A high-precision method for locating road traffic safety hazards

Abstract

This invention provides a high-precision method for locating road traffic safety hazards, comprising: constructing a dynamic coordinate system based on the sensing range of a lidar sensor; locating hazards by using traffic image data acquired by a camera device and applying the traffic image data to a traffic hazard identification model; determining real-time coordinate data of the traffic hazard based on the hazard location and the dynamic coordinate system; and determining the relative position data and safe distance data of the traffic hazard from the vehicle based on the real-time coordinate data. This invention can accurately locate the area of ​​a safety hazard while the vehicle is in motion, based on real-time detection and judgment of safety hazards. By locating the safety hazard, it can determine a relatively safe position and a distance-based safe position, thereby providing precise alerts to the user.

Description

Technical Field

[0001] This invention relates to the field of traffic safety assessment technology, and in particular to a method and system for evaluating the improvement of road traffic safety hazards. Background Technology

[0002] Currently, with the rapid development of my country's economy, the construction level of my country's highway transportation infrastructure and the scale of highway transportation network have been greatly improved. In addition, with the increasing number of cars, the accompanying traffic safety issues have attracted widespread attention.

[0003] Currently, most assessments of potential traffic safety hazards on roads are based on relevant regulations, driving comfort, and the highway operating environment, often involving simple warning signs at accident sites. If drivers fail to notice these signs, serious accidents can still occur. At present, there is no real-time and effective road traffic safety hazard prediction system, making it difficult to promptly detect potential road safety hazards. This hinders the application of advanced highway safety assessment technologies in engineering practice, resulting in poor traffic safety warnings for different vehicles on the road. Therefore, this invention provides a method and system for predicting road traffic safety hazards. By constructing a road traffic safety hazard prediction model, it analyzes the driving characteristic parameters of passing vehicles to accurately predict the probability of traffic safety hazards, thereby facilitating early warnings to drivers, improving the reliability and accuracy of eliminating traffic safety hazards, and ensuring driver safety. Summary of the Invention

[0004] This invention provides a high-precision positioning method for road traffic safety hazards, which can be used to solve the problem of drivers encountering safety hazards.

[0005] This invention proposes a high-precision method for locating road traffic safety hazards, comprising:

[0006] Based on the sensing range of the lidar, a dynamic coordinate system based on the lidar is constructed;

[0007] Traffic image data is acquired by camera devices, and the traffic image data is used to locate potential hazards through a traffic hazard identification model;

[0008] Based on the location of the hazard and the dynamic coordinate system, determine the real-time coordinate data of the traffic hazard;

[0009] Based on real-time coordinate data, the relative position data and safe distance data of vehicles from traffic hazards are determined.

[0010] Preferably, the dynamic coordinate system includes:

[0011] Pre-set the vehicle's geometric parameters and determine the vehicle's geometric endpoints;

[0012] Based on the geometric endpoints, construct the vehicle's geometric coordinate system; where,

[0013] The geometric coordinate system includes: a first coordinate system set for the left endpoint of the vehicle, a second coordinate system set for the right endpoint of the vehicle, a third coordinate system set for the front endpoint of the vehicle, and a fourth coordinate system set for the rear endpoint of the vehicle;

[0014] Obtain the vehicle's satellite navigation data to determine the vehicle's traffic route;

[0015] Integrate the geometric coordinate system into the traffic route to generate a dynamic geometric coordinate system based on traffic trajectory;

[0016] A set of dynamic coordinate systems based on geometric endpoints is generated using dynamic geometric coordinate systems.

[0017] The user and dynamic coordinate system set are used to update the LiDAR data in real time.

[0018] Preferably, the method further includes:

[0019] Based on the sensing range of the lidar, the lidar's sensing accuracy is divided into rings, generating multiple ring-shaped sensing areas; among them,

[0020] Each ring-shaped sensing area corresponds to a sensing accuracy of the lidar.

[0021] Based on the sensing accuracy, determine the deviation ring sensing area that is lower than the preset sensing accuracy;

[0022] An ultrasonic-assisted measurement sensing device is set up according to the deviation ring sensing area;

[0023] Based on the ultrasonic-assisted measurement and sensing device, obstacle data in the deviation ring sensing area are synchronously sensed;

[0024] Based on synchronous sensing, induction compensation is performed.

