Automatic pedestrian signal extension system
The automatic pedestrian signal extension system addresses inefficiencies in conventional systems by using deep learning to identify vulnerable road users and dynamically adjust signals, ensuring safe and efficient pedestrian crossings while maintaining traffic flow.
Patent Information
- Application Number
- KR1020250088331
- Authority / Receiving Office
- KR · KR
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2025-07-02
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2045-07-02
AI Technical Summary
Conventional pedestrian signal control systems fail to account for individual differences in mobility ability, such as age, disability, and use of mobility aids, leading to inefficient or unsafe pedestrian signal extensions that can disrupt overall traffic flow.
An automatic pedestrian signal extension system that uses deep learning-based object detection and analysis to identify vulnerable road users, dynamically extending signal time based on individual characteristics and environmental conditions, while adjusting traffic flow to maintain safety and efficiency.
Precisely determines vulnerable road users and dynamically extends pedestrian signals, ensuring safe crossing and optimizing traffic flow by considering individual needs and environmental factors, thereby enhancing pedestrian safety and minimizing traffic disruptions.
Smart Images

Figure 112025074397900-PAT00001_ABST
Abstract
Description
Technology Field
[0001] The present invention relates to an automatic pedestrian signal extension system. Background Technology
[0003] Pedestrian safety is one of the most fundamental and important elements in the urban traffic environment, and to this end, crosswalks equipped with traffic lights are operated in a way that guarantees pedestrians the right of way for a certain period of time.
[0004] In particular, since crosswalks at intersections or two-way roads are points where vehicles and pedestrians approach simultaneously, control through a signal system is essential, and the proper provision of pedestrian signals is directly linked to accident prevention.
[0005] Conventional pedestrian signal control methods are mostly fixed-time systems, based on a structure that provides the same signal time to all pedestrians.
[0006] This method is based on the average walking speed set for a typical adult (e.g., 1.2 to 1.5 m / s) and has structural limitations in that it cannot take into account individual differences in mobility ability, such as the pedestrian's age, disability, and use of mobility aids.
[0007] To complement this, there are cases where manual request methods via traffic light buttons or simple object detection methods using sensors have been applied; however, even in these cases, they merely detect the presence or absence of a pedestrian and do not provide the functionality to determine whether the pedestrian can actually pass within the standard time.
[0008] In addition, there are cases where signals are unnecessarily extended or, conversely, necessary extensions are not implemented due to a failure to consider complex factors such as the number of pedestrians or movement patterns.
[0009] Meanwhile, since crosswalks are not limited to affecting only a single intersection but are organically connected to adjacent intersections and the overall traffic flow, the extension or reduction of pedestrian signals is not merely confined to the specific crosswalk but can expand into an issue of integration with the entire urban signaling network or smart transportation systems.
[0010] In other words, a precise and balanced signal control system is required to ensure the safety of vulnerable road users while not hindering the overall traffic flow. Prior art literature
[0012] Korean Patent Publication No. 10-2185100 (December 1, 2020) The problem to be solved
[0013] The present invention provides an automatic pedestrian signal extension system capable of supporting the safe crossing of vulnerable road users by analyzing a pedestrian's shape, movement pattern, speed, and the presence or absence of additional equipment to precisely determine whether the pedestrian is a vulnerable road user, and dynamically extending the pedestrian signal time based on the determination result. means of solving the problem
[0015] According to one aspect of the present invention, a pedestrian signal automatic extension system is provided, comprising: a pedestrian detection unit that detects a pedestrian waiting to cross a crosswalk or crossing a crosswalk; a vulnerable road user determination unit that determines a vulnerable road user by analyzing the pedestrian's shape, movement pattern, movement speed, and the presence or absence of additional equipment; a pedestrian signal control unit that grants additional pedestrian signal time to a standard pedestrian signal time depending on whether the pedestrian determined to be a vulnerable road user crosses the crosswalk; and a traffic flow adjustment unit that adjusts traffic flow in conjunction with a vehicle signal when the pedestrian signal is extended.
