A method, system and medium for warning abnormal status of bus stop parking area

Abstract

The present invention relates to an early warning method, system, and medium for abnormal conditions in bus stop parking areas. The method comprises: U1. The V2X OBU in the unmanned bus system receives the longitude and latitude information of the center point of the station parking space transmitted by the V2X RSU within the direct communication range of the PC5 interface. The V2X OBU calculates and matches the position information of the unmanned bus and the center point of the station parking space in real time based on its own GPS module. If the distance difference is within 100m, it is determined that the unmanned bus is about to enter the station, and then performs abnormal state detection of the parking space to determine whether there is an abnormal state. The present invention not only integrates multi-source data such as roadside lidar and cameras to obtain bus stop parking area information from different angles and dimensions, but also calculates the dwell time and density of pedestrians in pedestrian occupancy detection, greatly reducing misjudgments and missed judgments. Compared with traditional single detection methods, the accuracy of abnormality detection is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of abnormality early warning for bus stop parking areas, and in particular to an early warning method, system and medium for abnormal conditions in bus stop parking areas. Background Art

[0002] With the continuous development of unmanned driving technology, smart bus stops are important places for buses to stop and passengers to get on and off. The normal operation of their parking areas is crucial to the efficient operation of the bus system.

[0003] However, in actual operation, various abnormal conditions often occur in bus stop parking areas, such as vehicles occupying parking spaces for a long time, non-bus vehicles parking illegally, and some passengers and pedestrians occupying parking spaces. These abnormal conditions will not only affect the normal parking and departure of buses, causing bus operation delays and reducing bus service quality, but may also cause traffic congestion and affect the overall smoothness of urban traffic.

[0004] In the prior art, a Chinese patent (application number: 202310270205.0, publication number: CN 116386328A) discloses an intelligent bus control system and a control method thereof, including an intelligent bus operation data database module, which is used to collect in real time the intelligent bus operation monitoring data, monitoring intelligent bus operation status data and guidance continuity valid data that need to be monitored in real time and encrypt them into an initial information set and send them to the intelligent bus operation data different analysis model modules; the intelligent bus operation data different analysis model modules are provided with multiple bus operation feature analysis models, and a feature interleaving algorithm is provided between the multiple bus operation feature analysis models. The initial information set is transmitted to the bus operation feature analysis model, and the initial information set is transmitted through the feature interleaving algorithm, and then The initial information set is then sent to the remote cloud-based human-computer interaction module through the bus operation characteristic analysis model. The remote cloud-based human-computer interaction module includes an intelligent bus operation history archive containing intelligent bus operation characteristic data. The remote cloud-based human-computer interaction module performs optimization model data calculation and optimization on the initial information set and the intelligent bus operation characteristic data to obtain intelligent bus operation data for different paths. This intelligent bus operation data is then transmitted to the bus operation characteristic analysis model via the characteristic interleaving algorithm and then to the intelligent bus operation path adjustment module. This solution does not analyze and determine whether pedestrians occupy bus parking spaces in bus parking areas, which affects bus entry into the station and reduces the overall efficiency of smart bus operations.

[0005] In the prior art, the patent (application number: 202110796785.8, publication number: CN 113538887A) discloses an intelligent bus terminal system. Step 1: Establish a network model of the intelligent bus system based on the network hierarchy characteristics of the Internet of Things; Step 2: Design a bus hardware system based on the intelligent bus system network model established in step 1, and write an intelligent bus software.

[0006] The patent application is designed to create a hardware system and install it on each intelligent bus system platform. Step 3: The intelligent bus system platform obtains data from the bus system through the hardware system designed in Step 2 and uses the software system to send and receive data between platforms as needed. Step 4: The intelligent bus system processes data between platforms, enabling the bus dispatch center to implement intelligent bus dispatching, buses to obtain travel data change information, and bus stops and passenger clients to obtain and display information. This patent has weak data processing and analysis capabilities, performing only simple processing for the bus dispatch system. It lacks the ability to comprehensively analyze large amounts of data and cannot provide more valuable information for bus operations. Summary of the Invention

[0007] In view of the above shortcomings of the existing technology, the present invention provides an early warning method, system and medium for abnormal conditions in bus stop parking areas. It not only integrates multi-source data such as roadside lidar and cameras to obtain bus stop parking area information from different angles and dimensions, but also calculates the pedestrian stay time and density in pedestrian occupancy detection, greatly reducing misjudgments and missed detections. Compared with traditional single detection methods, the accuracy of anomaly detection is greatly improved.