[0025] Preferably, the traffic image data includes:

[0026] The system acquires multiple traffic images captured by a camera device, inertial navigation data measured by an inertial measurement unit, and positioning data obtained by a positioning system.

[0027] The camera, inertial measurement unit, and positioning system are mounted on the same traffic image acquisition vehicle;

[0028] Based on traffic images, inertial navigation data, and positioning data, determine whether the speed of the vehicle collecting the traffic images is lower than a preset speed threshold.

[0029] If it is determined that the speed of the vehicle collecting the traffic image is lower than the preset speed threshold, the traffic image will be considered invalid data.

[0030] If it is determined that the speed of the vehicle collecting the traffic image is higher than a preset speed threshold, then the traffic image is considered valid data.

[0031] Preferably, the traffic hazard identification model includes a weather-based hazard model, a traffic flow-based hazard model, a geographical hazard model, an obstacle-based collision hazard model, a road-based road hazard model, and a navigation signal-based communication hazard model; wherein,

[0032] The meteorological hazard model is used to identify potential driving hazards for vehicles under different weather conditions;

[0033] Traffic flow hazard model is used to identify potential vehicle accident hazards under different traffic flow density scenarios;

[0034] Geographic hazard models are used to identify potential driving hazards for vehicles in different geographical environments.

[0035] Collision hazard models are used to identify vehicle collision hazards under different obstacle conditions;

[0036] The road hazard model is used to determine the speed hazards of vehicles under different road conditions;

[0037] The communication vulnerability model is used to identify communication vulnerabilities in vehicles under different communication conditions and intensities.

[0038] Preferably, the method further includes:

[0039] Receive traffic image data from the camera device and input them into the corresponding traffic hazard identification model;

[0040] Target hazard factors corresponding to the target traffic hazard identification model are determined from multiple traffic images; among them...

[0041] Each traffic image stores a baseline feature for comparing potential traffic hazards.

[0042] By comparing traffic images with at least one hazard comparison benchmark feature in a traffic hazard model, at least one target comparison result is obtained.

[0043] The target identification result is determined based on the comparison result of at least one target, and the hazard location data of the traffic hazard is determined.

[0044] Preferably, determining the real-time coordinate data of the traffic hazard based on the hazard location and dynamic coordinate system includes:

[0045] Based on the hazard location data, the coordinates of the corresponding location on the dynamic coordinate system are determined using lidar.

[0046] Calculate the normal vector of each coordinate point in the coordinate data, and set the coordinate measurement grid for the location of the potential hazard;

[0047] By using a coordinate measurement grid, the coordinate points and normal vectors of each side of the vehicle are clustered to generate clusters;

[0048] Calculate the consistency of clusters based on the clusters;

[0049] Based on the consistency of the clusters, a 3D outline of the vehicle hazard location is generated;

[0050] Based on the 3D contour, determine the real-time coordinate data of traffic hazards.

[0051] Preferably, the relative position data includes:

[0052] Based on real-time coordinate data of traffic hazards, determine the relative angle information of the location of the traffic hazard with respect to different endpoints of the vehicle;

[0053] Based on the relative angle information, determine the direction angle information of the traffic hazard relative to the vehicle;

[0054] Based on the azimuth information, determine the distribution data of vehicles corresponding to traffic hazards in the traffic image;

[0055] Based on the distribution data, a three-dimensional scene coordinate system is constructed with the vehicle as the origin.

[0056] Based on the three-dimensional scene coordinate system, the relative position information of different traffic hazards relative to vehicles is marked, and relative position modeling is performed.

[0057] Preferably, the safe distance data includes:

[0058] Acquire target ranging images of different endpoints of traffic hazards relative to vehicles;

[0059] First and second measurement points are set at the locations of safety hazards in the target ranging image, and reference lines are set in the target ranging image;

[0060] Based on the actual shape and size of traffic hazards and their corresponding pixel distances, calibration parameters are established for the actual distances in the image and their corresponding pixel distances.

[0061] The angle between the distance measuring line formed by the line connecting the first and second measuring points and the reference line is used as the basis for selecting calibration parameters. At this time, the product of the pixel distance of the distance measuring line and the selected calibration parameters is the measurement distance of the traffic hazard.

