JP7793902B2 - Train accident avoidance system, train accident avoidance method, and program - Google Patents

Description

The present invention relates to a train accident avoidance system, a train accident avoidance method, and a program for avoiding train accidents.

When a danger occurs on the platform, a passenger or station staff member who sees it presses the emergency stop button, bringing the train to an emergency stop and attempting to avoid a train accident. In this case, because a person is involved from the time the danger is confirmed to the time the emergency stop button is pressed, there is no immediacy, and there is the issue of the danger being overlooked. Additionally, because the emergency stop is only made after the danger is confirmed, there is also the issue of delayed initial action to avoid a train accident.

In response to this, a train accident avoidance system is known that uses a camera to capture images of the tracks and platform, detects incidents such as a person falling from the platform based on the captured images, and stops the train based on the detection results (see, for example, Patent Documents 1 and 2).

International Publication No. 2020/158197 Japanese Patent Application Laid-Open No. 2020-135243 JP 2012-056535 A

However, there is a demand for a more reliable way to prevent train accidents by detecting signs of train accidents on platforms and taking early action.

The purpose of this disclosure is to provide a train accident avoidance system, a train accident avoidance method, and a program that solve the above-mentioned problems.

One aspect of the present invention to achieve the above object is to
an imaging means for imaging the track on which the train runs and the platform where people get on and off the train;
a communication means for transmitting the images of the tracks and platforms photographed by the photographing means by a wireless communication system;
a sign detection means for learning images showing signs of train accidents on the platform that may be the cause of the train accident, and for detecting signs of train accidents on the platform based on the learning results and the images of the platform transmitted by the communication means;
an object detection means for learning images of objects that may cause train accidents on the tracks, and detecting objects that may cause train accidents on the tracks based on the learning results and the images of the tracks transmitted by the communication means;
an instruction means for issuing a running instruction to the train,
The instruction means
When the sign of a train accident on the platform is detected by the sign detection means, a deceleration command is issued to trains that are located within a predetermined distance from the platform and approaching the platform,
When an object on the tracks is detected by the object detection means, a stop command is issued to a train that is located within a predetermined distance from the platform and is approaching the platform.
It is a train accident avoidance system.
One aspect of the present invention to achieve the above object is to
taking photographs of the tracks on which the train runs and the platforms where people get on and off the train;
transmitting the captured images of the tracks and platforms by wireless communication;
a step of learning images showing signs of a train accident on the platform that may cause the train accident, and detecting signs of a train accident on the platform based on the learning results and the transmitted images of the platform;
learning images of objects that may cause train accidents on the tracks, and detecting objects that may cause train accidents on the tracks based on the learning results and the transmitted images of the tracks;
a step of issuing a deceleration command to a train that is located within a predetermined distance from the platform and is approaching the platform when a sign of a train accident on the platform is detected;
a step of issuing a stop instruction to a train that is located within a predetermined distance from the platform and is approaching the platform when an object on the track is detected;
The present invention may be a train accident avoidance method including the steps of:
One aspect of the present invention to achieve the above object is to
A process of learning images showing signs of train accidents on platforms that are the cause of train accidents, and detecting signs of train accidents on platforms based on the learning results and photographed images of the platforms;
A process of learning images of objects that may cause train accidents on the tracks, and detecting objects that may cause train accidents on the tracks based on the learning results and photographed images of the tracks;
When a sign of a train accident on the platform is detected, a process of issuing a deceleration command to trains that are located within a predetermined distance from the platform and approaching the platform;
When an object on the track is detected, a process of issuing a stop instruction to a train that is located within a predetermined distance from the platform and is approaching the platform;
The program may be a program that causes a computer to execute the above.

This disclosure provides a train accident avoidance system, train accident avoidance method, and program that solve the above-mentioned problems.