[0016] The pedestrian detection unit can detect objects including pedestrians, strollers, wheelchairs, and walking aids within a detection area using an algorithm including at least one of YOLO, SSD, or Mask R-CNN among deep learning-based object recognition algorithms, and analyze the location, size, contour, and direction of movement of each object.
[0017] The road vulnerable determination unit may determine a pedestrian as a road vulnerable by analyzing movement characteristics including the pedestrian's cumulative distance traveled over time, the rate of change in movement speed, and the regularity of the movement trajectory, if the movement speed is below a preset speed or the change in the movement trajectory exceeds a preset standard.
[0018] The road vulnerable determination unit detects auxiliary equipment including wheels located together with the pedestrian among the shapes of the pedestrian, and if the auxiliary equipment corresponds to a stroller or wheelchair, the pedestrian can be determined to be a road vulnerable.
[0019] The pedestrian signal control unit may grant additional pedestrian signal time during nighttime or inclement weather by referring to external environmental information, including the time of day and weather information, in addition to whether the pedestrian is a vulnerable road user.
[0020] The pedestrian signal control unit can grant additional pedestrian signal time when more than a preset number of vulnerable road users are detected within the same detection area.
[0021] The pedestrian signal control unit can differentially allocate additional pedestrian signal time according to the type of pedestrian with a disability.
[0022] The pedestrian detection unit detects abnormal behaviors, including stopping, falling, and moving in the reverse direction, within the crosswalk, and the pedestrian signal control unit may grant additional pedestrian signal time to ensure the safety of the pedestrian.
[0023] The pedestrian signal control unit can restore the balance of the entire signal cycle by reducing a portion of the standard pedestrian signal time in the next pedestrian signal cycle when no vulnerable road user is detected after the extended pedestrian signal is applied based on the additional pedestrian signal time granted due to a vulnerable road user.
[0024] The pedestrian signal control unit may include an AI learning module that predicts pedestrian patterns according to specific regions and time periods by cumulatively learning the shape, movement pattern, movement speed, and presence or absence of additional equipment of pedestrians, and preemptively applies an extension if it is determined that additional pedestrian signal time will be needed in advance. Effects of the invention
[0026] According to the present invention, the shape, movement pattern, speed, and presence or absence of additional equipment of a pedestrian are analyzed to precisely determine whether the pedestrian is a vulnerable road user, and the pedestrian signal time is dynamically extended based on the result of this determination, thereby supporting the safe crossing of vulnerable road users. Brief explanation of the drawing
[0028] FIG. 1 is a diagram of the relationship of an automatic pedestrian signal extension system according to one embodiment of the present invention. FIG. 2 is a diagram showing each configuration of a pedestrian signal automatic extension system according to one embodiment of the present invention. FIG. 3 is a flowchart of determining a traffic vulnerable person and deciding whether to extend a pedestrian signal in an automatic pedestrian signal extension system according to an embodiment of the present invention. Specific details for implementing the invention
[0029] The present invention is capable of various modifications and may have various embodiments; specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, and it should be understood that it includes all modifications, equivalents, and substitutions that fall within the spirit and scope of the present invention. In describing the present invention, detailed descriptions of related prior art are omitted if it is determined that such detailed descriptions may obscure the essence of the present invention.
[0030] Terms such as "first," "second," etc., may be used to describe various components, but said components should not be limited by said terms. These terms are used solely for the purpose of distinguishing one component from another.
[0031] The terms used in this application are used merely to describe specific embodiments and are not intended to limit the invention. The singular expression includes the plural expression unless the context clearly indicates otherwise. In this application, terms such as "comprising" or "having" are intended to specify the presence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.
[0032] In addition, regarding the contact relationship between each component, the term "combination" is used not only to mean cases where each component is in direct physical contact with each other, but also to encompass cases where another component is interposed between each component and each component is in contact with that other component.
[0033] Hereinafter, an embodiment of the pedestrian signal automatic extension system (100) according to the present invention will be described in detail with reference to the attached drawings. In describing with reference to the attached drawings, identical or corresponding components are given the same reference numerals, and redundant descriptions thereof will be omitted.