[0008] In order to achieve the above-mentioned and other related purposes, the present invention provides the following technical solutions:

[0009] A method for early warning of abnormal conditions in a bus stop parking area, the method comprising:

[0010] U1. The V2X OBU in the unmanned bus system is within the direct communication range of the PC5 interface and receives the longitude and latitude information of the center point of the station parking space transmitted by the V2X RSU. The V2X OBU uses its built-in GPS module to calculate and match the location of the unmanned bus with the center point of the station parking space in real time. If the distance difference is within 100 meters, it determines that the unmanned bus is about to enter the station.

[0011] The U2.V2X OBU transmits the upcoming arrival information to the station-train collaborative control platform, and the V2X RSU forwards it to the MEC. After receiving the upcoming arrival information, the MEC performs parking space abnormality detection based on the raw data of parking space obstacles received and stored in real time by the platform sensor system to determine whether there is any abnormality.

[0012] U3. When abnormal parking space status information is detected in step U2, the MEC directly transmits the information to the V2X OBU via the V2X RSU. The V2X OBU transmits the parking space status information to the autonomous driving controller.

[0013] U4. When step U3 obtains information about abnormal parking space status, the information is transmitted to the station-vehicle collaborative control platform and the station-vehicle collaborative management platform via 4G / 5G. By statistically analyzing the pedestrian parking time, a hierarchical early warning strategy is implemented.

[0014] Furthermore, in step U1, determining that the unmanned public transport vehicle is about to enter the station includes:

[0015] U11. The longitude and latitude of the parking space center is expressed as P1=(γ1,θ1), then the parking space center and other points P x =(γ x ,θ x ) of the distance function d(P1,P x )for,

[0016] ,

[0017] Among them, γ1 is longitude, θ1 is latitude, γ x is the longitude of other points, θ x is the latitude of other points;

[0018] U12. Get the location B of the unmanned bus at different time points t t =(γ b (t),θ b (t)), the position of V2X RSU Q = (γ r ,θ r ), the coverage of V2X RSU is D r =0.3-0.5km, the coverage range varies depending on the actual working conditions;

[0019] U13.V2X RSU communication range determination:

[0020] ,

[0021] Among them, when in_range Bt,Q =1, the V2X OBU determines the distance to the station based on the received coordinates of the center point of the platform parking space:

[0022] ,

[0023] Among them, D ɑ =0.1km, which is the threshold distance for entering the station. The value can be manually adjusted dynamically according to the actual working conditions.Bt,P1 =1 indicates that the unmanned bus is about to arrive at the station.

[0024] Furthermore, in step U2, the abnormal state detection of the parking space and determining whether there is an abnormal state include:

[0025] U21. Based on the real-time reception and storage of the original data of the parking space obstacle of the platform sensor system, the original laser radar point cloud data information and the camera image data information are obtained and preprocessed to obtain the preprocessed laser radar point cloud data information and the camera image data information;

[0026] U22. Based on the preprocessed lidar point cloud data and camera image data, an improved Hungarian algorithm is used to detect and continuously track pedestrian targets to obtain pedestrian target recognition and tracking results;

[0027] U23. Based on the results of pedestrian target recognition and tracking, construct a pedestrian stay time function T in the parking area stay (i, t) and the density function of the point cloud in the parking area ρ(W, t),

[0028] ,

[0029] ,

[0030] Among them, S is the indicator function, which uses 0-1 to quantify the state, i is a positive integer, t is a different time point, T is the time window, in_region() is the judgment function of whether the pedestrian is in the parking area, p i is the result of pedestrian target recognition and tracking, n is the sample size, W is the point cloud, area(W) is the point cloud in the parking area, and the pedestrian's residence time in the parking area and the density of the point cloud in the parking area are calculated to obtain the data information of the pedestrian's residence time in the parking area and the density of the point cloud in the parking area;

[0031] U24. Based on the pedestrian's parking time and the density of the point cloud in the parking area, establish an abnormality scoring function G for the parking area.

