A crossing signal control method for vulnerable pedestrian groups

By installing video detectors on zebra crossings, identifying and protecting vulnerable pedestrians and extending pedestrian green flash signals, the neglect of the convenience and safety of vulnerable groups in the existing technology is solved, and the safety of pedestrians crossing and other traffic participants are realized. Waiting for the balance of time.

CN115762117BActive Publication Date: 2024-11-12JIANGSU AEROSPACE DAWEI TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202211398787.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-11-09
Publication Date
2024-11-12
Estimated Expiration
2042-11-09

AI Technical Summary

Technical Problem

The existing pedestrian crossing signal control method fails to effectively take into account the convenience of crossing the street of vulnerable groups (such as the elderly, pregnant women, disabled people, etc.), and individual speed differences are ignored when pedestrians cross the street, which may increase the waiting time for other traffic participants. Or accidentally triggers a crossing request by the person.

Method used

By installing video detectors at both ends of the zebra crossing, deep learning methods are used to analyze pedestrian video images, identify and mark vulnerable pedestrian individuals, calculate their remaining time to cross the street, and extend the signal when pedestrians are green flashing to protect vulnerable pedestrians from crossing the street and ensure their safety Sex and convenience.

Benefits of technology

Improve the safety and convenience of crossing the street by vulnerable pedestrians, avoid traffic accidents caused by failure to cross the street in time, enhance the accuracy of identifying groups of vulnerable pedestrians, and improve the data sample size and identification of the auxiliary system through feedback training and data collection. accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115762117B_ABST
    Figure CN115762117B_ABST
Patent Text Reader

Abstract

The present invention discloses a method for controlling a crossing signal for a vulnerable pedestrian group, comprising the steps of: establishing a corresponding pedestrian individual label data set; using a video detector to output the coordinate information of the pedestrian individual; calculating the information in the current pedestrian individual matrix; using a deep learning method to analyze the video, classify the pedestrians, and calculate the remaining crossing time of the current vulnerable pedestrian individual; when the pedestrian crossing phase is green, using the video detectors at both ends of the zebra crossing to perform morphological recognition on all pedestrian individuals, and if there are vulnerable pedestrian individuals, calculate the remaining crossing time of the largest vulnerable pedestrian group; the output results of the video detectors in each direction are transmitted to the recognition auxiliary system, and the system transmits the green flash extension time to the signal machine. The present invention avoids traffic accidents caused by the vulnerable pedestrian group failing to complete the crossing in time, collects data from each networked assembly intersection, increases the data sample size, and performs cluster training and learning.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of automation technology, and in particular relates to a street crossing signal control method for vulnerable pedestrian groups. Background Art

[0002] Pedestrian crossing is something that must be taken into consideration in urban traffic. In view of the current actual situation in the city, it is not recommended to build a large number of pedestrian overpasses or underground passages. Pedestrian crossing at zebra crossings is an issue that needs to be given special consideration in traffic release.

[0003] Some current related technologies focus on signal control for elderly people crossing the street, but fail to fully consider that vulnerable groups are not limited to the elderly. People with immobility, pregnant women, and the disabled are all issues that should be considered when pedestrians cross the street. There are also some that only allow pedestrians to cross the street when a certain number of pedestrians reach the crossing point, or that crossing control buttons are installed on lamp poles, which do not take into account the convenience or necessity of crossing the street.

[0004] Pedestrian crossing signal control methods often consider pedestrians as a whole, without taking into account the situations that often occur in vulnerable pedestrian groups when crossing the street. In current pedestrian crossing control, the remaining crossing time of pedestrians is generally calculated, and then the pedestrian crossing phase time is extended.