[0035] According to one embodiment of the present invention, a pedestrian signal automatic extension system (100) is provided, comprising: a pedestrian (10) detection unit (110) for detecting a pedestrian (10) waiting to cross a crosswalk or crossing a crosswalk; a traffic vulnerable person determination unit (120) for determining a traffic vulnerable person (20) by analyzing the shape, movement pattern, movement speed, and presence or absence of additional equipment of the pedestrian (10); a pedestrian signal control unit (130) for granting an additional pedestrian signal time (30) to a standard pedestrian signal time (30) depending on whether the pedestrian (10) determined to be a traffic vulnerable person (20) crosses the crosswalk; and a traffic flow adjustment unit (140) for adjusting the traffic flow in conjunction with a vehicle signal (40) when the pedestrian signal (30) is extended.
[0036] The pedestrian signal automatic extension system (100) according to the present embodiment analyzes in real time the shape, movement pattern, speed, and presence or absence of additional equipment of a pedestrian (10) using a crosswalk, and determines whether the pedestrian is a person with a disability (20) based on this, and automatically extends the pedestrian signal (30) time, thereby supporting the person with a disability (20) to safely pass through the crosswalk within the standard signal time.
[0037] This embodiment utilizes deep learning-based object detection technology to accurately detect objects including not only pedestrians (10) but also additional equipment such as strollers, wheelchairs, walking aids (including crutches and canes), and umbrellas, and can precisely determine the characteristics of the road-vulnerable person (20) by comprehensively analyzing the speed, trajectory change, and presence of accompanying equipment of the detected objects.
[0038] Based on the results of this judgment, additional pedestrian signal time (30) may be assigned in a weighted or differential manner according to the number or type of pedestrians (20), and even when the mobility of pedestrians (10) is expected to decrease due to external environmental conditions such as nighttime or bad weather, the pedestrian signal control unit (130) may flexibly adjust the signal time by reflecting this.
[0039] In addition, this embodiment can detect abnormal behavior such as a pedestrian (10) stopping, falling, or moving in the reverse direction while crossing, and in this case, it may automatically provide additional pedestrian signal time (30) or perform follow-up measures such as a warning or a request for rescue if necessary.
[0040] Furthermore, if there is no detection result of a pedestrian (10) or if a person with a disability (20) is not actually detected, the system may autonomously restore the balance of the overall signal flow by partially reducing the reference pedestrian signal (30) time in the next signal cycle to correct the signal system that has become unbalanced due to the previous signal extension.
[0041] In addition, the pedestrian signal automatic extension system (100) of the present embodiment operates in conjunction with traffic flow information of an adjacent intersection to minimize the impact of the pedestrian signal (30) extension on vehicle flow and can improve the overall efficiency of the smart transportation system.
[0042] Therefore, this embodiment can simultaneously achieve pedestrian (10) protection and traffic flow optimization by simultaneously solving various technical challenges such as ensuring the practical safety of vulnerable road users (20), preventing unnecessary signal extension, controlling the recovery of pedestrian signals (30), and strengthening connectivity between signals.
[0043] Each component of the pedestrian signal automatic extension system (100) according to the present embodiment will be described in more detail below with reference to FIGS. 1 to 3.
[0045] First, as illustrated in FIG. 2, the pedestrian (10) detection unit (110) is one of the core components of the pedestrian signal automatic extension system (100) according to the present embodiment, and can perform the role of detecting in real time from the time when a pedestrian (10) enters or attempts to enter the crosswalk, and providing basic data for subsequent determination of a vulnerable road user (20) and control of the pedestrian signal (30).
[0046] More specifically, the pedestrian (10) detection unit (110) may include a camera (sensor) (112) and an object recognition algorithm module (114), which may be interconnected to enable high-speed processing and accurate analysis.
[0047] The camera (sensor) (112) may be implemented as a fixed camera or intelligent CCTV system installed on the top or side of the crosswalk, and in some embodiments, may be integrated with a traffic signal or pedestrian signal (10).