[0032] G=ω1H time +ω2H density +ω3H pose ,

[0033] Among them, ω1, ω2 and ω3 are weight coefficients, H time Score the residence time, H density is the density score, H pose Score the posture and perform abnormal parking state detection to determine whether there is any abnormal state.

[0034] Furthermore, the function in_region() for judging whether the pedestrian is in the parking area is:

[0035] ,

[0036] Among them, p i is the result of pedestrian target recognition and tracking, and area(W) is the point cloud in the parking area.

[0037] Furthermore, in step U21, the preprocessing function P of the data information of the original laser radar point cloud is object for,

[0038] P object =P total -{g i ||n i ×g|>cosθ ground},

[0039] Among them, p total is the data information of the original laser radar point cloud, g i is the i-th point of the original lidar point cloud, n i is the point cloud g i The normal vector, g is the direction of gravity, θ ground is the threshold;

[0040] The preprocessing function I of the data information of the image of the camera 预 for,

[0041] I 预 =CLAHE(GaussianBlur(I,λ)),

[0042] Where I is the data information of the camera image, CLAHE is the histogram equalization processing function, GaussianBlur is the Gaussian noise processing, and λ is the standard deviation of Gaussian blur.

[0043] Furthermore, in step U22, the use of the improved Hungarian algorithm to detect and continuously track pedestrian targets includes:

[0044] U221. Based on the pre-processed laser radar point cloud data information and the camera image data information, establish the point cloud and image association cost function c ij ,

[0045] c ij =η1×d geo (h i ,h j )+η2×d app (fi ,f j ),

[0046] Among them, d geo (h i ,h j ) is the point h in the point cloud cluster i With the image bounding box h j The geometric distance, d app (f i ,f j ) is the laser point cloud cluster feature f i With image feature f j The distance, η1 and η2 are penalty factors;

[0047] U222. Based on the association cost function c of the point cloud and the image ij , establish pedestrian target detection function C,

[0048] ,

[0049] Among them, x ij To associate the lidar point cloud cluster i with the pedestrian box j detected by the camera.

[0050] Furthermore, the abnormal state detection of the parking space and determining whether there is an abnormal state further includes:

[0051] U25. Based on the parking area anomaly scoring function G, an anomaly score for the parking area is obtained and a preset threshold is set. If the parking area anomaly score is less than the preset threshold, the parking space is considered normal, and the autonomous driving controller executes a stop according to the preset trajectory and strategy.

[0052] U26. If the parking area's anomaly score exceeds the preset threshold, the parking space status is abnormal. The autonomous driving controller immediately pulls over to the side of the road, and a human driver takes over. After the pedestrian leaves the parking space, the driver manually drives the vehicle into the station and parks.

[0053] Furthermore, in step U4, the hierarchical implementation of the early warning strategy by statistically analyzing the pedestrians' occupancy and residence time includes:

[0054] U41. Pedestrian parking time T stayq ≥T threshold1 , currently in a mild warning state of abnormal parking area status;

[0055] U42. Pedestrian occupancy time T stayq ≥T threshold2 The station-train collaborative control platform defines that the parking area is currently in a severe abnormal warning state.

[0056] In order to achieve the above-mentioned objectives and other related objectives, the present invention also provides an early warning system for abnormal conditions in bus stop parking areas, comprising a computer device programmed or configured to execute the steps of any one of the early warning methods for abnormal conditions in bus stop parking areas.