[0062] Preferably, the method further includes:

[0063] Acquire radar sensing data within the target sensing area of ​​the lidar, where,

[0064] Each radar sensing data corresponds to the image captured by the camera device at the corresponding viewpoint, and the set of video images includes images of the safety hazard from various viewpoints;

[0065] When a safety hazard appears in multiple video frames within a set of video footage, the ranging data of the safety hazard in the multiple video frames is accumulated;

[0066] An alarm will be triggered based on the ranging data.

[0067] The beneficial effects of this invention are as follows:

[0068] This invention can detect and judge safety hazards in real time while the vehicle is in motion. By judging the safety hazards, it can accurately locate the area of ​​the safety hazard, determine the relatively safe position and the distance safe position, and then give the user an accurate reminder.

[0069] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0070] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0071] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0072] Figure 1 This is a flowchart illustrating a high-precision positioning method for road traffic safety hazards according to an embodiment of the present invention.

[0073] Figure 2 This is a classification diagram of the traffic hazard identification model in an embodiment of the present invention;

[0074] Figure 3 This is a diagram showing the radar sensing loop in an embodiment of the present invention. Detailed Implementation

[0075] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0076] This invention proposes a high-precision method for locating road traffic safety hazards, comprising:

[0077] Based on the sensing range of the lidar, a dynamic coordinate system based on the lidar is constructed;

[0078] Traffic image data is acquired by camera devices, and the traffic image data is used to locate potential hazards through a traffic hazard identification model;

[0079] Based on the location of the hazard and the dynamic coordinate system, determine the real-time coordinate data of the traffic hazard;

[0080] Based on real-time coordinate data, the relative position data and safe distance data of vehicles from traffic hazards are determined.

[0081] The principle behind the above technical solution is as follows:

[0082] As attached Figure 1 As shown, this invention proposes a method for locating safety hazards, which can achieve high-precision positioning compared to existing technologies.

[0083] In this process, firstly, based on the sensing range of the lidar, a dynamic coordinate system of the lidar is constructed. This is equivalent to being able to perform depth sensing of the surroundings while the vehicle is moving dynamically. Based on the dynamic coordinate system, a model is built to achieve measurement of the surroundings.

[0084] For traffic hazard identification, this invention is based on image recognition. It identifies traffic hazard data on traffic images to determine road hazards. However, this invention is not limited to road hazards. Furthermore, it locates the traffic hazard using image location and a dynamic coordinate system. Existing technologies only locate distances, but this invention is based on endpoint positioning. This approach better prevents collisions and other issues, making it easier for drivers to drive precisely. Finally, based on relative position data and safe distance data, it ensures vehicle driving safety.

[0085] The beneficial effects of the above technical solution are as follows:

[0086] This invention can detect and judge safety hazards in real time while the vehicle is in motion. By judging the safety hazards, it can accurately locate the area of ​​the safety hazard, determine the relatively safe position and the safe distance position, and then give the user an accurate reminder.

[0087] Preferably, the dynamic coordinate system includes:

[0088] Pre-set the vehicle's geometric parameters and determine the vehicle's geometric endpoints;

[0089] Based on the geometric endpoints, construct the vehicle's geometric coordinate system; where,

[0090] The geometric coordinate system includes: a first coordinate system set for the left endpoint of the vehicle, a second coordinate system set for the right endpoint of the vehicle, a third coordinate system set for the front endpoint of the vehicle, and a fourth coordinate system set for the rear endpoint of the vehicle;

[0091] Obtain the vehicle's satellite navigation data to determine the vehicle's traffic route;

[0092] Integrate the geometric coordinate system into the traffic route to generate a dynamic geometric coordinate system based on traffic trajectory;

[0093] A set of dynamic coordinate systems based on geometric endpoints is generated using dynamic geometric coordinate systems.

[0094] The user and dynamic coordinate system set are used to update the LiDAR data in real time.

[0095] The principle behind the above technical solution is as follows:

[0096] In constructing a dynamic coordinate system, the present invention addresses the issue that existing technologies only set up a dynamic coordinate system based on the vehicle. However, a dynamic coordinate system cannot measure distances accurately because the vehicle's own geometric parameters must be input in advance.