[0048] The detection angle of the camera (sensor) (112) can be adjusted to capture the front and side of the crosswalk simultaneously, and an infrared (IR) camera or a thermal imaging camera may be used in combination to obtain a suitable image even at night or in bad weather conditions.
[0049] These cameras (sensors) (112) may include a high-resolution lens for object tracking and a real-time streaming function.
[0050] Additionally, when sensors other than the camera are used, composite sensors consisting of LiDAR, ultrasonic, radar sensors, etc., may be used, and they can be used in conjunction with the camera to improve object detection accuracy.
[0051] For example, since objects with low size or reflectivity, such as strollers or wheelchairs, may be difficult to recognize with an optical camera alone, the combined use of the aforementioned supplementary sensor can contribute to maximizing detection reliability.
[0052] The pedestrian (10) detection unit (110) may be systematically designed to process video data in real time for privacy protection, extract and store or transmit only the object recognition result, and automatically discard the original video after a certain period of time.
[0053] Meanwhile, the object recognition algorithm module (114) is composed of a deep learning-based learning network and can utilize an algorithm that includes at least one of the following algorithms: YOLO (You Only Look Once), SSD (Single Shot MultiBox Detector), or Mask R-CNN (Region-based Convolutional Neural Network).
[0054] This object recognition algorithm module (114) can classify and recognize various objects, including people, strollers, wheelchairs, walking aids (including crutches and canes), in video frames input from a camera (sensor) (112) at high speed.
[0055] In addition, the trained neural network identifies objects based on their contours, sizes, and motion patterns, and post-processing techniques such as Non-Maximum Suppression (NMS) can be applied in parallel to prevent interference between objects.
[0056] In addition, the pedestrian (10) detection unit (110) can extract not only the presence of a simple object, but also the location coordinates (x, y) of each object, the direction and speed of movement in frames, and the relative distance information from other objects, and this information can then be transmitted to the road vulnerable person determination unit (120) and the pedestrian signal control unit (130).
[0057] In another embodiment, the pedestrian (10) detection unit (110) can maintain consistency in tracking by removing duplicates between objects and assigning a unique identifier to each object when multiple objects are detected simultaneously.
[0058] Meanwhile, the pedestrian (10) detection unit (110) can predict the forward behavior of the pedestrian (10). For example, an object that is loitering around a crosswalk and is not moving can be judged as not having started crossing yet, and an object that shows a pattern of gradually moving toward the edge of the road can be judged as having an intention to enter.
[0059] This forward behavior prediction may also be used to determine whether to extend the prior signal in the pedestrian signal control unit (130).
[0060] As another example, when the detected pedestrian (10) detection unit (110) detects a pedestrian (10) and shows behaviors such as stopping or changing direction while crossing, it can transmit the current situation to the pedestrian signal control unit (130) and the traffic flow adjustment unit (140) to temporarily suspend the vehicle signal and simultaneously transmit a rescue signal to nearby vehicles and the road control center while adjusting the traffic flow.
[0061] As illustrated in FIG. 2, the traffic vulnerable person determination unit (120) can determine whether the pedestrian (10) is a traffic vulnerable person (20) by analyzing object data received from the pedestrian (10) detection unit (110) within the pedestrian signal automatic extension system (100).
[0062] The judgment result of the traffic vulnerable person judgment unit (120) serves as a key criterion for determining whether to extend the pedestrian signal (30) and calculating the extension time, and the accuracy of the judgment can be directly linked to the reliability of the entire system.
[0063] In this embodiment, the transportation vulnerable person (20) refers to people who have difficulty passing through a crosswalk within a limited time due to reasons such as moving at a relatively slower speed compared to a general adult pedestrian (10) or using a means of mobility assistance.
[0064] For example, the transportation vulnerable (20) may include the elderly, wheelchair users, guardians with strollers, and users of walking aids (including crutches and canes).
[0065] More specifically, as illustrated in FIG. 2, the traffic vulnerable determination unit (120) may include a walking characteristic analysis module and an additional equipment recognition module (126).