[0057] In order to achieve the above-mentioned and other related purposes, the present invention also provides a computer-readable storage medium, which stores a computer program programmed or configured to execute any one of the above-mentioned methods for warning of abnormal conditions in bus stop parking areas.

[0058] The present invention has the following positive effects:

[0059] 1. This invention integrates multi-source data, including roadside lidar and camera data, to capture bus stop parking area information from various angles and dimensions. In pedestrian occupancy detection, it calculates pedestrian dwell time and density, significantly reducing false positives and missed detections. Compared to traditional single-detection methods, the accuracy of anomaly detection is significantly improved.

[0060] 2. The system's timely and accurate anomaly detection and warning functions provide strong guarantees for the normal operation of public buses. When a pedestrian is detected occupying a parking space, the bus can adjust its driving strategy in advance based on the warning information to avoid parking delays caused by parking area anomalies. The bus operation management department can also quickly activate emergency plans based on the warning and reasonably dispatch vehicles, reducing the impact on the normal operation of bus lines, improving the punctuality and operating efficiency of buses, and thus enhancing the quality of bus services and enhancing public satisfaction and trust in public transportation. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 This is a schematic diagram of a parking area illegal parking scenario of the present invention;

[0062] Figure 2 Schematic diagram of the method flow of the present invention;

[0063] Figure 3 This is a flow chart of detecting abnormal parking conditions and determining whether there is an abnormal condition according to the present invention. DETAILED DESCRIPTION

[0064] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0065] Example 1: Figure 1 or Figure 2 As shown, a method for early warning of abnormal conditions in a bus stop parking area includes:

[0066] U1. The V2X OBU in the unmanned bus system is within the direct communication range of the PC5 interface and receives the longitude and latitude information of the center point of the station parking space transmitted by the V2X RSU. The V2X OBU uses its built-in GPS module to calculate and match the location of the unmanned bus with the center point of the station parking space in real time. If the distance difference is within 100 meters, it determines that the unmanned bus is about to enter the station.

[0067] The U2.V2X OBU transmits the upcoming arrival information to the station-train collaborative control platform, and the V2X RSU forwards it to the MEC. After receiving the upcoming arrival information, the MEC performs parking space abnormality detection based on the raw data of parking space obstacles received and stored in real time by the platform sensor system to determine whether there is any abnormality.

[0068] U3. When abnormal parking space status information is detected in step U2, the MEC directly transmits the information to the V2X OBU via the V2X RSU. The V2X OBU transmits the parking space status information to the autonomous driving controller.

[0069] U4. When step U3 obtains information about abnormal parking space status, the information is transmitted to the station-vehicle collaborative control platform and the station-vehicle collaborative management platform via 4G / 5G. By statistically analyzing the pedestrian parking time, a hierarchical early warning strategy is implemented.

[0070] In this embodiment, in step U1, determining that the unmanned public transport vehicle is about to enter the station includes:

[0071] U11. The longitude and latitude of the parking space center is expressed as P1=(γ1,θ1), then the parking space center and other points P x =(γ x ,θ x ) of the distance function d(P1,P x )for,

[0072] ,

[0073] Among them, γ1 is longitude, θ1 is latitude, γ x is the longitude of other points, θ x is the latitude of other points;

[0074] U12. Get the location B of the unmanned bus at different time points t t =(γ b (t),θ b (t)), the position of V2X RSU Q = (γ r,θ r ), the coverage of V2X RSU is D r =0.3-0.5km, the coverage range varies depending on the actual working conditions;

[0075] U13.V2X RSU communication range determination:

[0076] ,

[0077] Among them, when in_range Bt,Q =1, the V2X OBU determines the distance to the station based on the received coordinates of the center point of the platform parking space:

[0078] ,

[0079] Among them, D ɑ =0.1km, which is the threshold distance for entering the station. The value can be manually adjusted dynamically according to the actual working conditions. Bt,P1 =1 indicates that the unmanned bus is about to arrive at the station.

