CN115285150A

CN115285150A - Intelligent anti-pinch method, device, equipment and medium for rail transit platform door

Info

Publication number: CN115285150A
Application number: CN202210680511.7A
Authority: CN
Inventors: 李晶; 汪小勇
Original assignee: Casco Signal Ltd
Current assignee: Casco Signal Ltd
Priority date: 2022-06-15
Filing date: 2022-06-15
Publication date: 2022-11-04

Abstract

The invention relates to an intelligent anti-pinch method, an intelligent anti-pinch device, intelligent anti-pinch equipment and an intelligent anti-pinch medium for a rail transit platform door. Compared with the prior art, the method has the advantages of improving the safety in the specific scene operation environment and the like.

Description

Intelligent anti-pinch method, device, equipment and medium for rail transit platform door

Technical Field

The invention relates to a rail transit signal control system, in particular to an intelligent rail transit platform door anti-pinch method, device, equipment and medium based on machine vision and passenger track analysis.

Background

In an urban rail transit signal system based on vehicle-vehicle communication, control and state acquisition of a platform door are completed by a rail-side target controller, but the rail-side target controller can only sense whether the platform door is closed and locked through equipment such as an infrared sensor and provide related warning, once passengers get on and off the platform door and are clamped by an operation center, the operation center can only know that the platform door cannot be closed, specific reasons cannot be found in time, manual discovery needs to be carried out through station staff, and then the problem is solved according to manual operation of the process. However, in the existing unmanned train control system, a driver is not used, and once the station staff removes the shielding platform door from the system by operation errors, the possibility that the driver has danger when the platform door clamps the person due to secondary confirmation of the state of the platform door is avoided.

Along with machine vision's development, the camera is used for detecting the foreign matter between platform door and the train door, and infrared ray sensor before the contrast, the distinguishable foreign matter classification of camera carries out the basis to the reason that the platform door can't be closed and detects. The existing vision sensor uses methods such as 3D vision detection, namely only identifying a target object in a single video frame and not logically associating the same object in different video frames, and cannot be applied to certain scenes needing logical judgment. Whether an object clamped by the platform door is related to a passenger or not can not be accurately detected, for example, the passenger is clamped by a bag on the passenger, the object is judged to be clamped by the passenger, but the platform door can not be closed only due to the fact that foreign matters are found in the conventional visual perception scheme, and if the situation is similar, the operation efficiency of the rail transit system is judged to be influenced by the fact that the platform door is clamped by the passenger every time.

Through retrieval, chinese patent publication No. CN110091879a discloses a platform door anti-pinch detection system based on image recognition and a control method thereof, and specifically discloses a system in which a laser emitter, a projection plate and an image recognition unit are correspondingly arranged, a laser emitter emits visible laser to the projection plate, and the image recognition unit correspondingly recognizes the number of detection spots, and by comparing the number of spots with the number of working laser emitters, whether a person or an object is pinched between the platform door and a train can be effectively determined, but the existing patent cannot accurately detect whether an object pinched by the platform door is associated with a passenger, and the existing image recognition can only find the condition that the platform door cannot be closed due to a foreign object, so how to protect the passenger in the whole train getting on and off process becomes a technical problem to be solved.

Disclosure of Invention

The invention aims to overcome the defects in the prior art and provide an intelligent anti-pinch method, device, equipment and medium for a platform door of rail transit.

The purpose of the invention can be realized by the following technical scheme:

according to the first aspect of the invention, the intelligent anti-pinch method for the platform door of the rail transit is provided, the method tracks passengers getting on and off the platform through a machine vision algorithm and captures dynamic tracks of the passengers, the actions of the passengers are captured through the dynamic tracks, and the possibility and the danger of people pinching the platform door are judged by combining the current states of the platform door and the train door, so that the protection of getting on and off the train is realized.

According to the method, a visual perception device is used for monitoring the train getting-on and getting-off area of passengers, and the monitoring range of the visual perception device is required to cover the platform door and the area close to the train door inside the train under the condition that the platform door and the train door are opened simultaneously.

As a preferred technical scheme, the method identifies and tracks passengers in a visual perception area, and assigns identification codes to the passengers, wherein the visual perception area is a boarding and alighting area including the interior of a platform and a train.