1 is a block diagram showing a schematic system configuration of a train accident avoidance system according to an embodiment of the present invention. A block diagram showing a schematic system configuration of an AI server and a control server according to this embodiment. FIG. 10 is a diagram showing a configuration for transmitting an image of the platform that is the source of a deceleration instruction to the train management unit. FIG. 10 is a diagram showing an example of a flow for releasing a train from a stopped state. 3 is a flowchart showing the flow of a train accident avoidance method according to the present embodiment. 1 is a block diagram showing a schematic system configuration of a train accident avoidance system according to an embodiment of the present invention. 1 is a block diagram showing a schematic system configuration of a train accident avoidance system according to an embodiment of the present invention.

Embodiment 1
Hereinafter, an embodiment of the present invention will be described with reference to the drawings. Fig. 1 is a block diagram showing a schematic system configuration of a train accident avoidance system according to this embodiment. The train accident avoidance system 1 according to this embodiment includes a plurality of cameras 2, a wireless communication network 3, an AI server 4, and a control server 5.

Camera 2 is a specific example of a photographing means. Camera 2 is installed on the tracks along which train 6 travels, on the platform where people board and disembark the train, etc. Camera 2 photographs the tracks and the platform. Camera 2 is configured as a camera 2 that supports ultra-high image quality of, for example, 4K (3840 x 2160) or 8K (7680 x 4320) pixels. Camera 2 may also be configured as a compound eye surveillance camera.

Camera 2 transmits the captured images of the tracks and platforms to AI server 4 via wireless communication network 3.

The wireless communication network 3 is a specific example of a communication means. The wireless communication network 3 connects, for example, multiple cameras 2, an AI server 4, a control server 5, and a train 6 via a high-speed wireless communication method such as 5G (5th Generation Mobile Communication System) wireless communication or 6G wireless communication. This allows high-quality, large-volume image data to be transmitted from the cameras 2 to the AI server 4 at high speed.

The wireless communication network 3 includes wireless base stations, networks, etc. The AI server 4 and control server 5 may be integrated into a single server.

The AI server 4 and control server 5 have a typical computer hardware configuration, including, for example, a processor 4a, 5a such as a CPU (Central Processing Unit) or GPU (Graphics Processing Unit), internal memory 4b, 5b such as RAM (Random Access Memory) or ROM (Read Only Memory), storage devices 4c, 5c such as HDD (Hard Disk Drive) or SSD (Solid State Drive), input/output I/Fs 4d, 5d for connecting peripheral devices such as displays, and communication I/Fs 4e, 5e for communicating with devices external to the device.

Figure 2 is a block diagram showing the general system configuration of the AI server and control server according to this embodiment. The AI server 4 has a sign detection unit 41 and an object detection unit 42. Note that the sign detection unit 41 and the object detection unit 42 may be configured as separate AI servers. Alternatively, the sign detection unit 41 and the object detection unit 42 may be configured as an integrated unit.

The sign detection unit 41 is a specific example of a sign detection means. The sign detection unit 41 detects signs of a train accident on the platform. A sign of a train accident on the platform is, for example, a sign that there is a possibility that a person or object on the platform may fall onto the tracks and cause a collision with a train 6 traveling on the tracks in the future.

The sign detection unit 41 includes a learning device. The learning device may perform deep learning. The learning device is configured, for example, with a neural network such as a recurrent neural network (RNN). This RNN may have a long short-term memory (LSTM) in the intermediate layer. Instead of a neural network, the learning device may be configured with another learning device such as a support vector machine (SVM).

The learning device of the sign detection unit 41 learns images that indicate signs of a train accident on a platform that could cause a train accident. The images include still images and videos. Images that indicate signs of a train accident on a platform that could cause a train accident include, for example, images of the facial expression or movement of a person on the platform who is thought to be about to fall onto the tracks, or images of the movement of an object on the platform.

The sign detection unit 41 detects signs of a train accident on the platform based on the learning results from the learning device and images of the platform transmitted from the camera 2 via the wireless communication network 3. The sign detection unit 41 inputs the images of the platform transmitted from the camera 2 via the wireless communication network 3 into the trained learning device. The sign detection unit 41 detects signs of a train accident on the platform based on the output values output from the learning device.