[0080] In this embodiment, if Figure 3 As shown, in step U2, the abnormal state detection of the parking space and the determination of whether there is an abnormal state include:

[0081] U21. Based on the real-time reception and storage of the original data of the parking space obstacle of the platform sensor system, the original laser radar point cloud data information and the camera image data information are obtained and preprocessed to obtain the preprocessed laser radar point cloud data information and the camera image data information;

[0082] U22. Based on the preprocessed lidar point cloud data and camera image data, an improved Hungarian algorithm is used to detect and continuously track pedestrian targets to obtain pedestrian target recognition and tracking results;

[0083] U23. Based on the results of pedestrian target recognition and tracking, construct a pedestrian stay time function T in the parking area stay (i, t) and the density function of the point cloud in the parking area ρ(W, t),

[0084] ,

[0085] ,

[0086] Among them, S is the indicator function, which uses 0-1 to quantify the state, i is a positive integer, t is a different time point, T is the time window, in_region() is the judgment function of whether the pedestrian is in the parking area, p iis the result of pedestrian target recognition and tracking, n is the sample size, W is the point cloud, area(W) is the point cloud in the parking area, and the pedestrian's residence time in the parking area and the density of the point cloud in the parking area are calculated to obtain the data information of the pedestrian's residence time in the parking area and the density of the point cloud in the parking area;

[0087] U24. Based on the pedestrian's parking time and the density of the point cloud in the parking area, establish an abnormality scoring function G for the parking area.

[0088] G=ω1H time +ω2H density +ω3H pose ,

[0089] Among them, ω1, ω2 and ω3 are weight coefficients, H time Score the residence time, H density is the density score, H pose Score the posture and perform abnormal parking state detection to determine whether there is any abnormal state.

[0090] Table 1 Meaning and value selection rules of scoring indicators

[0091]

[0092] In this embodiment, the function in_region() for determining whether the pedestrian is in the parking area is:

[0093] ,

[0094] Among them, p i is the result of pedestrian target recognition and tracking, and area(W) is the point cloud in the parking area.

[0095] In this embodiment, in step U21, the preprocessing function P of the data information of the original laser radar point cloud is object for,

[0096] P object =P total -{g i ||n i ×g|>cosθ ground},

[0097] Among them, p total is the data information of the original laser radar point cloud, g i is the i-th point of the original lidar point cloud, n i is the point cloud g i The normal vector, g is the direction of gravity, θ ground is the threshold;

[0098] The preprocessing function I of the data information of the image of the camera 预 for,

[0099] I 预 =CLAHE(GaussianBlur(I,λ)),

[0100] Where I is the data information of the camera image, CLAHE is the histogram equalization processing function, GaussianBlur is the Gaussian noise processing, and λ is the standard deviation of Gaussian blur.

[0101] In this embodiment, in step U22, detecting and continuously tracking pedestrian targets using the improved Hungarian algorithm includes:

[0102] U221. Based on the pre-processed laser radar point cloud data information and the camera image data information, establish the point cloud and image association cost function c ij ,

[0103] c ij =η1×d geo (h i ,h j )+η2×d app (f i ,f j ),

[0104] Among them, d geo (h i ,h j ) is the point h in the point cloud cluster i With the image bounding box h j The geometric distance, d app (f i ,f j ) is the laser point cloud cluster feature f i With image feature f j The distance, η1 and η2 are penalty factors;

[0105] U222. Based on the association cost function c of the point cloud and the image ij , establish pedestrian target detection function C,

[0106] ,

[0107] Among them, x ij To associate the lidar point cloud cluster i with the pedestrian box j detected by the camera.

[0108] In this embodiment, the detection of abnormal parking state and determination of whether there is an abnormal state further includes:

[0109] U25. Based on the parking area anomaly scoring function G, an anomaly score for the parking area is obtained and a preset threshold is set. If the parking area anomaly score is less than the preset threshold, the parking space is considered normal, and the autonomous driving controller executes a stop according to the preset trajectory and strategy.

[0110] U26. If the parking area's anomaly score exceeds the preset threshold, the parking space status is abnormal. The autonomous driving controller immediately pulls over to the side of the road, and a human driver takes over. After the pedestrian leaves the parking space, the driver manually drives the vehicle into the station and parks.