[0066] The walking characteristic analysis module can analyze the location data of the pedestrian (10) cumulatively over time to analyze the distance traveled per object, the change in speed between frames, the linearity of the trajectory, and the pattern of change in direction.
[0067] The pedestrian characteristics analysis module can determine that the object is a person with a disability (20) if the average movement speed is maintained below a certain standard (e.g., 0.6 m / s) or if irregular changes in direction are repeated in the movement trajectory.
[0068] In addition, if the stopping state persists for a certain period of time or if repeated stopping and restarting is detected in the middle of walking, it is determined that there is a lack of consistency in walking and that there is a high possibility that the person is a person with a disability (20), and thus the person can be evaluated as a person with a disability (20).
[0069] The additional equipment recognition module (126) can determine whether additional equipment, such as a stroller, wheelchair, walking aid (including crutches and canes), is present by analyzing shape data around the pedestrian (10) detected during the object recognition process.
[0070] The additional equipment recognition module (126) can apply an instance splitting algorithm to separate the boundary between the pedestrian (10) and the equipment, and analyze their relative positions and whether they are moving continuously to determine whether they are physically connected.
[0071] For example, a user who operates a wheelchair can make a highly reliable judgment because the pedestrian (10) object and the wheelchair object move continuously at the same speed and in the same direction, and are located in spatially adjacent to each other.
[0072] For example, the result of determining the pedestrian (10) may be assigned tags such as "wheelchair user," "guardian accompanying a stroller," and "user of a walking aid (including crutches and canes)" along with a unique identifier for each pedestrian (10), and this information may be transmitted to the pedestrian signal control unit (130) to enable differential extension of the pedestrian signal (30) time according to conditions.
[0073] The tag is updated in real time depending on the situation, and for example, even if it is classified as a normal pedestrian (10) at the initial detection, it may be reclassified as a vulnerable road user (20) if the walking speed decreases rapidly or an abnormal trajectory is observed.
[0074] Additionally, the traffic vulnerable person determination unit (120) can be configured to operate by lowering the conditions in an environment where the likelihood of traffic vulnerable people (20) appearing is high by referring to region and time zone data, thereby including all traffic vulnerable people (20).
[0075] For example, the weight for detecting stroller users can be increased at an intersection near a daycare center around 8 a.m., and the sensitivity for detecting users of mobility aids (including crutches and canes) or wheelchair users can be increased around a senior welfare center after 5 p.m.
[0076] In addition to these children and the elderly, for example, an object that repeatedly moves in a zigzag pattern may be detected, and based on the irregular pattern of such movement, the possibility of cognitive or physical disability may be estimated, and the object may be determined to be a person with a disability (20) that is not included in the existing classification of persons with disabilities (20).
[0077] In this way, the traffic vulnerable person judgment unit (120) can function as a core component that simultaneously satisfies the protection of pedestrians (10) and the harmony of traffic flow by precisely analyzing the characteristics of pedestrians (10) in various situations and making flexible judgments based on real-time data.
[0078] These pedestrian (10) detection unit (110) and traffic vulnerable person determination unit (120) can determine a traffic vulnerable person (20) through a process as shown in FIG. 3 and transmit information to the pedestrian signal control unit (130).
[0079] As illustrated in FIG. 2, the pedestrian signal control unit (130) can determine whether to extend the pedestrian signal (30) time and the extension time based on the pedestrian (10) status information received from the traffic vulnerable person determination unit (120) and the traffic vulnerable person (20) determination result information, and reflect this in the actual signal control.
[0080] The pedestrian signal control unit (130) can determine whether to extend a certain amount of time during the standard pedestrian signal (30) time based on the pedestrian (10) status tag (e.g., wheelchair user, caregiver accompanying a stroller, etc.) and movement speed information received from the pedestrian vulnerable person determination unit (120).
[0081] For example, if the detected pedestrian (10) is a wheelchair user, and shows a speed of 0.5 m / s which is significantly lower than the standard walking speed (1.2 m / s), an additional +5 seconds of time may be granted.