When the sign detection unit 41 detects a sign of a train accident on the platform, it sends a sign detection signal indicating the detection result to the control server 5.

The object detection unit 42 is a specific example of an object detection means. The object detection unit 42 detects objects that could cause a train accident on the tracks. Objects that could cause a train accident on the tracks include, for example, people or objects that fall from a platform or a train. The object detection unit 42 includes a learning device. The learning device of the object detection unit 42 learns images that show objects that could cause a train accident on the tracks. The images include still images and videos.

The object detection unit 42 detects objects on the tracks that could cause a train accident based on the learning results from the learning device and images of the tracks transmitted from the camera 2 via the wireless communication network 3. The object detection unit 42 inputs the images of the tracks transmitted from the camera 2 via the wireless communication network 3 into the trained learning device. The object detection unit 42 detects objects on the tracks that could cause a train accident based on the output values output from the learning device.

When the object detection unit 42 detects an object on the tracks that could cause a train accident, it sends an object detection signal indicating the detection result to the control server 5.

The control server 5 is a specific example of an instruction means. The control server 5 has a train instruction unit 51. The train instruction unit 51 issues running instructions to the running control unit 61 of the train 6 running on the tracks. The train instruction unit 51 transmits a deceleration instruction signal to decelerate the train 6 and a stop instruction signal to stop the train 6 to the running control unit 61 of the train 6 via the wireless communication network 3.

When the train 6's running control unit 61 receives a deceleration instruction signal from the train instruction unit 51, it controls the braking unit 62 and other units to decelerate the train 6. Similarly, when the train 6's running control unit 61 receives a stop instruction signal from the train instruction unit 51, it controls the braking unit 62 and other units to stop the train 6.

The train control unit 51 issues deceleration and stop instructions to trains 6 that are located within a predetermined distance from the platform and approaching the platform. The running control unit 61 of each train 6 detects the current position of the train 6, for example, based on signals from GPS satellites, and transmits this current position to the train control unit 51. Based on the current position of each train transmitted from the running control unit 61 of each train 6, the train control unit 51 can determine whether the train is located within a predetermined distance from the platform and is approaching the platform.

However, there is a demand for more reliable prevention of train accidents by detecting signs of train accidents on platforms and taking early action.

In contrast, the train accident avoidance system 1 according to this embodiment issues a deceleration command to trains 6 that are located within a predetermined distance from the platform and approaching the platform when the sign detection unit 41 detects a sign of a train accident on the platform.

For example, when the train control unit 51 receives a sign detection signal from the sign detection unit 41, it transmits a deceleration command signal to the running control unit 61 of the train 6 that is located within a predetermined distance from the platform and is approaching the platform, based on the current position transmitted from the running control unit 61 of each train 6. The running control unit 61 of the train 6 decelerates the train 6 in response to the deceleration command signal transmitted from the train control unit 51.

This means that if signs of a train accident on the platform are detected, train 6 approaching the platform can be slowed down early and countermeasures can be taken, making it possible to more reliably avoid train accidents.

Furthermore, when an object on the tracks is detected by the object detection unit 42, the train accident avoidance system according to this embodiment issues a stop command to trains 6 that are located within a predetermined distance from the platform and approaching the platform.

For example, when the train control unit 51 receives an object detection signal from the object detection unit 42, it transmits a stop instruction signal to the running control unit 61 of a train 6 that is located within a predetermined distance from the platform and is approaching the platform, based on the current position transmitted from the running control unit 61 of each train 6. The running control unit 61 of the train 6 stops the train 6 in response to the stop instruction signal transmitted from the train control unit 51.

As a result, if an object on the tracks is detected, the train 6 approaching the platform will be immediately stopped, preventing a collision between the train 6 and the object on the tracks and more reliably avoiding train accidents.

Here, for example, as mentioned above, once Train 6 has made an emergency stop, there is the issue that it takes a considerable amount of time for it to resume operation even if there are no problems, due to on-site inspections by station staff.