[0097] However, vehicle parameters may change. Therefore, when performing geometric coordinate processing, this invention constructs a large number of geometric coordinate systems, performs geometric measurements on the overall sides of the vehicle, and determines the distance between the endpoints of the vehicle's sides and the location of the traffic hazard.

[0098] In this process, the present invention also achieves traffic trajectory fusion based on satellite navigation data. Through traffic trajectory fusion, the ranging data is updated in real time by lidar to determine high-precision ranging data.

[0099] The beneficial effects of this invention are as follows:

[0100] First, the present invention is based on geometric endpoints, which can ensure that collisions will not occur to the greatest extent. In the prior art, a safe distance must be set, and safety cannot be guaranteed in the presence of visual or sensory blind spots.

[0101] Secondly, the present invention fuses the endpoints of multiple coordinate systems, and then uses satellite navigation data to achieve rapid, short-range, high-speed positioning of road traffic routes.

[0102] In addition, the present invention uses a dynamic coordinate system. The problem with dynamic coordinate systems is that the measurement is inaccurate. Although the present invention uses a dynamic coordinate system, it can ensure the accuracy of measurement and the rapid updating of measurement data because it combines endpoint detection with endpoint detection of multiple coordinate systems.

[0103] In determining the geometric endpoints of a vehicle, existing technologies, to ensure the accuracy of the endpoints, may determine the geometric endpoints based on the vehicle's own manufacturing parameters. This is highly accurate for new cars, but insufficient for older cars. Furthermore, the geometric parameters of a vehicle cannot be guaranteed to meet user requirements after leaving the factory, as vehicles experience wear and tear, collisions, or even traffic accidents. Therefore, this invention employs the following method to optimize accuracy detection:

[0104] Step 1: This invention first segments the dynamic geometric data acquired from the vehicle's dynamic geometric coordinate system into frames and performs windowing processing to generate a data frame set:

[0105] F(x) = f1, f2, f3…f x

[0106] Among them, f x This represents the dynamic geometric data of the x-th frame; x is a positive integer, x∈n;

[0107] This invention divides the geometric signal into frames, and after framing, it can perform calculations on each detail of the geometric signal after framing.

[0108] Step 2: Determine the phase value of the dynamic geometric data for each frame by performing frame transformation on the frame-segmented geometric set.

[0109]

[0110] Where j represents the imaginary number; k represents the component; and n represents the total number of frames;

[0111] The phase value determines the stability of the real-time detected data in the dynamic geometric coordinate system, and can also represent the accuracy of the acquired geometric data.

[0112] Step 3: Calculate the entropy value of the dynamic geometric data based on the phase value;

[0113]

[0114] Where S represents the entropy value of the geometric signal;

[0115] By assessing the accuracy of geometric data, we can determine the entropy value of dynamic geometric data. The entropy value can determine the importance of changing geometric data, which corresponds to the accuracy of vehicle endpoint data. Under normal circumstances, the vehicle's power failure will not cause any changes.

[0116] Step 4: Calculate the total distance confidence of the geometric signal based on the entropy value:

[0117]

[0118] Where D represents the total distance confidence.

[0119] Distance reliability can be used to determine the reliability of endpoint distances when performing geometric endpoint detection, based on the total distance reliability, to ensure that the vehicle's dynamic detection coordinate system is not erroneous.

[0120] Step 5: Calculate the expected loss of the geometric signal based on the distance confidence level:

[0121]

[0122] Where Q represents the expected loss value.

[0123] Step 6: Based on the expected loss, optimize each frame of the geometric signal separately, and determine the judgment value of the optimized target geometric signal:

[0124] L = [|f x *r(a x )| Q ]-β|p(a x )| Q

[0125] Where L represents the determination value of the target geometric signal; r(a x p(a) represents the weighted expectation function of the geometric signal; x ) represents the weighting function of the geometric signal; β represents the weighting loss coefficient;

[0126] In steps 5 and 6, step 5 determines the expected loss, which is the potential loss in detection accuracy. This loss in detection accuracy is calculated by subtracting the actual value from the expected value in step 6 to determine the optimized geometric signal. However, the judgment value of this target geometric signal is used to judge the correctness of the expected loss. If L is greater than 1, it means that the actual loss does not exceed the expected loss, and the dynamic geometric data is reliable and can be used as the target dynamic geometric data. If the judgment value is less than 1, it means that the expected loss is exceeded and the data detected by the dynamic coordinate system is incorrect.

