication with the management device 20. For example, the communication unit 12 is a communication interface having a communication IC (Integrated Circuit) and a communication connector, and performs data communication via a communication network. The communication unit 12 also outputs image data acquired by the imaging unit 11 to the management device 20.

2 shows an example of how to install the imaging device 10 for acquiring image data used in the passenger approach detection system 100 according to the present invention. In FIG. 2, a train T is shown stopped at a platform H.
The imaging device 10 is installed on a cable rack or the like provided near the ceiling of the platform H, and is located at a position where it can capture an image in a direction approximately parallel to the traveling direction of the train T. In addition, the imaging devices 10 are installed, for example, two units per car of the train T at intervals.
The height from the platform H of the position where the imaging device 10 is installed is height h (e.g., 3 m or more) shown in FIG. 2. The height h is not limited to 3 m, and may be any height that allows a clear image of the roof part TR of the train T and the platform edge part Ha of the platform H. The imaging range of the imaging device 10 includes at least a part of the train T (e.g., the roof part TR) and a predetermined range w that is a predetermined distance inside the platform H from the platform edge part Ha shown in FIG. 2, and in this embodiment, the predetermined range w is, for example, a range of 250 mm. By installing the imaging device 10 at this position, the roof part TR of the train T and the platform edge part Ha can be clearly imaged from directly above the platform edge part Ha.

1 , the management device 20 is installed, for example, in a station office, and includes a control unit 21, a storage unit 22, and a communication unit 23. The management device 20 may also be installed in a central command room, a security guard station, or the like.
The control unit 21 is a part that controls the operation of the management device 20, and is configured with, for example, a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), etc., and controls each part of the management device 20 through cooperation between the program data stored in the memory unit 22 and the CPU.
The control unit 21 also includes a determination unit 211 , a detection unit 212 , and a notification unit 213 .

The determination unit 211 analyzes the image of the roof portion TR of the train T acquired by the imaging device 10 to determine whether the train T has departed from the platform H and started moving.

When the judgment unit 211 determines that the train T has departed from the platform H and started moving, the detection unit 212 analyzes images of a specified range w of the platform edge Ha subsequently acquired by the imaging device 10, and detects the presence of an approaching passenger located within the specified range w of the platform edge Ha.
Furthermore, the detection unit 212 detects the presence of an approaching passenger by image analysis using machine learning. Since various algorithms and deep neural networks (DNNs) using artificial intelligence have already been developed for detecting the presence of an approaching passenger, such publicly known technologies can be used.

When the detection unit 212 detects the approach of a passenger, the notification unit 213 notifies staff, the driver, the approaching passenger, or passengers present on platform H, etc., via the communication unit 23 by controlling the alarm device 30, which is a specified device.

The memory unit 22 is a part that stores various information required for the operation of the management device 20, and is composed of, for example, an HDD (Hard Disk Drive), semiconductor memory, etc., and stores data required for the operation of the management device 20, such as program data, in a manner that can be read and written by the control unit 21.

The communication unit 23 is a part used for communication between the imaging device 10 and the notification device 30. For example, it is a communication interface having a communication IC and a communication connector, and performs data communication via a communication network under the control of the control unit 21.

The notification device 30 is a special signal light emitting device, a departure signal, a display device carried by a crew member or station staff, a train emergency stop device, etc.
The notification device 30 includes a notification unit 31 and a communication unit 32 .
The notification unit 31 is a light emitting unit of the special signal light emitting device and the departure signal, a display unit of the display device, and an audio unit of the train emergency stop device. The notification unit 31 also issues a detection alert when an approaching passenger is detected by the detection unit 212. Upon receiving the detection alert from the notification unit 31, a staff member stops the train by pressing the emergency stop button, thereby preventing the approaching passenger from coming into contact with the running train.
The communication unit 32 is a part used for communication with the management device 20. For example, it is a communication interface having a communication IC and a communication connector, and performs data communication via a communication network. Furthermore, when an approaching passenger is detected by the detection unit 212, the communication unit 32 receives an instruction to issue a detection alert from the notification unit 213.

3A to 3E are diagrams showing an example of setting a departure detection area A in an image acquired by the imaging device 10 and judging the departure of a train T by the judging unit 211. FIG.
3A is a diagram showing a case where a train T is not present on platform H. A rectangular area provided on the tracks is a departure detection area A. The rectangular area provided on platform H is a predetermined range w that is a predetermined distance inside platform H from the platform edge Ha.
FIG. 3B is a diagram showing a case where a train T has arrived at a station and a roof portion TR of the train T has entered a departure detection area A.
3C is a diagram showing a case where the train T stops at the platform H and passengers get on and off the train T. The roof portion TR of the train T is within the departure detection area A.
3D is a diagram showing a case where the train T departs from the platform H. The determination unit 211 determines that the train T has departed by performing image analysis of the roof portion TR of the train T within the departure detection area A in the image data acquired by the imaging device 10.
Fig. 3E is a diagram showing the state after the train T has departed from the station. In this case, the roof part TR of the train T has departed from the departure detection area A. The determination unit 211 performs a process of detecting the presence of an approaching passenger by the detection unit 212 during the period from the time when it is determined that the train T shown in Fig. 3D has departed from the platform H and started moving until the roof part TR of the train T departs from the departure detection area A in the train travel direction, and determines whether or not there is an approaching passenger.