In contrast to this, in this embodiment, as described above, when the train instruction unit 51 of the control server 5 issues a deceleration instruction to the running control unit 61 of the train 6, it may send an image of the platform that is the basis for the deceleration instruction to the management unit 7 that manages the train 6 (Figure 3). Also, when the train instruction unit 51 of the control server 5 issues a stop instruction to the running control unit 61 of the train 6, it may send an image of the tracks that is the basis for the stop instruction to the management unit 7 that manages the train 6. The management unit 7 that manages the train is implemented, for example, on a terminal of a station employee who manages the train 6 or a terminal in a control room that manages the train 6.

As a result, after issuing an instruction to slow down or stop the train 6, station staff or others can check the images of the cause of the instruction to slow down or stop on a terminal equipped with the management unit 7, allowing them to quickly identify any problems and shorten the time it takes to resume operations if there are no problems.

For example, as shown in Figure 4, the train instruction unit 51 of the control server 5 sends a stop instruction signal to the train 6's running control unit 61 via the wireless communication network 3, and also sends an image of the tracks that is the source of the stop instruction to the station staff terminal (step S401).

The station attendant checks the image of the tracks on the terminal (step S402), and if they determine that there are no problems, performs an operation on the terminal to cancel the stop instruction (step S403). In response to the cancellation operation, the terminal transmits a cancellation signal to the train control unit 51 via the wireless communication network 3 to cancel the stop of train 6.

In response to the release signal sent from the terminal, the train control unit 51 transmits a running start signal to the train 6's running control unit 61 via the wireless communication network 3 to cause the train to start running (step S404). The train 6's running control unit 61 starts running in response to the running start signal from the train control unit 51 (step S405). In this way, station staff and others can check the image that caused the stop command on their terminal, quickly identifying the problem situation, and can immediately resume operation if there is no problem.

Next, the train accident avoidance method according to this embodiment will be explained in detail. Figure 5 is a flowchart showing the flow of the train accident avoidance method according to this embodiment.

The camera 2 captures images of the tracks and platform, and transmits the captured images of the tracks and platform to the sign detection unit 41 and object detection unit 42 of the AI server 4 via the wireless communication network 3 (step S501).

The sign detection unit 41 detects signs of a train accident on the platform based on the learning results of the learning device and the images of the platform transmitted from the camera 2, and transmits a sign detection signal to the train control unit 51 (step S502).

In response to the sign detection signal transmitted from the sign detection unit 41, the train control unit 51 transmits a deceleration instruction signal to the running control unit 61 of the train 6 that is located within a predetermined distance from the platform and is approaching the platform (step S503).

The train 6's running control unit 61 decelerates the train 6 in response to the deceleration instruction signal from the train instruction unit 51 (step S504).

The object detection unit 42 detects an object on the tracks that could cause a train accident based on the learning results of the learning device and the image of the tracks transmitted from the camera 2, and transmits an object detection signal to the train control unit 51 (step S505).

In response to the object detection signal transmitted from the object detection unit 42, the train control unit 51 transmits a stop instruction signal to the running control unit 61 of the train 6 that is located within a predetermined distance from the platform and is approaching the platform (step S506).

The train 6 running control unit 61 stops the train 6 in response to the stop instruction signal from the train instruction unit 51 (step S507).

As described above, in the train accident avoidance system according to this embodiment, when the sign detection unit 41 detects a sign of a train accident on the platform, the train instruction unit 51 instructs trains 6 that are located within a predetermined distance from the platform and are approaching the platform to decelerate. As a result, when a sign of a train accident on the platform is detected, train accidents can be avoided more reliably by quickly slowing down the train 6 approaching the platform and taking measures to prevent it.

In addition, when an object on the tracks is detected by the object detection unit 42, the train instruction unit 51 issues a stop instruction to trains 6 that are located within a predetermined distance from the platform and are approaching the platform. As a result, when an object on the tracks is detected, the train 6 approaching the platform can be immediately stopped, preventing a collision between the train 6 and the object on the tracks and more reliably avoiding train accidents.