[0127] Preferably, the method further includes:

[0128] Based on the sensing range of the lidar, the lidar's sensing accuracy is divided into rings, generating multiple ring-shaped sensing areas; among them,

[0129] Each ring-shaped sensing area corresponds to a sensing accuracy of the lidar.

[0130] Based on the sensing accuracy, determine the deviation ring sensing area that is lower than the preset sensing accuracy;

[0131] An ultrasonic-assisted measurement sensing device is set up according to the deviation ring sensing area;

[0132] Based on the ultrasonic-assisted measurement and sensing device, obstacle data in the deviation ring sensing area are synchronously sensed;

[0133] Based on synchronous sensing, induction compensation is performed.

[0134] The principle behind the above technical solution is as follows:

[0135] When using lidar for sensing, this invention addresses the issue that lidar sensing covers a circular area, and that sensing accuracy decreases with distance. Therefore, this invention divides the sensing area into rings based on sensing accuracy. During the ring-division process, each ring sensing area is synchronously sensed using an ultrasonic-assisted measuring device. This synchronous sensing improves the lidar's sensing accuracy and achieves accuracy compensation.

[0136] The beneficial effects of the above technical solution are as follows:

[0137] The ring diagram is attached. Figure 3 As shown, the sensing accuracy within each ring has different sensing standards. By dividing the data into different sensing accuracies and comparing the initial accuracy deviation value, the areas where the lidar sensing is inaccurate are identified. Ultrasonic sensing devices are then used for sensing, and the two assist and synchronize with each other, resulting in more accurate sensing.

[0138] In terms of sensing accuracy, ultrasonic detection is suitable for parking lot environments. However, existing technologies do not integrate sensing data from lidar and ultrasonic sensors. This invention integrates sensing data from these two different sources to achieve more efficient accuracy compensation.

[0139] Preferably, the traffic image data includes:

[0140] The system acquires multiple traffic images captured by a camera device, inertial navigation data measured by an inertial measurement unit, and positioning data obtained by a positioning system.

[0141] The camera, inertial measurement unit, and positioning system are mounted on the same traffic image acquisition vehicle;

[0142] Based on traffic images, inertial navigation data, and positioning data, determine whether the speed of the vehicle collecting the traffic images is lower than a preset speed threshold.

[0143] If it is determined that the speed of the vehicle collecting the traffic image is lower than the preset speed threshold, the traffic image will be considered invalid data.

[0144] If it is determined that the speed of the vehicle collecting the traffic image is higher than a preset speed threshold, then the traffic image is considered valid data.

[0145] The principle behind the above technical solution is as follows:

[0146] In the process of locating traffic hazards using traffic image data, this invention requires constant identification of valid and invalid data because the ranging data is updated in real time. During the valid and invalid data detection process, traffic hazard data is only collected and identified after the vehicle exceeds a certain speed due to its own inertia. Generally, traffic hazard data is only collected when the vehicle speed is 0.

[0147] The beneficial effects of the above technical solution are as follows:

[0148] This invention restricts the camera device, inertial measurement unit, and positioning system to be mounted on the same traffic image acquisition vehicle; this is to ensure the consistency of the data from the three components, allowing for synchronous measurement and mutual compensation. The preset speed threshold is designed to adapt the vehicle's sensing accuracy and data update speed to the highest speeds in existing technologies, preventing excessively high data rates that could lead to unclear data acquisition.

[0149] Preferably, the traffic hazard identification model includes a weather-based hazard model, a traffic flow-based hazard model, a geographical hazard model, an obstacle-based collision hazard model, a road-based road hazard model, and a navigation signal-based communication hazard model; wherein,

[0150] The meteorological hazard model is used to identify potential driving hazards for vehicles under different weather conditions;

[0151] Traffic flow hazard model is used to identify potential vehicle accident hazards under different traffic flow density scenarios;

[0152] Geographic hazard models are used to identify potential driving hazards for vehicles in different geographical environments.

[0153] Collision hazard models are used to identify vehicle collision hazards under different obstacle conditions;

[0154] The road hazard model is used to determine the speed hazards of vehicles under different road conditions;

[0155] The communication vulnerability model is used to identify communication vulnerabilities in vehicles under different communication conditions and intensities.