[0082] Meanwhile, the pedestrian signal control unit (130) may include a time weight calculation module and an AI learning module (134).
[0083] The weight calculation module can apply weights to the extended time based on external environmental information such as time zone, weather conditions, regional characteristics, and the number of pedestrians (10).
[0084] For example, during the nighttime (21:00 to 06:00) hours, an additional extension of +5 seconds can be automatically applied to the basic extension time, and on rainy or snowy days, an additional extension of +3 seconds can be applied. Also, if there are 3 or more people with disabilities (20) within the same detection area, an additional +2 seconds can be applied, and flexible extensions according to various conditions are possible.
[0085] In addition, the pedestrian signal control unit (130) learns pedestrian pattern data that occurs repeatedly in a specific time period and area through an AI learning module (134), and may apply an extension of the pedestrian signal (30) in advance if the appearance of a person with a disability (20) is predicted even before a pedestrian (10) enters.
[0086] For example, if a pattern is accumulated showing a high frequency of pedestrians with strollers (10) at the crosswalk in front of the kindergarten between 3 PM and 4 PM, then +5 seconds may be added to the standard signal preemptively during that time period.
[0087] The pedestrian signal control unit (130) can restore the balance of the overall traffic flow by partially reducing the reference pedestrian signal (30) time in the next signal cycle to prevent errors from occurring in the relationship between the previously extended pedestrian signal (30) and other vehicle signals in the nearby area.
[0088] For example, if there are no pedestrians crossing the crosswalk or no vulnerable road users (20), the signal may be omitted at the next pedestrian signal (30), or the signal may be partially reduced within the standard time to compensate for the traffic flow. This process operates in conjunction with adjacent intersection signal information, thereby contributing to maintaining the synchronization of the entire city traffic network.
[0089] The pedestrian signal control unit (130) also tracks the crossing status of the pedestrian (10) in real time, and if abnormal behavior (stopping, falling, reversing direction, etc.) is detected, it can automatically apply an extension of the pedestrian signal (30) immediately or simultaneously transmit a rescue request signal.
[0090] Furthermore, the pedestrian signal control unit (130) can set an upper limit on the extension time to minimize the impact of the signal extension on the actual traffic flow.
[0091] For example, the extension is allowed only within a range not exceeding 10 seconds from the standard pedestrian signal (30) time, thereby maintaining a balance with the traffic flow. This upper limit time can be dynamically adjusted by collecting factors such as intersection congestion, vehicle queue length, and vehicle signal cycle in real time.
[0092] The pedestrian signal control unit (130) can also finely adjust the extension time by taking into account the physical length of the crosswalk, the density of pedestrians (10), and the predicted time to complete crossing. For example, if the length of the crosswalk is 30 m and the average speed of the detected pedestrians (10) is 0.7 m / s, the minimum time required for the pedestrians (10) to completely cross is calculated to be about 43 seconds, so it is possible to quantitatively calculate that at least 8 seconds of additional extension is required when the standard signal time is 35 seconds. This calculation is performed automatically within the control algorithm.
[0093] In this way, the pedestrian signal control unit (130) can simultaneously achieve safety for pedestrians (10) and harmony of the entire city's traffic flow based on a multi-layered judgment system that encompasses real-time detection results of vulnerable road users (20), external environmental conditions, prediction-based signal control, and post-recovery control.
[0094] As illustrated in FIG. 2, the traffic flow adjustment unit (140) is not limited to a single crosswalk, but can adjust the traffic flow by analyzing in real time the impact of the pedestrian signal (30) on adjacent intersections and the overall traffic flow, thereby ensuring interoperability and consistency between each signal cycle.
[0095] More specifically, the traffic flow control unit (140) can determine whether to allow signal extension and the upper limit of the extension time through real-time congestion detection, and to do so, it collects data from traffic sensors that detect the length of the vehicle queue in the road, the number of vehicles, the flow of turning vehicles, and the waiting time, and if the data exceeds a reference value, it can limit the extension of the pedestrian signal (30) or, if necessary, open the vehicle signal first to minimize the imbalance in the flow.