Embodiment 2
6 is a block diagram showing a schematic system configuration of the train accident avoidance system according to this embodiment. In this embodiment, the camera 20 may be provided movably. For example, a rail for moving the camera is laid on the roof of the platform or the like. The camera 20 moves along the rail by driving an actuator 21 such as a motor.

For example, construction equipment or vending machines may be temporarily placed or moved on the platform, causing obstructions and creating blind spots in the camera's field of view.

In response to this, the control server 50 may control the movement of the camera 20 so as to reduce blind spots in the field of view of the camera 20. This reduces blind spots in the field of view of the camera 20, making it possible to more reliably avoid train accidents.

The control server 50 according to this embodiment further includes a camera instruction unit 52 that instructs the movement of the camera 20. The camera instruction unit 52 controls the movement of the camera 20 by sending a control signal to the camera 20.

The AI server 4 may determine whether a blind spot has occurred in the field of view of the camera 20, for example, based on images of the platform and tracks captured by the camera 20. More specifically, the AI server 4 may determine that a blind spot has occurred in the field of view of the camera 20 when it detects construction equipment or a vending machine that is not normally installed on the platform. When the AI server 4 determines that a blind spot has occurred in the field of view of the camera 20, it transmits a blind spot occurrence signal indicating the determination result to the camera instruction unit 52 of the control server 50.

In response to the blind spot occurrence signal transmitted from the AI server 4, the camera instruction unit 52 transmits a control signal to the camera 20 in which the blind spot occurs, causing the camera 20 to move. The camera instruction unit 52 may, for example, move the camera 20 by a predetermined amount in a predetermined direction.

Embodiment 3
FIG. 7 is a block diagram showing a schematic system configuration of the train accident avoidance system according to this embodiment.

In this embodiment, the control server 60 controls the lighting of the platform illuminators 8 installed near objects that indicate signs of a train accident detected by the sign detection unit 41 and objects on the tracks that could cause a train accident detected by the object detection unit 42. By controlling the lighting of the illuminators 8, station staff and others can easily identify dangerous areas on the platform and tracks that could cause a train accident. In addition, by controlling the lighting of the illuminators 8, it is possible to alert people who are in danger and could cause a train accident.

The control server 60 according to this embodiment further includes a lighting instruction unit 53 that issues instructions to the illuminator 8. The lighting instruction unit 53 controls the lighting of the illuminator 8 by sending control signals to the illuminator 8. The lighting instruction unit 53 changes the lighting of the illuminator 8, for example, by turning it on, off, or blinking, changing the color of the lighting, or increasing or decreasing the illuminance.

The lighting instruction unit 53 may change the lighting of the illuminator 8 that is closest to an object that indicates a sign of a train accident detected by the sign detection unit 41, and an object that may cause a train accident on the tracks detected by the object detection unit 42.

While several embodiments of the present invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These novel embodiments may be embodied in a variety of other forms, and various omissions, substitutions, and modifications may be made without departing from the spirit of the invention. These embodiments and their variations are within the scope and spirit of the invention, and are also included in the scope of the invention and its equivalents as set forth in the claims.

The present invention can also be implemented, for example, by having a processor execute a computer program to perform the processing shown in Figure 5.

The program includes instructions (or software code) that, when loaded into a computer, cause the computer to perform one or more functions described in the embodiments. The program may be stored on a non-transitory computer-readable medium or a tangible storage medium. By way of example and not limitation, computer-readable media or tangible storage media include random-access memory (RAM), read-only memory (ROM), flash memory, solid-state drive (SSD) or other memory technology, CD-ROM, digital versatile disc (DVD), Blu-ray (registered trademark) disc or other optical disk storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage device. The program may also be transmitted on a transitory computer-readable medium or communication medium. By way of example and not limitation, transitory computer-readable media or communication media include electrical, optical, acoustic, or other forms of propagated signals.