[0005] There are several different schemes for pedestrian crossing at present, all of which have certain shortcomings. For example, the remaining crossing time calculation of pedestrians ignores the fact that pedestrians will subconsciously speed up when they see the green flashing pedestrian light group, thus wasting a few seconds of waiting time for vehicles in other phases; for example, a pedestrian waiting area is set up, and the pedestrian green light phase is turned on to allow pedestrians to cross the street only when the number of pedestrians reaches a threshold; for example, a pedestrian crossing button is installed on a pedestrian light pole, and the pedestrian crossing request is triggered by the button, and other uncertain factors may misinterpret the pedestrian crossing request, etc. Summary of the invention

[0006] In view of this, taking into account the different psychological factors of pedestrians of different behavior groups when crossing the zebra crossing and the differences in individual crossing speeds, the present invention proposes a crossing signal control method for vulnerable pedestrian groups, aiming to improve the safety of pedestrian crossing and the convenience of vulnerable pedestrian groups on the basis of not affecting the waiting time of traffic participants in other phases during the pedestrian crossing phase.

[0007] The present invention discloses a method for controlling a street crossing signal for vulnerable pedestrian groups, comprising the following steps:

[0008] Step 1: Establish a pedestrian individual label dataset;

[0009] Step 2: Calibrate the two video detectors installed at both ends of the zebra crossing according to their parameters, transform the video coordinate system into the intersection coordinate system with the center of the intersection as the origin, and use the two video detectors to output the actual spatial position of the pedestrian crossing once, and output the individual coordinate information L of the pedestrian. per ;

[0010] Step 3: Based on the pedestrian individual coordinate information L obtained in Step 2 per , calculate the current pedestrian individual V per dir in matrix per 、a per 、v per Information, where dir per Indicates the direction of pedestrians, a per represents the pedestrian acceleration, v per Indicates the current speed of individual pedestrians;

[0011] Step 4: Use deep learning methods to analyze and process the video images collected by the video detector, perform morphological perception and recognition on individual pedestrians, and classify pedestrians. per Classify and determine whether they belong to the vulnerable pedestrian group. If they are determined to belong to the vulnerable group, give o per Assign 1 and calculate the remaining crossing time t for the current disadvantaged pedestrian individual per ;

[0012] Step 5: When the pedestrian crossing phase is green, use the video detectors at both ends of the zebra crossing to identify the morphology of all pedestrians. If there are weak pedestrians, activate the pedestrian green flash phase extension signal; before the pedestrian green flash starts, calculate the remaining crossing time of all weak pedestrians, generate a weak pedestrian group set, and take the largest v in the set per .t per The value is used as the remaining time for all vulnerable pedestrian groups to cross the street. If another vulnerable pedestrian individual enters the pedestrian identification area before the pedestrian flashes green, the remaining time for the largest vulnerable pedestrian group to cross the street is recalculated and the original remaining time is replaced. When the pedestrian phase flashes green, v per .t per -t 行人绿闪相位时长 The result is transmitted to the signal machine as the green flash phase time that needs to be extended;

[0013] Step 6: The output results of the video detectors in each direction are transmitted to the recognition assistance system through the gateway. After the system calculates the output results, the green flash extension time is transmitted to the traffic light through the network to protect vulnerable pedestrians crossing the street.

[0014] Furthermore, the pedestrian individual label dataset is shown in the following formula:

[0015] V per =[o per ,dir per ,a per ,v per ,t per ,thick per ]

[0016] Among them, per Indicates the pedestrian category, dir per Indicates the direction of pedestrians, a per represents the pedestrian acceleration, v per represents the current speed of the pedestrian, t per Indicates the remaining time for pedestrians to cross the street. If there is a secondary crossing island, if the pedestrian has not reached the secondary crossing island, the remaining time for the pedestrian to cross the street is the time to reach the secondary crossing island. If the pedestrian has passed the secondary crossing island, the remaining time for the pedestrian to cross the street is the time to reach the end of the zebra crossing. per is the pedestrian density coefficient around the pedestrian at the current moment.