[0096] And, the traffic flow adjustment unit (140) can automatically correct the vehicle signal offset of an adjacent intersection according to the time change of the entire signal cycle that occurs after the pedestrian signal (30) is extended.
[0097] For example, if the pedestrian signal (30) is extended by 8 seconds due to the crossing of a specific group of pedestrians (10), the adjustment unit can slightly adjust the timing of vehicle entry into adjacent intersections to prevent sequential congestion and maintain driving consistency throughout the entire road network.
[0098] The traffic flow control unit (140) can separately open a portion of the vehicle signal to quickly restore traffic flow after the pedestrian (10) has finished crossing the crosswalk.
[0099] The traffic flow control unit (140) can be operated by partially opening only the vehicle signal in a specific direction when only pedestrians (10) in a specific direction are detected at a two-way crosswalk or when pedestrians have completed walking in one direction, and this method enables faster recovery of vehicle flow than the entire signal switching.
[0100] For example, assume a case where a pedestrian (10) is detected at a complex intersection in downtown Seoul and the pedestrian signal (30) is extended by 10 seconds beyond the standard time. In this case, if a vehicle queue is formed of 50m or more and the number of vehicles exceeds the standard value of 30, the traffic flow control unit (140) may limit the extension of the pedestrian signal (30) to a maximum of 5 seconds and synchronize the vehicle signal to open it first to some directions.
[0101] In addition, the traffic flow control unit (140), together with the pedestrian signal control unit (130), can perform learning-based control in a way that if the extension of the pedestrian signal (30) occurs repeatedly but the pedestrian vulnerable person (20) is not actually detected or is determined to be a prediction failure, the extension data is stored in an internal recording system, and subsequently, the extension of the pedestrian signal (30) at the same time is automatically suppressed or the upper limit of the extension time is lowered from 10 seconds to 5 seconds.
[0102] In this way, the traffic flow control unit (140) according to the present invention can secure stability and flexibility of the entire traffic system by predicting complex situations that may occur in various intersection environments within the city based on the real-time status of vehicles and pedestrians (10) and controlling the pedestrian signal (30) and vehicle signal (40) in conjunction accordingly.
[0104] In addition, the technical details described above may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., either individually or in combination. The program instructions recorded on the medium may be those specifically designed and configured for the embodiments, or they may be those known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include programming language code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. Hardware devices may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.
[0105] Although an embodiment of the present invention has been described above, those skilled in the art may modify and change the present invention in various ways by adding, changing, deleting, or adding components, etc., without departing from the spirit of the present invention as described in the claims, and such modifications and changes are also to be included within the scope of the rights of the present invention. Explanation of the symbols
[0107] 10: Pedestrian 20: People with disabilities 30: Pedestrian signal 40: Vehicle signal 100: Automatic pedestrian signal extension system 110: Pedestrian detection unit 112: Camera (Sensor) 114: Object Recognition Algorithm Module 120: Determination of Transportation Vulnerable Persons 122: Speed Analysis Module 124: Movement Trajectory Analysis Module 126: Additional Equipment Recognition Module 130: Pedestrian signal control unit 132: Time Weight Judgment Module 134: AI Learning Module 140: Traffic Flow Control Division 142: Adjacent Signal Interlocking Module
Claims
Claim 1 A pedestrian detection unit that detects pedestrians waiting to cross a crosswalk or crossing a crosswalk; a vulnerable road user determination unit that determines a vulnerable road user by analyzing the shape, movement pattern, movement speed, and presence or absence of additional equipment of said pedestrian; and a pedestrian signal control unit that grants additional pedestrian signal time to the standard pedestrian signal time depending on whether said pedestrian determined to be a vulnerable road user crosses the crosswalk.The system includes a traffic flow adjustment unit that adjusts traffic flow in conjunction with vehicle signals when pedestrian signals are extended; the vulnerable road user determination unit dynamically adjusts detection sensitivity for determining vulnerable road users in environments with a high probability of appearance by differentially