[0111] Taking a city's smart bus demonstration line as an example to verify the practical application effect of the present invention, 10 typical bus stops were selected for a one-month test, and relevant data were collected and analyzed.

[0112] During the implementation process, V2X RSUs were first deployed at each test station, followed by GPS-enabled V2X OBUs installed on unmanned buses. The stations were equipped with a variety of sensors, including millimeter-wave radars, ultrasonic sensors, and high-definition cameras, to collect environmental data from parking areas. MEC nodes were equipped with NVIDIA Jetson Xavier NX edge computing devices, running an obstacle detection algorithm based on YOLOv5.

[0113] Test results show that during peak hours in the morning and evening, the system detected an average of 15 parking area anomalies per day, mainly including:

[0114] Pedestrians illegally occupying parking spaces: 60%

[0115] Illegally parked vehicles: 25%

[0116] Other obstacles: 15%

[0117] Comparative experiments have shown that the method of the present invention achieves an abnormal state detection accuracy rate of over 95%, and an early warning response time of less than 300ms. The system can maintain high detection stability, especially under complex weather conditions.

[0118] Example 2: Based on the early warning method for abnormal status of a bus stop parking area in Example 1, the present invention is further illustrated and described below.

[0119] like Figure 1 or Figure 2 As shown, a method for early warning of abnormal conditions in a bus stop parking area includes:

[0120] U1. The V2X OBU in the unmanned bus system is within the direct communication range of the PC5 interface and receives the longitude and latitude information of the center point of the station parking space transmitted by the V2X RSU. The V2X OBU uses its built-in GPS module to calculate and match the location of the unmanned bus with the center point of the station parking space in real time. If the distance difference is within 100 meters, it determines that the unmanned bus is about to enter the station.

[0121] The U2.V2X OBU transmits the upcoming arrival information to the station-train collaborative control platform, and the V2X RSU forwards it to the MEC. After receiving the upcoming arrival information, the MEC performs parking space abnormality detection based on the raw data of parking space obstacles received and stored in real time by the platform sensor system to determine whether there is any abnormality.

[0122] U3. When abnormal parking space status information is detected in step U2, the MEC directly transmits the information to the V2X OBU via the V2X RSU. The V2X OBU transmits the parking space status information to the autonomous driving controller.

[0123] U4. When step U3 obtains information about abnormal parking space status, the information is transmitted to the station-vehicle collaborative control platform and the station-vehicle collaborative management platform via 4G / 5G. By statistically analyzing the pedestrian parking time, a hierarchical early warning strategy is implemented.

[0124] In this embodiment, in step U4, the hierarchical implementation of the early warning strategy by statistically analyzing the pedestrian's occupancy time includes:

[0125] U41. Pedestrian parking time T stayq ≥T threshold1 , currently in a mild warning state of abnormal parking area status;

[0126] U42. Pedestrian occupancy time T stayq ≥T threshold2 The station-train collaborative control platform defines that the parking area is currently in a severe abnormal warning state.

[0127] First, the V2X OBU receives the longitude and latitude of the center point of the station parking space from the V2X RSU within the direct communication range of the PC5 interface. The built-in GPS module calculates the relative position of the bus and the platform parking space in real time. When the distance is less than 100 meters, the bus is considered to be approaching the station.

[0128] Secondly, a station-vehicle collaborative control platform is established as the core processing unit. This platform forwards upcoming arrival information to the MEC edge computing node via the V2X RSU. Based on real-time data received from the platform sensor system, the MEC uses a deep learning algorithm to detect abnormal parking space obstacles.

[0129] Finally, a tiered warning mechanism was designed. When an abnormal condition is detected, it is directly transmitted to the V2X OBU via the V2X RSU and simultaneously sent to the autonomous driving controller. Simultaneously, this abnormal information is uploaded to the station-vehicle collaborative control platform via the 4G / 5G network. Combined with statistical analysis of pedestrian occupancy time, different levels of warning strategies are implemented.