[0017] Furthermore, in the intersection coordinate system, according to the plane coordinate x of the current pedestrian individual in the intersection coordinate system per ,y per The coordinate difference within unit time t is obtained by dir perx Value and dir pery , which is calculated as shown in the following formula:

[0018] if y per |t2-y per |t1>0

[0019] dir pery =1

[0020] else dir pery =2

[0021] if x per |t2-x per |t1>0

[0022] dir perx =3

[0023] else dir perx =4

[0024] For north-south and east-west crossings, the v per , calculated according to the following formula:

[0025]

[0026]

[0027] According to the above calculation v per , we can calculate the a of individual pedestrians when crossing the street in the north-south direction and the east-west direction. pery and a perx , the calculation formula is as follows:

[0028]

[0029]

[0030] Furthermore, the remaining pedestrian crossing time t per , the calculation process is shown in the following formula:

[0031]

[0032]

[0033] In the formula, o per Determine the value for the vulnerable pedestrian individual, thick per is the density coefficient of the crowd around the current vulnerable pedestrian individual, and The total length of north-south and east-west routes for individual vulnerable pedestrians crossing the street. is the length of the street currently walked, is the maximum moving speed of the vulnerable pedestrian within the current time t1.

[0034] Furthermore, the pedestrian recognition auxiliary system records the data of all pedestrians who have not been identified as vulnerable pedestrians, and records their crossing data matrix V per , analyze the crossing behavior and identify whether it is a strong non-vulnerable pedestrian individual. If it is not a strong non-vulnerable pedestrian, detect the pedestrian's morphological data and feed it back to the system as feedback input to improve the accuracy of the recognition system in identifying individuals in the vulnerable pedestrian group; in addition, the feedback data of the above identification at various intersections in the city is transmitted to the recognition system through the central gateway to expand its training sample set.

[0035] The beneficial effects of the present invention are as follows:

[0036] 1) Taking care of vulnerable pedestrian groups: When the pedestrian crossing phase is green, it is impossible to increase their crossing speed, which may lead to traffic accidents due to their failure to cross the street in time.

[0037] 2) The vulnerable pedestrian crossing assistance system adopts a feedback training method, focusing on the recognition accuracy of various vulnerable pedestrian groups, and re-identifying abnormal individual data, such as non-vulnerable pedestrians who fail to accelerate through the intersection under green flash conditions.

[0038] 3) It can collect data from each networked intersection, which can not only increase the data sample size of the pedestrian crossing assistance system for vulnerable pedestrians, but also enable the system to serve as a cluster training and learning system. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 Schematic diagram of the intersection coordinate system of the present invention;

[0040] Figure 2 A schematic diagram of a network connection between the vulnerable pedestrian group crossing identification auxiliary system and a signal machine of the present invention;

[0041] Figure 3 Schematic diagram of video detector installation of the present invention. DETAILED DESCRIPTION

[0042] The present invention is further described below in conjunction with the accompanying drawings, but the present invention is not limited in any way. Any changes or substitutions made based on the teachings of the present invention belong to the protection scope of the present invention.

[0043] The detailed steps of the present invention are as follows:

[0044] Step 1: According to the functions described and implemented in this technical solution, establish the corresponding pedestrian individual label data set, as shown in the following formula 1:

[0045] V per =[o per ,dir per ,a per ,v per ,t per ,thick per ]

[0046] Among them, per Indicates the pedestrian category, dir per Indicates the direction of the pedestrian (defined as 1 if going from south to north, 2 if going from north to south, 3 if going from east to west, and 4 if going from west to east), a per represents the pedestrian acceleration, v per represents the current speed of the pedestrian, t per Indicates the remaining crossing time for pedestrians (if there are two crossing islands, if the pedestrian has not reached the second crossing island, the crossing island time is used; if the pedestrian has passed the second crossing island, the zebra crossing end time is used), d per is the pedestrian density coefficient around the pedestrian at the current moment.