As a preferred technical solution, the identification code is multi-level;

if the passengers are changed from the overlapped individuals into independent individuals in the visual perception area, the previous identification codes are reserved and the next-level identification codes are added;

if the passengers are changed from independent individuals to overlapped individuals in the visual perception area, the previous identification codes are reserved and the next-level identification codes are added;

if the objects are not separated in the visual perception area, the overlapped objects are judged to be the same object and have an identification code.

As a preferred technical solution, the method records the trajectory of each passenger and associates the identification code thereof, wherein the starting point of the trajectory is that the passenger enters the visual perception area, and the ending point of the trajectory is that the passenger leaves the visual perception area.

As a preferred technical scheme, the track of an object with a plurality of identification codes is spliced according to the sequence of the identification codes, so that the continuity of the track is ensured until the object leaves a sensing area;

for each level of identification code, the uniqueness of the identification code needs to be ensured, and finally, a plurality of tracks are allowed to be formed after a target object leaves a sensing area;

a unique identification code is set for each track.

As a preferred technical scheme, the identification code and the historical track of the passenger are emptied after the track display passenger leaves the visual perception area, and a new identification code is given when the passenger reenters.

As a preferable technical scheme, the method forms a complete behavior logic scene of passengers when getting on or off the vehicle through passenger trajectory analysis.

As a preferred technical scheme, when the real-time track displays the set time for stopping the passenger, the passenger is identified as a passenger stopping point;

and dividing an area which can cause the platform door to be clamped in the visual perception area, wherein the passenger track and the divided area which can be clamped in the platform door are unified under the same pixel coordinate system.

Preferably, the method includes acquiring a time point when the train control system closing command executed by the platform door fails, and confirming that the single passenger gets on or off the train through a historical track of the single passenger before the time point.

According to the method, the passenger historical behaviors are integrated, and the passenger logic scene is formed by the real-time passenger track and the passenger stopping point after the closing of the screen door fails so as to judge the passenger clamping risk.

As a preferred technical solution, the final result of the method integrates the risk of the platform door clamping people and the risk thereof by integrating the risks of clamping all passengers appearing in the visual perception area when the platform door is failed to close.

As a preferred technical scheme, the method unifies the possibility and the danger of clamping the platform door into three grades, namely low, common and high, and configures corresponding processing schemes for the three grades.

According to a second aspect of the invention, the device for the intelligent anti-pinch method for the rail transit platform door comprises a visual perception device, a trackside target controller and a trackside edge calculation device;

the arrangement of the visual perception equipment can cover the areas in the train, near the train door and the areas near the platform outside the platform door when the train door and the platform door are opened;

the trackside target controller acquires the current states of the platform door and the train door and whether the platform door and the train door successfully execute a door closing instruction issued by the train control system;

the track side edge computing equipment is responsible for structuring video data so as to track the same object in different video frames and obtain a dynamic track of the same object; meanwhile, the position of the train door of the platform door under the pixel coordinate system can be prestored by the track side edge computing device, and the current platform door state and the target object track captured by the visual perception device are fused for carrying out logic operation so as to identify a dangerous scene and send an alarm.

According to a third aspect of the invention, there is provided an electronic device comprising a memory having stored thereon a computer program and a processor implementing the method when executing the program.

According to a fourth aspect of the invention, a computer-readable storage medium is provided, on which a computer program is stored which, when being executed by a processor, carries out the method.

Compared with the prior art, the invention has the following advantages:

1) According to the invention, the identification accuracy of the fault scene generation reason of the platform door closing failure is improved through the track and behavior logic of the passengers in the passenger getting-on and getting-off processes.

2) The invention unifies the possibility of clamping people by the platform door and the danger degree thereof into a measurement standard, thereby reducing the complexity of the system.

3) The invention increases the safety of the system by identifying the risk probability of people clamping the platform door.

4) The invention can accurately distinguish the risk of people clamping the platform door and the incapability of closing the platform door caused by other reasons, thereby enhancing the operation efficiency of the system.

5) The invention can be coupled with the existing unmanned train control system with low coupling, thereby enhancing the system redundancy.