[0130] In order to more intuitively demonstrate the technical process of the present invention, the following data table can be constructed to illustrate the key parameters of each link:

[0131]

[0132] Compared with traditional manual monitoring methods, the present invention has significant advantages in detection accuracy, response speed, and degree of automation. The stability and reliability of the system have been fully verified, especially in scenarios with large passenger flows during peak hours.

[0133] In this embodiment, the present invention provides an early warning system for abnormal conditions in bus stop parking areas, including a computer device programmed or configured to execute the steps of any one of the early warning methods for abnormal conditions in bus stop parking areas.

[0134] In this embodiment, the present invention provides a computer-readable storage medium storing a computer program programmed or configured to execute any one of the above-described methods for warning of abnormal conditions in a bus stop parking area.

[0135] Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM).

[0136] In summary, the present invention not only integrates multi-source data such as roadside lidar and cameras to obtain bus stop parking area information from different angles and dimensions, but also calculates pedestrian dwell time and density during pedestrian occupancy detection, greatly reducing misjudgments and missed detections. Compared with traditional single detection methods, the accuracy of anomaly detection is greatly improved.

[0137] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A method for early warning of abnormal conditions in bus stop parking areas, characterized in that: The method comprises: U1. The V2X OBU in the unmanned bus system is within the direct communication range of the PC5 interface and receives the longitude and latitude information of the center point of the station parking space transmitted by the V2X RSU. The V2X OBU uses its built-in GPS module to calculate and match the location of the unmanned bus with the center point of the station parking space in real time. If the distance difference is within 100 meters, it determines that the unmanned bus is about to enter the station. The U2.V2X OBU transmits the upcoming arrival information to the station-train collaborative control platform, and the V2X RSU forwards it to the MEC. After receiving the upcoming arrival information, the MEC performs parking space abnormality detection based on the raw data of parking space obstacles received and stored in real time by the platform sensor system to determine whether there is any abnormality. U3. When abnormal parking space status information is detected in step U2, the MEC transmits the information directly to the V2X OBU via the V2X RSU. The V2X OBU transmits the parking space status information to the autonomous driving controller. U4. When abnormal parking space status information is obtained in step U3, the information is transmitted via 4G / 5G to the station-train collaborative control platform and the station-train collaborative management platform. A hierarchical early warning strategy is implemented based on statistical analysis of pedestrian parking time. In step U2, the abnormal state detection of the parking space and determining whether there is an abnormal state include: U21. Based on the real-time reception and storage of the original data of the parking space obstacle of the platform sensor system, the original laser radar point cloud data information and the camera image data information are obtained and preprocessed to obtain the preprocessed laser radar point cloud data information and the camera image data information; U22. Based on the preprocessed lidar point cloud data and camera image data, an improved Hungarian algorithm is used to detect and continuously track pedestrian targets to obtain pedestrian target recognition and tracking results; U23. Based on the results of pedestrian target recognition and tracking, construct a pedestrian stay time function T in the parking area stay (i, t) and the density function of the point cloud in the parking area ρ(W, t), , ， Among them, S is the indicator function, which uses 0-1 to quantify the state, i is a positive integer, t is a different time point, T is the time window, in_region() is the judgment function of whether the pedestrian is in the parking area, p i is the result of pedestrian target recognition and tracking, n is the sample size, W is the point cloud, area(W) is the point cloud in the parking area, and the pedestrian's residence time in the parking area and the density of the point cloud in the parking area are calculated to obtain the data information of the pedestrian's residence time in the parking area and the density of the point cloud in the parking area; U24. Based on the pedestrian's parking time and the density of the point cloud in the parking area, establish an abnormality scoring function G for the parking area. G=ω1H time +ω2H density +ω3H pose , Among them, ω1, ω2 and ω3 are weight coefficients, H time Score the residence time, H density is the density score, H pose Score the posture and perform abnormal parking state detection to determine whether there is any abnormal state.