adjusting detection weights for each type of vulnerable road user based on regional and time-of-day data; the pedestrian signal control unit controls the system to restore balance to the overall signal cycle by reducing a portion of the standard pedestrian signal time in the next pedestrian signal cycle if no vulnerable road user is detected after the extended pedestrian signal is applied based on the additional pedestrian signal time granted due to the vulnerable road user; the pedestrian signal control unit includes an AI learning module that predicts the probability of appearance of vulnerable road users by continuously accumulating pattern data regarding pedestrian shapes, movement patterns, movement speeds, and the presence or absence of additional equipment that occur repeatedly in specific regions and time-of-days during actual operation, and preemptively applies the extension if it is determined that additional pedestrian signal time is required in advance; the traffic flow adjustment unit is linked to both cases where the additional pedestrian signal time is granted by the pedestrian signal control unit and cases where the standard pedestrian signal time is reduced, and corresponds to the time corresponding to the extension or reduction of the pedestrian signal. A pedestrian signal automatic extension system that preserves the total signal cycle duration by adjusting vehicle signals at a crosswalk, and the traffic flow adjustment unit automatically adjusts the vehicle signal offset of an adjacent intersection based on the preservation of the total signal cycle duration. In the case where the additional pedestrian signal time is preemptively applied by the AI learning module together with the pedestrian signal control unit but no vulnerable road users are actually detected, the traffic flow adjustment unit stores the extension data in an internal recording system and subsequently performs learning-based control to automatically suppress pedestrian signal extension at the same time or lower the upper limit of the extension time. Claim 2 A pedestrian signal automatic extension system according to claim 1, wherein the pedestrian detection unit detects objects including the pedestrian, stroller, wheelchair, and walking aid within a detection area using an algorithm including at least one of YOLO, SSD, or Mask R-CNN among deep learning-based object recognition algorithms, and analyzes the position, size, contour, and direction of movement of each object. Claim 3 In paragraph 2, the pedestrian signal automatic extension system, wherein the pedestrian vulnerable determination unit analyzes movement characteristics including the pedestrian's cumulative travel distance over time, rate of change in travel speed, and regularity of the travel trajectory, and determines the pedestrian as a pedestrian vulnerable if the travel speed is below a preset speed or the change in the travel trajectory exceeds a preset standard. Claim 4 In paragraph 2, the pedestrian signal automatic extension system, wherein the above-mentioned vulnerable road user determination unit detects the above-mentioned auxiliary equipment including wheels located together with the pedestrian among the shapes of the pedestrian, and determines the pedestrian as a vulnerable road user if the said auxiliary equipment corresponds to a stroller or wheelchair. Claim 5 In paragraph 2, the pedestrian signal control unit refers to external environment information including the time of day and weather information in addition to whether the pedestrian is a traffic-vulnerable person, and grants the additional pedestrian signal time by adding weight in the case of nighttime or bad weather. This is an automatic pedestrian signal extension system. Claim 6 In paragraph 2, the pedestrian signal control unit is a pedestrian signal automatic extension system that grants the additional pedestrian signal time by increasing the weight when more than a preset number of vulnerable road users are detected within the same detection area. Claim 7 In paragraph 2, the pedestrian signal control unit is a pedestrian signal automatic extension system that differentially grants additional pedestrian signal time according to the type of pedestrian with a traffic disability. Claim 8 In paragraph 2, the pedestrian detection unit detects when the pedestrian exhibits abnormal behavior including stopping, falling, and moving in the reverse direction within the crosswalk, and the pedestrian signal control unit grants additional pedestrian signal time to ensure the safety of the pedestrian, in a pedestrian signal automatic extension system. Claim 9 delete Claim 10 delete
Citation Information
Patent Citations
Public transport priority control method and device
CN106683452A
Smart crosswalk controlling system
KR1020180063643A
Method and System for Detecting Weak Walking Person Based on Behavioral Cognition using Machine Learning Technique
KR1020190051128A
Method, apparatus and program for traffic signal control for pedestrians with limited mobility
KR102827836B1