Drawings

FIG. 1 is a schematic view of an intelligent anti-pinch device for a platform door of a rail transit system according to the present invention;

FIG. 2 is a flowchart of an overall passenger identification and tracking algorithm of the present invention;

FIG. 3 is a flowchart illustrating an algorithm for determining a possibility of a platform door being caught by a person according to the present invention;

fig. 4 is a schematic diagram illustrating a situation that the getting-off of passengers is clamped at a high possibility when the train door is successfully closed due to the failure of closing the platform door;

FIG. 5 is a schematic diagram illustrating a situation where the passenger getting-off is clamped low in probability when the train door is successfully closed due to the failed platform door closing;

FIG. 6 is a schematic diagram illustrating a general possibility scenario of passengers getting on and being caught when a train door is successfully closed due to a failed platform door closing;

FIG. 7 is a schematic view of a general possibility scene of passengers getting on the train and being caught when the platform door and the train door fail to be closed simultaneously;

fig. 8 is a schematic view of a general possibility scene of a passenger getting-off jam under a failure of simultaneous closing of a platform door and a train door;

fig. 9 is a schematic view illustrating a situation where passengers who have failed to get on or off the train simultaneously with the platform door are clamped at a low possibility;

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, shall fall within the scope of protection of the present invention.

The invention relates to a rail transit intelligent platform door anti-pinch method based on machine vision passenger track analysis, which is characterized in that a machine vision algorithm is used for identifying and tracking passengers in a vision perception area to record unique tracks of the passengers, behavior logics of the passengers are judged by combining the tracks with the states of a platform door and a train door, so that the possibility and the danger of platform door pinching are generated from the perspective of passenger behaviors, and the possibility and the danger of related hazards are unified to be low, common and high, and meanwhile, different processing schemes are provided for the passengers. The invention can realize integral protection to the whole process of getting on and off the bus of passengers, forms a passenger behavior logic scene by judging the passenger track and behavior, and judges the possibility of people clamping at the platform door and the danger thereof. The intelligent protection system is beneficial to safety protection of passengers in the process of getting on and off the train, can be used as a supplement of the existing system after being linked with the existing signal system to improve the safety under the operation environment of a specific scene, can be adapted to platform operation in the existing rail transit unmanned and manned systems, and realizes intelligent protection of platform doors.

In order to realize the identification and tracking of passengers, a target capture deep learning algorithm and a multi-target tracking algorithm are selected in the implementation. A neural network algorithm based on Draknet 53 is selected to identify a small target object, realize detail detection of the target object and have excellent real-time performance. The multi-target tracking algorithm is evolved from the Sort algorithm, the SORT algorithm uses simple Kalman filtering to process the relevance of frame-by-frame data and uses the Hungarian algorithm to carry out relevance measurement, and the simple algorithm obtains good performance under a high frame rate. However, since the SORT ignores the surface characteristics of the detected object, the estimation is accurate only when the uncertainty of the object state estimation is low, and when the method is implemented, more reliable measurement is used for replacing the associated measurement, a CNN network is used for training in a large-scale pedestrian data set, the characteristics are extracted, and the robustness of the network to loss and obstacles is increased. The whole algorithm is equivalent to a second-order structure in target detection, and target identification and tracking are adopted. The overall algorithm implementation is shown in fig. 2. And (3) obtaining the unique ID of each passenger in the video monitoring area by using an algorithm, and obtaining the moving track of each passenger in the monitoring area under a pixel coordinate system, wherein each unique ID corresponds to a pedestrian track for subsequent logic judgment.

The probability of the occurrence of the person clamping scene is judged according to the action track of the passenger in the visual perception area and the behavior logic after door closing failure, the probability is divided into three parts, namely low, common and high through a track logic analysis algorithm, as shown in fig. 3, different possibilities correspond to different processing flows, the risk of person clamping is reduced, and meanwhile, the operation efficiency is considered. And if the person clamping risk is judged to be low, performing secondary door closing attempt remotely, and if the person clamping risk is judged to be common, performing secondary confirmation by linking the CCTV through the operation center to confirm whether person clamping occurs. And if the risk of people involved is judged to be high, a train remote braking instruction is directly and remotely sent to ensure that the train does not move and the safety problem does not occur.

The normal state is when the train door and the platform door are successfully closed at the same time.

When the train door is successfully closed but the platform door is not closed successfully, the current moment is recorded as t1, all passengers in the video monitoring area are identified at the moment, and the tracks of the passengers in the pixel coordinate system are inquired.