[0047] Step 2: Install a video detector at each end of the zebra crossing, calibrate it according to the video detector parameters, transform the video coordinate system into the intersection coordinate system with the center of the intersection as the origin, and use two video detectors to output the coordinates of the actual spatial position of pedestrians crossing the zebra crossing once. The output matrix is ​​shown in the following formula 2:

[0048] L per =[x per ,y per ,t per ]

[0049] Among them, x per ,y per They are the plane coordinates of the current pedestrian in the intersection coordinate system, t per It is the time information of the current coordinates.

[0050] Step 3: According to the pedestrian individual coordinate information L obtained in step 2 per , calculate the current pedestrian individual V per dir in matrix per 、a per 、v per information.

[0051] according to Figure 1 In the intersection coordinate system shown (the coordinates in the north and east directions are positive, and the coordinates in the south and west directions are negative), according to x per ,y per The coordinate difference within unit time t can be obtained by dir per The value is calculated as shown in the following formulas 3 and 4:

[0052] if y per |t2-y per |t1>0

[0053] dir pery =1

[0054] else dir pery =2

[0055] if x per |t2-x per |t1>0

[0056] dir perx =3

[0057] else dir perx =4

[0058] For north-south and east-west crossings, the v per It can be calculated according to the following formulas 5 and 6:

[0059]

[0060]

[0061] According to the above calculation v per , we can calculate the a of individual pedestrians when crossing the street in the north-south direction and the east-west direction. per , the calculation formula is shown below7:

[0062]

[0063]

[0064] Step 4: Use deep learning methods to analyze and process the video images collected by the video detector, perform morphological perception and recognition on individual pedestrians, and calculate the o in the above formula 1. per Pedestrians are classified into different categories to determine whether they belong to the vulnerable pedestrian group. If they are determined to belong to the vulnerable group, o per Assign 1 and calculate the remaining time t for the current vulnerable pedestrian to cross the street per The calculation process is shown in the following formula:

[0065]

[0066]

[0067] In the formula, o per Determine the value for the vulnerable pedestrian individual, thick per is the density coefficient of the crowd around the current vulnerable pedestrian individual, is the total length of the current individual crossing behavior route of vulnerable pedestrians, is the length of the street currently walked, is the maximum moving speed of the vulnerable pedestrian within the current time t1.

[0068] Step 5: When the pedestrian crossing phase is green, use the video detectors at both ends of the zebra crossing to identify the morphology of all pedestrians. If there are weak pedestrians, activate the pedestrian green flash phase extension signal. Before the pedestrian green flash starts, calculate the remaining crossing time of all weak pedestrians and generate the weak pedestrian group set S per = {P1,P2,…,P n}, take the maximum V per .t perAs the remaining time for all vulnerable pedestrian groups to cross the street, if another vulnerable pedestrian individual enters the pedestrian identification area before the pedestrian phase flashes green, the remaining time for the largest vulnerable pedestrian group to cross the street is recalculated and the original remaining time is replaced. When the pedestrian phase flashes green, V per .t per -t 行人绿闪相位时长 This value is transmitted to the signal machine as the green flash phase time that needs to be extended.

[0069] Step 6: The vulnerable pedestrian group crossing identification auxiliary system is connected to the signal machine using a network transmission method. Its structure is shown in the following figure. Figure 2 The output of the video detectors in each direction is transmitted to the identification auxiliary system through the gateway. After the system calculates the output result, it transmits the green flash extension time to the signal machine through the network to protect vulnerable pedestrians crossing the street.

[0070] Step 7: At the same time, the pedestrian recognition auxiliary system records the data of all pedestrians who have not been identified as vulnerable pedestrians, and records their crossing data matrix V per , analyze the crossing behavior and identify whether it is a strong non-weak pedestrian individual. If it is not a strong non-weak pedestrian, the pedestrian's morphological data is detected and fed back to the system as feedback input to improve the accuracy of the recognition system in identifying individuals in the vulnerable pedestrian group. In addition, the feedback data of the above identification at various intersections in the city is transmitted to the recognition system through the central gateway to expand its training sample set.