REFERENCE SIGNS LIST 1 Train accident avoidance system 2 Camera 3 Wireless communication network 4 AI server 5 Control server 6 Train 7 Management unit 8 Illuminator 20 Camera 21 Actuator 41 Predictor detection unit 42 Object detection unit 50 Control server 51 Train instruction unit 52 Camera instruction unit 53 Illumination instruction unit 60 Control server 61 Travel control unit

Claims (4)

an imaging means for imaging the track on which the train runs and the platform where people get on and off the train;
a communication means for transmitting the images of the tracks and platforms photographed by the photographing means by a wireless communication system;
a sign detection means for learning images showing signs of train accidents on the platform that may be the cause of the train accident, and for detecting signs of train accidents on the platform based on the learning results and the images of the platform transmitted by the communication means;
an object detection means for learning images of objects that may cause train accidents on the track, and for detecting objects that may cause train accidents on the track based on the learning results and images of the track transmitted by the communication means;
an instruction means for issuing a running instruction to the train,
The instruction means
When the sign detection means detects a sign of a train accident on the platform, a command to decelerate is issued to trains that are located within a predetermined distance from the platform and approaching the platform,
When an object on the tracks is detected by the object detection means, a stop instruction is issued to a train that is located within a predetermined distance from the platform and is approaching the platform;
further comprising a moving means for moving the photographing means so as to reduce a blind spot in the field of view of the photographing means;
The instruction means controls changes in lighting by turning on, off, blinking, or increasing or decreasing the illuminance of the platform lighting devices installed near the object indicating a sign of a train accident detected by the sign detection means and the object detected by the object detection means.
Train accident avoidance system. 2. The train accident avoidance system according to claim 1,
When the instruction means issues a deceleration instruction to the train, the instruction means transmits an image of the platform that is the basis of the deceleration instruction to a management unit that manages the train,
When the instruction means issues a stop instruction to the train, the instruction means transmits an image of the track that is the basis of the stop instruction to a management unit that manages the train.
Train accident avoidance system. a step of photographing, by an imaging means, a track on which a train runs and a platform where people get on and off the train;
transmitting the captured images of the tracks and platforms by wireless communication;
a step of learning images showing signs of a train accident on the platform that may cause the train accident, and detecting signs of a train accident on the platform based on the learning results and the transmitted images of the platform;
learning images of objects that may cause train accidents on the tracks, and detecting objects that may cause train accidents on the tracks based on the learning results and the transmitted images of the tracks;
a step of issuing a deceleration command to a train that is located within a predetermined distance from the platform and is approaching the platform when a sign of a train accident on the platform is detected;
When an object on the track is detected, issuing a stop instruction to a train that is located within a predetermined distance from the platform and is approaching the platform;
Including,
moving the photographing means so as to reduce a blind spot in the field of view of the photographing means;
Controlling changes in lighting by turning on, off, or blinking the lights of platform lighting devices installed near the detected object indicating a sign of a train accident and the object detected by the object detection means, or by increasing or decreasing the illuminance.
How to avoid train accidents. A process of learning images showing signs of train accidents on platforms that may cause train accidents, and detecting signs of train accidents on platforms based on the learning results and images of the tracks on which trains run and the platforms taken by an image taking device for taking pictures of the platforms where people get on and off the trains ;
A process of learning images of objects on the tracks that may cause train accidents, and detecting objects on the tracks that may cause train accidents based on the learning results and images of the tracks photographed by the photographing means ;
When a sign of a train accident on the platform is detected, a process of issuing a deceleration command to trains that are located within a predetermined distance from the platform and approaching the platform;
When an object on the tracks is detected, a process of issuing a stop instruction to a train that is located within a predetermined distance from the platform and is approaching the platform;
on the computer,
moving the photographing means so as to reduce a blind spot in the field of view of the photographing means;
Controlling changes in lighting by turning on, off, or blinking the lights of platform lighting devices installed near the detected object indicating a sign of a train accident and the object detected by the object detection means, or by increasing or decreasing the illuminance.
program.