[0156] The principle behind the above technical solution is as follows:

[0157] As attached Figure 2As shown, when identifying traffic hazards, existing technologies only rely on a single identification model trained with big data to determine the corresponding traffic hazards. However, they cannot identify traffic hazards affected by weather conditions, geographical factors, vehicle flow density, collision environments, road-related hazards, or communication-related hazards. In contrast, this invention, in addition to identifying physical traffic hazards, also sets up corresponding traffic hazard models based on specific scenarios. Through these models, corresponding traffic hazard data is determined, and traffic hazards are identified.

[0158] The beneficial effects of this invention are as follows:

[0159] This invention identifies and detects different types of traffic hazards by setting up different models. Because different traffic models are used for different traffic hazards, the accuracy is higher.

[0160] Preferably, the method further includes:

[0161] Receive traffic image data from the camera device and input them into the corresponding traffic hazard identification model;

[0162] Target hazard factors corresponding to the target traffic hazard identification model are determined from multiple traffic images; among them...

[0163] Each traffic image stores a baseline feature for comparing potential traffic hazards.

[0164] By comparing traffic images with at least one hazard comparison benchmark feature in a traffic hazard model, at least one target comparison result is obtained.

[0165] The target identification result is determined based on the comparison result of at least one target, and the hazard location data of the traffic hazard is determined.

[0166] The principle behind the above technical solution is as follows:

[0167] When identifying traffic hazards using image data, this invention inputs the traffic image data into a preset traffic hazard identification model. It then compares the model with the baseline features of traffic hazard identification models in various countries using the simplest comparison method to identify whether a traffic hazard exists, locates the traffic hazard, and determines the corresponding hazard data.

[0168] The beneficial effects of the above technical solution are as follows:

[0169] Target hazard factors are pre-defined descriptive features for hazard identification, such as the characteristics of injuries sustained in a traffic accident. Then, through feature comparison, the identification result of a specific traffic hazard is determined, and the hazard location data causing the traffic hazard is identified.

[0170] Preferably, determining the real-time coordinate data of the traffic hazard based on the hazard location and dynamic coordinate system includes:

[0171] Based on the hazard location data, the coordinates of the corresponding location on the dynamic coordinate system are determined using lidar.

[0172] Calculate the normal vector of each coordinate point in the coordinate data, and set the coordinate measurement grid for the location of the potential hazard;

[0173] By using a coordinate measurement grid, the coordinate points and normal vectors of each side of the vehicle are clustered to generate clusters;

[0174] Calculate the consistency of clusters based on the clusters;

[0175] Based on the consistency of the clusters, a 3D outline of the vehicle hazard location is generated;

[0176] Based on the 3D contour, determine the real-time coordinate data of traffic hazards.

[0177] The principle behind the above technical solution is as follows:

[0178] When determining the real-time coordinates of traffic hazards using hazard location data, this invention addresses the issue that traffic hazard objects have a certain outline. To achieve high-precision offsetting of objects posing traffic hazards, this invention first determines the coordinate data of the traffic hazard location and performs orientation determination. After orientation determination, precise measurement is performed using a coordinate measurement grid to accurately determine the 3D outline of the traffic hazard.

[0179] The beneficial effects of the above technical solution are as follows:

[0180] In the process of dynamically acquiring coordinate data, this invention can determine the location of traffic hazards by calculating normal vectors, and then locate the hazards using a coordinate grid. However, to achieve more accurate hazard identification, calculations are required. The coordinate grid can arrange the data points to achieve data clustering, generating clusters. 3D contour positioning is then performed using these clusters to determine the real-time coordinate data of the traffic hazards.

[0181] Preferably, the relative position data includes:

[0182] Based on real-time coordinate data of traffic hazards, determine the relative angle information of the location of the traffic hazard with respect to different endpoints of the vehicle;

[0183] Based on the relative angle information, determine the direction angle information of the traffic hazard relative to the vehicle;

[0184] Based on the azimuth information, determine the distribution data of vehicles corresponding to traffic hazards in the traffic image;

[0185] Based on the distribution data, a three-dimensional scene coordinate system is constructed with the vehicle as the origin.