[0071] According to whether there is a pedestrian secondary crossing island, two embodiments of pedestrian crossing signals for different vulnerable groups are provided. The pedestrian secondary crossing island is a pedestrian safety island set on the crosswalk at a wider intersection, dividing the crosswalk into two parts, thereby improving the safety of pedestrians crossing the street at a wider crosswalk intersection. Both Embodiment 1 and Embodiment 2 require the installation of a video detector according to step 2 in the detailed description of this technical solution, and the installation diagram is shown in FIG. Figure 3 As shown, the dotted box is the joint detection area of ​​the two video detectors: the difference between the two lies in whether there is a safety island in the middle of the pedestrian crossing, which will lead to differences in the subsequent calculation of the green light extension time for the pedestrian phase.

[0072] Example 1: Pedestrian island without secondary crossing

[0073] S1: Before the pedestrian crossing phase turns green, the data matrix collected by the two video collectors in the previous cycle is cleared to avoid affecting the data content of the current signal cycle.

[0074] S2: When the pedestrian crossing phase turns green, the two video detectors start working, process the video content, calculate and output the individual coordinate matrix of the pedestrian, and obtain the individual coordinate matrix of the pedestrian at time t.

[0075] V per =[o per ,dir per ,a per ,v per ,t per ,thick per ]

[0076] S3: Use deep learning methods to perceive and identify the morphological behaviors of individual pedestrians in the video content, and label the vulnerable pedestrians to form an S per = {P1,P2,…,P n} a collection of vulnerable pedestrian groups.

[0077] S4: Always update this S before the pedestrian crossing phase green flashes per Matrix, when green flashes, get the set, t per The maximum value is used as the basis for extending the green flash phase signal time.

[0078] S5: Finally, V per .t per -t 行人绿闪相位时长 The result is transmitted to the traffic light through the gateway to extend the green flashing time during the pedestrian crossing phase.

[0079] S6: All abnormal results of video recognition are uploaded to the central routing server, which recalculates these abnormal data and sends the updated model to the vulnerable pedestrian group crossing assistance system.

[0080] Example 2 has two pedestrian islands for crossing the street

[0081] S1: Before the pedestrian crossing phase turns green, the data matrix collected by the two video collectors in the previous cycle is cleared to avoid affecting the data content of the current signal cycle.

[0082] S2: When the pedestrian crossing phase turns green, the two video detectors start working, process the video content, calculate and output the individual coordinate matrix of the pedestrian, and obtain the individual coordinate matrix of the pedestrian at time t.

[0083] V per =[o per ,dir per ,a per ,v per ,t per ,thick per ]

[0084]

[0085] In the formula, o per Determine the value for the vulnerable pedestrian individual, thick peris the density coefficient of the crowd around the current vulnerable pedestrian individual, is the total length of the current individual crossing behavior route of vulnerable pedestrians, is the length of the street currently walked, is the maximum moving speed of the vulnerable pedestrian within the current time t1.

[0086] Calculate t per If the pedestrian has not reached the secondary crossing island and the pedestrian green flashing phase is on, then 1 / 2 of the total crossing distance is taken as the total crossing distance. If the pedestrian green flashing phase is on after passing the second crossing island, then As the total journey across the street.

[0087] S3: Use deep learning methods to perceive and identify the morphological behaviors of individual pedestrians in the video content, and label the vulnerable pedestrians to form an S per = {P1,P2,…,P n} a collection of vulnerable pedestrian groups.

[0088] S4: Always update this S before the pedestrian crossing phase green flashes per Matrix, when green flashes, get the set, t per The maximum value is used as the basis for extending the green flash phase signal time.

[0089] S5: Finally, V per .t per -t 行人绿闪相位时长 The result is transmitted to the traffic light through the gateway to extend the green flashing time during the pedestrian crossing phase.