As shown in fig. 4, when the passenger behavior track shows that the passenger behavior intends to get off the train, and the passenger track does not obviously move within 7 seconds from the time t1 to judge that the track stops, and the track stop point is within a dangerous area (within one meter in the longitudinal and transverse directions of the platform door), if the passenger track in the monitoring area conforms to the description, the possibility that the passenger is clamped when getting off the train is judged to be high, therefore, the train door is successfully closed, if the platform staff does not handle the movement risk of the automatic driving train, the platform door is judged to be high in the risk level of clamping the passenger.

When the passenger behavior track shows that the passenger behavior intention is to get off as shown in fig. 5, no stopping point is formed or the stopping point is outside the monitoring area from the time t1 to the time when the passenger track leaves the monitoring area, if the passenger track in the monitoring area conforms to the description, the passenger is judged to be low in clamping possibility, if the tracks of all the passengers do not have the ordinary or high risk level, the passenger clamping risk of the platform door is judged to be low, if the tracks of all the passengers include the higher-level passenger clamping risk, the alarm is given according to the highest possibility, and the next processing is carried out.

When the passenger behavior track shows that the passenger behavior is intended to get on the bus as shown in fig. 6, the passenger track does not obviously move within 7 seconds from the time t1 to judge that the track stops, the track stop point is within a dangerous area (within one meter in the longitudinal direction and the transverse direction of the departure platform door), if the passenger track in the monitoring area conforms to the description, the passenger behavior may be that the passenger does not get on the bus and waits on the platform, if the tracks of all the passengers do not have high dangerous grades, the passenger clamping risk of the platform door is judged to be common, if the tracks of all the passengers contain higher-grade passenger clamping risks, the alarm is given according to the highest possibility, and the next processing is carried out.

When the train door is failed to close and the platform door is failed to close, the current moment is recorded as t1, all passengers in the video monitoring area are identified at the moment, and the tracks of the passengers in the pixel coordinate system are inquired.

As shown in fig. 7 and 8, when the passenger behavior track shows that the passenger behavior is intended to get off or get on, the passenger track does not obviously move within 7 seconds from the time t1 to judge that the track stops, and the track stop point is judged to be in a normal risk level of the platform door being clamped in the dangerous area, therefore, although the passenger is clamped possibly, the train door is closed and the locking fails, at the moment, the automatic train cannot take the movement authorization, and the automatic train does not have the movement risk and does not need to directly send emergency braking.

As shown in fig. 9, the passenger trajectory from time t1 to leaving the monitoring area does not form a stopping point or the stopping point is outside the monitoring area, if the passenger trajectory in the monitoring area conforms to the above description, it is determined that the passenger is low in possibility of being caught, and if the trajectories of all passengers do not have a normal or high risk level, it is determined that the platform door is low in risk of catching people, and if the trajectories of all passengers include a higher risk of catching people, the alarm is given according to the highest possibility, and the next step is performed.

The above is a description of method embodiments, and the embodiments of the present invention are further described below by way of apparatus embodiments.

As shown in fig. 1, the intelligent platform door anti-pinch device based on machine vision passenger trajectory capture of the present invention includes a camera, a trackside target controller, and trackside edge computing equipment. The arrangement of the camera can cover the area near the train door in the train and the area near the platform outside the platform door when the train door platform door is opened. The trackside target controller is responsible for acquiring the current states of the platform door and the train door and whether the platform door and the train door successfully execute a door closing instruction issued by the train control system, and the trackside edge computing equipment is responsible for structuring the video data so as to realize tracking of the same object in different video frames and acquisition of the dynamic track of the same object. Meanwhile, the position of the train door of the platform door train under the pixel coordinate system can be prestored by the track side edge computing equipment, and the current platform door state and the target object track captured by the visual perception equipment are fused for carrying out logic operation so as to identify a dangerous scene and send an alarm.

It can be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working process of the described module may refer to the corresponding process in the foregoing method embodiment, and is not described herein again.

The electronic device of the present invention includes a Central Processing Unit (CPU) that can perform various appropriate actions and processes according to computer program instructions stored in a Read Only Memory (ROM) or computer program instructions loaded from a storage unit into a Random Access Memory (RAM). In the RAM, various programs and data required for the operation of the device can also be stored. The CPU, ROM, and RAM are connected to each other via a bus. An input/output (I/O) interface is also connected to the bus.

A plurality of components in the device are connected to the I/O interface, including: an input unit such as a keyboard, a mouse, etc.; an output unit such as various types of displays, speakers, and the like; storage units such as magnetic disks, optical disks, and the like; and a communication unit such as a network card, modem, wireless communication transceiver, etc. The communication unit allows the device to exchange information/data with other devices via a computer network such as the internet and/or various telecommunication networks.