2. The method for early warning of abnormal conditions in bus stop parking areas according to claim 1, characterized in that: In step U1, determining that an unmanned public transport vehicle is about to enter the station includes: U11. The longitude and latitude of the parking space center is expressed as P1=(γ1,θ1), then the parking space center and other points P x =(γ x ,θ x ) of the distance function d(P1,P x )for, , Among them, γ1 is longitude, θ1 is latitude, γ x is the longitude of other points, θ x is the latitude of other points; U12. Get the location B of the unmanned bus at different time points t t =(γ b (t),θ b (t)), the position of V2X RSU Q = (γ r ,θ r ), the coverage of V2X RSU is D r =0.3-0.5km, the coverage range varies depending on the actual working conditions; U13.V2X RSU communication range determination: ， Among them, when in_range Bt,Q =1, the V2X OBU determines the distance to the station based on the received coordinates of the center point of the platform parking space: ， Among them, D ɑ =0.1km, which is the threshold distance for entering the station. The value can be manually adjusted dynamically according to the actual working conditions. Bt,P1 =1 indicates that the unmanned bus is about to arrive at the station.

3. The method for early warning of abnormal conditions in bus stop parking areas according to claim 1, characterized in that: The function in_region() for judging whether the pedestrian is in the parking area is: ， Among them, p i is the result of pedestrian target recognition and tracking, and area(W) is the point cloud in the parking area.

4. The method for early warning of abnormal conditions in bus stop parking areas according to claim 1, characterized in that: In step U21, the preprocessing function P of the data information of the original laser radar point cloud is object for, P object =P total -{g i ||n i ×g|>cosθ ground }, Among them, p total is the data information of the original laser radar point cloud, g i is the i-th point of the original lidar point cloud, n i is the point cloud g i The normal vector, g is the direction of gravity, θ ground is the threshold; The preprocessing function I of the data information of the image of the camera 预 for, I 预 =CLAHE(GaussianBlur(I,λ)), Where I is the data information of the camera image, CLAHE is the histogram equalization processing function, GaussianBlur is the Gaussian noise processing, and λ is the standard deviation of Gaussian blur.

5. The method for early warning of abnormal conditions in bus stop parking areas according to claim 1, characterized in that: In step U22, the use of the improved Hungarian algorithm to detect and continuously track pedestrian targets includes: U221. Based on the pre-processed laser radar point cloud data information and the camera image data information, establish the point cloud and image association cost function c ij , c ij =η1×d geo (h i ,h j )+η2×d app (f i ,f j ), Among them, d geo (h i ,h j ) is the point h in the point cloud cluster i With the image bounding box h j The geometric distance, d app (f i ,f j ) is the laser point cloud cluster feature f i With image feature f j The distance, η1 and η2 are penalty factors; U222. Based on the association cost function c of the point cloud and the image ij , establish pedestrian target detection function C, ， Among them, x ij To associate the lidar point cloud cluster i with the pedestrian box j detected by the camera.

6. The method for early warning of abnormal conditions in bus stop parking areas according to claim 1, characterized in that: The abnormal state detection of the parking space and determining whether there is an abnormal state further include: U25. Based on the parking area anomaly scoring function G, an anomaly score for the parking area is obtained and a preset threshold is set. If the parking area anomaly score is less than the preset threshold, the parking space is considered normal, and the autonomous driving controller executes a stop according to the preset trajectory and strategy. U26. If the parking area's anomaly score exceeds the preset threshold, the parking space status is abnormal. The autonomous driving controller immediately pulls over to the side of the road, and a human driver takes over. After the pedestrian leaves the parking space, the driver manually drives the vehicle into the station and parks.

7. An early warning system for abnormal conditions in bus stop parking areas, comprising computer equipment, characterized in that: The computer device is programmed or configured to execute the steps of the early warning method for abnormal status of a bus stop parking area as claimed in any one of claims 1 to 6.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program programmed or configured to execute the method for early warning of abnormal conditions in a bus stop parking area according to any one of claims 1 to 6.

Images