[0090] S6: All abnormal results of video recognition are uploaded to the central routing server, which recalculates these abnormal data and sends the updated model to the vulnerable pedestrian group crossing assistance system.

[0091] The beneficial effects of the present invention are as follows:

[0092] 1) Taking care of vulnerable pedestrian groups: When the pedestrian crossing phase is green, it is impossible to increase their crossing speed, which may lead to traffic accidents due to their failure to cross the street in time.

[0093] 2) The vulnerable pedestrian crossing assistance system adopts a feedback training method, focusing on the recognition accuracy of various vulnerable pedestrian groups, and re-identifying abnormal individual data, such as non-vulnerable pedestrians who fail to accelerate through the intersection under green flash conditions.

[0094] 3) It can collect data from each networked intersection, which can not only increase the data sample size of the pedestrian crossing assistance system for vulnerable pedestrians, but also enable the system to serve as a cluster training and learning system.

[0095] As used herein, the word "preferred" is intended to be used as an example, instance, or illustration. Any aspect or design described herein as "preferred" is not necessarily to be construed as being more advantageous than other aspects or designs. On the contrary, the use of the word "preferred" is intended to present concepts in a specific way. The term "or" as used in this application is intended to mean an inclusive "or" rather than an exclusive "or". That is, unless otherwise specified or clear from the context, "X uses A or B" means any one of the naturally included permutations. That is, if X uses A; X uses B; or X uses both A and B, then "X uses A or B" is satisfied in any of the foregoing examples.

[0096] Moreover, although the present disclosure has been shown and described with respect to one or implementations, those skilled in the art will think of equivalent variations and modifications based on the reading and understanding of this specification and the accompanying drawings. The present disclosure includes all such modifications and variations, and is limited only by the scope of the appended claims. In particular, with respect to the various functions performed by the above-mentioned components (such as elements, etc.), the terms used to describe such components are intended to correspond to any component (unless otherwise indicated) that performs the specified function of the component (such as it is functionally equivalent), even if the structure is not equivalent to the disclosed structure of the function in the exemplary implementation of the present disclosure shown herein. In addition, although the specific features of the present disclosure have been disclosed with respect to only one of several implementations, such features can be combined with one or other features of other implementations that may be desired and advantageous for a given or specific application. Moreover, insofar as the terms "including", "having", "containing" or their variations are used in specific embodiments or claims, such terms are intended to be included in a manner similar to the term "comprising".

[0097] The functional units in the embodiments of the present invention may be integrated into a processing module, or each unit may exist physically separately, or multiple or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium. The above-mentioned storage medium may be a read-only memory, a disk or an optical disk, etc. The above-mentioned devices or systems may execute the storage method in the corresponding method embodiment.

[0098] To sum up, the above embodiment is an implementation mode of the present invention, but the implementation mode of the present invention is not limited by the embodiment. Any other changes, modifications, substitutions, combinations, and simplifications that deviate from the spirit and principles of the present invention should be equivalent replacement methods and are included in the protection scope of the present invention.