The operation of the passenger approach detection system 100 according to this embodiment will be described below with reference to the flowchart of the departure decision process in FIG. 4 and the flowchart of the approaching passenger detection process in FIG. 5.

In the departure determination process, first, the determination unit 211 receives image data including the departure detection area A from the imaging device 10 via the communication unit 23 (step S1).
Next, the determination unit 211 performs image analysis of the departure detection area A in the image data (step S2).
Next, the determination unit 211 determines whether or not the roof portion TR of the train T is within the departure detection area A (step S3).
If the roof portion TR is in the departure detection area A (step S3; YES), the judgment unit 211 judges whether the roof portion TR in the departure detection area A is moving, that is, whether the train T is moving (step S4).
If the train T is moving (step S4; YES), the judgment unit 211 judges whether the train T is departing or not (step S5).
If the train T is departing (step S5; YES), the determination unit 211 starts the approaching passenger detection process shown in FIG. 5 (step S6).

Next, in the approaching passenger detection process, the detection unit 212 performs image analysis of a predetermined range w in the image data, a predetermined distance inside the platform H from the platform edge Ha (step S61).
Next, the detection unit 212 determines whether or not a passenger is present within a predetermined range w of the platform edge Ha image-analyzed in step S61 (step S62).
When a passenger is present within a predetermined range w of the platform edge Ha (step S62; YES), the detection unit 212 determines whether or not a detection alert has been issued (step S63).
If a detection alert has not yet been issued (step S63; NO), the notification unit 213 sends an instruction to the alarm device 30 via the communication unit 23 to issue a detection alert, and the alarm unit 31 of the alarm device 30 issues a detection alert (step S64), terminates the processing, and returns to the departure judgment processing shown in Figure 4.
Moreover, if a detection alert has already been issued (step S63; YES), the detection unit 212 ends the process and returns to the departure determination process shown in FIG.

Furthermore, if there is no passenger within the predetermined range w of the platform edge Ha (step S62; NO), the detection unit 212 determines whether or not a detection alert has been issued (step S65).
If a detection alert has already been issued (step S65; YES), the notification unit 213 sends an instruction to stop the detection alert to the alarm device 30 via the communication unit 23, and the alarm unit 31 of the alarm device 30 stops the detection alert (step S66).
Thereafter, the detection unit 212 ends the process and returns to the departure determination process shown in FIG.
Also, if a detection alert has not been issued (step S65; NO), the detection unit 212 ends the process and returns to the departure determination process shown in FIG.

Next, the determination unit 211 receives the next image data including the departure detection area A from the imaging device 10 via the communication unit 23 (step S7).
Next, the determination unit 211 performs image analysis of the departure detection area A in the next image data (step S8).
Next, the determination unit 211 determines whether or not the roof portion TR of the train T is within the departure detection area A in the next image data (step S9).
If the roof portion TR is within the departure detection area A (step S9; YES), that is, if the train T is still departing, the judgment unit 211 returns to step S5.
In addition, if the roof portion TR is not within the departure detection area A (step S3; NO), if the train T is not moving (step S4; NO), if the train T is not departing (step S5; NO), and if the roof portion TR is not within the departure detection area A (step S9; NO), in other words, if the roof portion TR has deviated from the departure detection area A, the judgment unit 211 judges that it is a time period in which it is not necessary to detect the presence of approaching passengers (step S10) and returns to step S1.

The above departure judgment process is performed continuously until the control unit 21 of the management device 20 receives an instruction to end the process.

As described above, the passenger approach detection system 100 according to the above embodiment comprises an imaging unit 11 that is installed so as to be able to capture images in a direction approximately parallel to the direction of travel of the train T and that continuously captures images of the train T and a specified range w of the platform edge Ha, a judgment unit 211 that analyzes the images of the train T continuously captured by the imaging unit 11 and determines whether the train T has departed the platform H and started moving, and a detection unit 212 that, when it is determined by the judgment unit 211 that the train T has departed the platform H and started moving, detects the presence of an approaching passenger located within the specified range w of the platform edge Ha from images of the specified range w of the platform edge Ha subsequently captured by the imaging unit 11.
Therefore, when it is determined that a train has left the platform and started moving, the presence of approaching passengers within a specified range of the edge of the platform is detected, which makes it possible to eliminate detection in areas other than the specified range of the edge of the platform and at times other than when the train is departing, thereby enabling efficient detection of approaching passengers to the train.
In addition, by eliminating detections other than those at the time of unnecessary train departures, the occurrence of false positives can be further reduced, preventing disruptions to train operations and ensuring stable transportation.
In addition, since the passenger approach detection system analyzes the images of the train acquired by the imaging unit to determine whether the train has left the platform and started moving, it does not require connection to devices inside the train, and the system can be constructed relatively easily.