The processing unit performs the various methods and processes described above, such as the method of the present invention. For example, in some embodiments, the inventive methods may be implemented as a computer software program tangibly embodied in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and/or installed onto the device via ROM and/or the communication unit. When the computer program is loaded into RAM and executed by a CPU, it may perform one or more of the steps of the method of the invention described above. Alternatively, in other embodiments, the CPU may be configured to perform the inventive method by any other suitable means (e.g. by means of firmware).

The functions described herein above may be performed, at least in part, by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems on a chip (SOCs), complex Programmable Logic Devices (CPLDs), and the like.

Program code for implementing the methods of the present invention may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the program codes, when executed by the processor or controller, cause the functions/operations specified in the flowchart and/or block diagram to be performed. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of the present invention, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

While the invention has been described with reference to specific embodiments, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. The intelligent anti-pinch method for the platform door of the rail transit is characterized in that passengers getting on and off the platform are tracked through a machine vision algorithm, dynamic tracks of the passengers are captured, behaviors of the passengers are captured through the dynamic tracks, and the possibility and danger of people pinching of the platform door are judged by combining the current states of the platform door and the train door, so that passenger getting on and off of a train are protected.

2. The intelligent rail transit platform door anti-pinch method as claimed in claim 1, wherein a visual perception device is used to monitor the passenger getting on and off the train in a region covering both the platform door and the train door inside the train when the platform door and the train door are opened simultaneously.

3. The intelligent rail transit platform door anti-pinch method according to claim 1, wherein the method identifies and tracks passengers in a visually perceptible area, wherein the visually perceptible area is a boarding and disembarking area, including inside of platforms and trains, and assigns identification codes to the passengers.

4. The intelligent rail transit platform door anti-pinch method of claim 3, wherein the identification code is multi-level;

5. The method of claim 3, wherein the method records the track of each passenger and associates the identification code thereof, wherein the starting point of the track is that the passenger enters the visual perception area, and the ending point of the track is that the passenger leaves the visual perception area.

6. The intelligent anti-pinch method for rail transit platform doors according to claim 5,

splicing the tracks of the objects with a plurality of identification codes according to the sequence of the identification codes, and ensuring the continuity of the tracks until the objects leave the sensing area;

a unique identification code is set for each track.

7. The intelligent anti-pinch method for rail transit platform doors according to claim 5, wherein the track indicates that the passenger leaves the visual perception area and clears his identification code and the historical track, and when he re-enters the track, the track is assigned a new identification code.

8. The intelligent anti-pinch method for rail transit platform doors according to claim 1, wherein the method forms a complete behavior logic scene of passengers when getting on or off the train through passenger trajectory analysis.

9. The intelligent anti-pinch method for rail transit platform doors according to claim 8, wherein a passenger stop point is identified when the real-time trajectory indicates a passenger stop set time;

10. The method as claimed in claim 8, wherein the method is used to confirm that a single passenger gets on or off the train through the historical track of the passenger before the time point by obtaining the time point when the train control system closing command fails.

11. The intelligent anti-pinch method for rail transit platform doors according to claim 8, wherein the method integrates historical passenger behaviors, real-time passenger trajectories and passenger stopping points after the closing of the screen door fails to form a passenger logic scene so as to judge the risk of pinching the passenger.

12. The intelligent anti-pinch method for rail transit platform doors according to claim 8, wherein the final result of the method is to determine the risk of platform door pinching and the risk thereof by integrating the risk of pinching of all passengers who appear in the visual perception area when the platform door fails to close.

13. The intelligent anti-pinch method for platform doors in rail transit as claimed in claim 8, wherein the method unifies possibility and danger of platform door pinching into three levels, low, normal, high, and configures corresponding processing schemes for them.

14. An apparatus for the intelligent rail transit platform door anti-pinch method of claim 1, comprising a visual perception device, a trackside target controller and a trackside edge computing device;

15. An electronic device comprising a memory and a processor, the memory having stored thereon a computer program, wherein the processor, when executing the program, implements the method of any of claims 1-13.

16. A computer-readable storage medium, on which a computer program is stored, which program, when being executed by a processor, is adapted to carry out the method of any one of claims 1 to 13.