Claims

1. A method for controlling street crossing signals for vulnerable pedestrian groups, characterized in that: The following steps are involved: Step 1: Establish a pedestrian individual label dataset; Step 2: Calibrate the two video detectors installed at both ends of the zebra crossing according to their parameters, transform the video coordinate system into the intersection coordinate system with the center of the intersection as the origin, and use the two video detectors to output the actual spatial position of the pedestrian crossing once, and output the individual coordinate information L of the pedestrian. per ; Step 3: Based on the pedestrian individual coordinate information L obtained in Step 2 per , calculate the current pedestrian individual V per dir in matrix per 、a per 、v per Information, where dir per Indicates the direction of pedestrians, a per represents the pedestrian acceleration, v per Indicates the current speed of individual pedestrians; Step 4: Use deep learning methods to analyze and process the video images collected by the video detector, perform morphological perception and recognition on individual pedestrians, and classify pedestrians. per Classify and determine whether they belong to the vulnerable pedestrian group. If they are determined to belong to the vulnerable group, give o per Assign 1 and calculate the remaining crossing time t for the current disadvantaged pedestrian individual per ; Step 5: When the pedestrian crossing phase is green, use the video detectors at both ends of the zebra crossing to identify the morphology of all pedestrians. If there are weak pedestrians, activate the pedestrian green flash phase extension signal; before the pedestrian green flash starts, calculate the remaining crossing time of all weak pedestrians, generate a weak pedestrian group set, and take the largest v in the set per .t per The value is used as the remaining time for all vulnerable pedestrian groups to cross the street. If another vulnerable pedestrian individual enters the pedestrian identification area before the pedestrian flashes green, the remaining time for the largest vulnerable pedestrian group to cross the street is recalculated and the original remaining time is replaced. When the pedestrian phase flashes green, v per ·t per -t 行人绿闪相位时长 The result is transmitted to the signal machine as the green flash phase time that needs to be extended; Step 6: The output results of the video detectors in each direction are transmitted to the recognition assistance system through the gateway. After the system calculates the output results, the green flash extension time is transmitted to the traffic light through the network to protect vulnerable pedestrians crossing the street.

2. The method for controlling street crossing signals for vulnerable pedestrian groups according to claim 1, characterized in that: The pedestrian individual label dataset is shown in the following formula: V per =[o per ,dir per ,a per ,v per ,t per ,thick per ] Among them, per Indicates the pedestrian category, dir per Indicates the direction of pedestrians, a per represents the pedestrian acceleration, v per represents the current speed of the pedestrian, t per Indicates the remaining time for pedestrians to cross the street. If there is a secondary crossing island, if the pedestrian has not reached the secondary crossing island, the remaining time for the pedestrian to cross the street is the time to reach the secondary crossing island. If the pedestrian has passed the secondary crossing island, the remaining time for the pedestrian to cross the street is the time to reach the end of the zebra crossing. per is the pedestrian density coefficient around the pedestrian at the current moment.

3. The method for controlling street crossing signals for vulnerable pedestrian groups according to claim 2, characterized in that: In the intersection coordinate system, according to the plane coordinate x of the current pedestrian individual in the intersection coordinate system per ,y per The coordinate difference within unit time t is obtained by dir perx Value and dir pery , which is calculated as shown in the following formula: if and per |t2-y per |t1>0 you pery =1 else you pery =2 if x per |t2-x per |t1>0 you perx =3 else you perx =4 For north-south and east-west crossings, the v per , calculated according to the following formula: According to the above calculation v per , we can calculate the a of individual pedestrians when crossing the street in the north-south direction and the east-west direction. pery and a perx , the calculation formula is as follows:

4. The method for controlling street crossing signals for vulnerable pedestrian groups according to claim 1, characterized in that: The remaining pedestrian crossing time t per , the calculation process is shown in the following formula: In the formula, o per Determine the value for the vulnerable pedestrian individual, thick per is the density coefficient of the crowd around the vulnerable pedestrian, and The total length of north-south and east-west routes for individual vulnerable pedestrians crossing the street. is the length of the street currently walked, is the maximum moving speed of the vulnerable pedestrian within the current time t1.

5. The method for controlling street crossing signals for vulnerable pedestrian groups according to claim 1, characterized in that: The pedestrian recognition auxiliary system records the data of all pedestrians who have not been identified as vulnerable pedestrians, and records their crossing data matrix V per , analyze the crossing behavior and identify whether it is a strong non-vulnerable pedestrian individual. If it is not a strong non-vulnerable pedestrian, detect the pedestrian's morphological data and feed it back to the system as feedback input to improve the accuracy of the recognition system in identifying individuals in the vulnerable pedestrian group; in addition, the feedback data of the above identification at various intersections in the city is transmitted to the recognition system through the central gateway to expand its training sample set.