In addition, the detection unit 212 detects the presence of approaching passengers located within a specified range w of the platform edge Ha from the time when the judgment unit 211 determines that the train T has departed from the platform H and started moving until the train T departs from a specified range in the direction of train T's travel.
Therefore, false detections can be eliminated during the time period other than the time from when the train departs until the train deviates from a specified range in the train's direction of travel, and the approach of passengers to the train can be detected more efficiently.

In addition, the detection unit 212 detects the presence of an approaching passenger located within a predetermined range w of the platform edge Ha through image analysis using machine learning.
Therefore, it is possible to accurately detect the approach of passengers to a train within a specified range at the edge of the platform when a train departs and the station platform is crowded.

In addition, the judgment unit 211 judges the movement of the train T by image analysis of the roof portion TR of the train T.
Therefore, since the passenger approach detection system analyzes the image of the train roof acquired by the imaging unit to determine whether the train has left the platform and started moving, it does not require connection to devices inside the train, etc., and the system can be constructed relatively easily.

The device also includes a notification unit 213 that notifies the user by controlling a specific device when the detection unit 212 detects an approaching passenger.
Therefore, a staff member who receives the notification can stop the train from moving by pressing the emergency stop button, thereby preventing approaching passengers from coming into contact with the moving train.

The specified device may be a special signal light emitting device, a departure signal, a display device carried by a train crew member or station staff member, or an emergency train stop device.
Therefore, a staff member who receives the notification can stop the train from moving by pressing the emergency stop button, thereby preventing approaching passengers from coming into contact with the moving train.

The imaging unit 11 is installed at a predetermined height h from the platform edge Ha.
Therefore, it is possible to clearly detect the approach of passengers within a specified range of the edge of the platform, ensuring safe transportation.

Although the preferred embodiment of the present invention has been described above, the present invention is not limited to the above embodiment. For example, the imaging devices 10 are installed at intervals of two per car of the train T, but this is not limited to this. It is also possible to install imaging devices 10 that can capture clear images of the roof portions TR and platform edge portions Ha of all cars of the train T. In this case, it is sufficient to install one imaging device 10.

In the above embodiment, the width of the predetermined range w of the platform edge Ha is 250 mm, but this is not limited to this. It can be any distance that is dangerous for passengers to approach the train T.

REFERENCE SIGNS LIST 10 Imaging device 11 Imaging section 12 Communication section 20 Management device 21 Control section 211 Determination section 212 Detection section 213 Notification section 22 Storage section 23 Communication section 30 Notification device 31 Notification section 32 Communication section 100 Passenger approach detection system H Platform Ha Platform edge T Train TR Roof section

Claims (8)

an imaging unit that is installed so as to be able to capture images in a direction substantially parallel to the direction of travel of the train and that continuously captures images of the train and a predetermined range of the edge of the platform;
a determination unit that analyzes the images of the train continuously acquired by the imaging unit and determines whether the train has departed from a platform and started moving;
a detection unit that detects the presence of an approaching passenger located within a predetermined range of the edge of the platform from images of the edge of the platform within the predetermined range acquired by the imaging unit after the judgment unit has judged that the train has departed from the platform and started moving;
Equipped with
A passenger approach detection system characterized in that the judgment unit judges the movement of the train by image analysis of the roof of the train . The passenger approach detection system according to claim 1, characterized in that the detection unit detects the presence of the approaching passenger located within a predetermined range of the edge of the platform from the time when the determination unit determines that the train has departed the platform and started moving until the train departs from a predetermined range in the train's travel direction. The passenger approach detection system according to claim 1 or 2, characterized in that the detection unit detects the presence of the approaching passenger located within a predetermined range of the edge of the platform by image analysis using machine learning. 4. The passenger approach detection system according to claim 1, further comprising a notification unit that notifies the user by controlling a predetermined device when the approaching passenger is detected by the detection unit. The passenger approach detection system according to claim 4 , characterized in that the specified device is any one of a special signal light emitting device, a departure signal, a display device carried by a crew member or station staff, and a train emergency stop device. The passenger approach detection system according to any one of claims 1 to 5 , wherein the imaging unit is installed at a position of a predetermined height from the edge of the platform. A passenger approach detection method performed by a passenger approach detection system, comprising:
an acquiring step of acquiring images of a predetermined range of the train and the edge of the platform, the image capturing step being performed by installing the camera so as to be capable of capturing images in a direction substantially parallel to the direction of travel of the train;
a determination step of analyzing the images of the train continuously acquired by the acquisition step to determine whether the train has departed from a platform and started moving;
a detection step of detecting the presence of an approaching passenger located within a predetermined range of the edge of the platform from images of the edge of the platform within the predetermined range acquired in the acquisition step when the determination step determines that the train has departed from the platform and started moving;
Including,
A passenger approach detection method, characterized in that in the judgment step, the movement of the train is judged by image analysis of the roof of the train . The passenger approach detection method described in claim 7, characterized in that in the detection process, the presence of the approaching passenger within a specified range of the edge of the platform is detected from the time when it is determined in the judgment process that the train has left the platform and started moving until the train deviates from a specified range in the direction of train travel.

Images