[0186] Based on the three-dimensional scene coordinate system, the relative position information of different traffic hazards relative to vehicles is marked, and relative position modeling is performed.

[0187] The principle behind the above technical solution is as follows:

[0188] When determining the relative location of a traffic hazard, this invention uses lidar to obtain the directional angles of the corresponding traffic hazard elements at the location of the traffic hazard. The directional angle data is used to determine the distribution location of the traffic hazard, and then a three-dimensional scene coordinate system is used to model the location of the traffic hazard.

[0189] The beneficial effects of the above technical solution are as follows:

[0190] In the process of relative position modeling, relative angle information—that is, the overall relative position of the traffic hazard location from the current vehicle—can be used to determine the relative orientation and relative distance. This allows for position modeling of different traffic hazard locations, whether there are multiple or a single hazard, achieving highly accurate position ranging. This invention solves the problem of fuzzy positioning in existing technologies, enabling precise positioning and thus achieving micro-distance detection of vehicles in mountainous and narrow areas, which is also more suitable for the field of autonomous driving.

[0191] Preferably, the safe distance data includes:

[0192] Acquire target ranging images of different endpoints of traffic hazards relative to vehicles;

[0193] First and second measurement points are set at the locations of safety hazards in the target ranging image, and reference lines are set in the target ranging image;

[0194] Based on the actual shape and size of traffic hazards and their corresponding pixel distances, calibration parameters are established for the actual distances in the image and their corresponding pixel distances.

[0195] The angle between the distance measuring line formed by the line connecting the first and second measuring points and the reference line is used as the basis for selecting calibration parameters. At this time, the product of the pixel distance of the distance measuring line and the selected calibration parameters is the measurement distance of the traffic hazard.

[0196] The working principle of the above technical solution is as follows:

[0197] This invention, when measuring vehicle distances using different endpoints, identifies different measurement points based on different target detection images. By setting reference lines at these measurement points—lines used to mark and represent distances displayed on the user's screen—the measured distance can be quickly displayed using the actual shape and size of the traffic hazard and its corresponding pixel distance.

[0198] The beneficial effects of the above technical solution are as follows:

[0199] When selecting the measurement distance, this invention uses the angle between the measuring line formed by the line connecting the first and second measuring points and the reference line as the basis for selecting the calibration parameter, which can ensure the accuracy of the measurement distance. Then, the product of the pixel distance of the measuring line and the selected calibration parameter is the measurement distance of the traffic hazard, which conforms to the geometric calculation principle and the calculation result is more accurate and faster.

[0200] Preferably, the method further includes:

[0201] Acquire radar sensing data within the target sensing area of ​​the lidar, where,

[0202] Each radar sensing data corresponds to the image captured by the camera device at the corresponding viewpoint, and the set of video images includes images of the safety hazard from various viewpoints;

[0203] When a safety hazard appears in multiple video frames within a set of video footage, the ranging data of the safety hazard in the multiple video frames is accumulated;

[0204] An alarm will be triggered based on the ranging data.

[0205] The working principle of the above technical solution is as follows:

[0206] In the process of alarming using lidar, under normal conditions, the lidar only displays a three-dimensional image and test data; a single video may show multiple perspectives. If multiple video feeds present safety hazards, distance measurement alarms are required.

[0207] The beneficial effects of the above technical solution are as follows:

[0208] This invention addresses the problem in existing technologies where lidar data and camera data cannot be fused and calculated. This invention is based on the combined calculation of multiple methods, including hazard accumulation, lidar, and perspective shooting, to achieve ranging alarm.

[0209] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A high-precision positioning method for road traffic safety hazards, characterized in that, The method comprises the following steps: constructing a dynamic coordinate system based on a laser radar according to a sensing range of the laser radar; obtaining traffic image data by a camera, and positioning hidden dangers by the traffic image data through a traffic hidden danger identification model; determining real-time coordinate data of traffic hidden dangers according to the hidden danger positioning and the dynamic coordinate system; determining relative position data and safety distance data of traffic hidden dangers from a vehicle according to the real-time coordinate data; the dynamic coordinate system comprises: pre-setting geometric parameters of the vehicle to determine geometric end points of the vehicle; constructing geometric coordinate systems of the vehicle respectively according to the geometric end points; wherein, the geometric coordinate system comprises: a first coordinate system set of a left end point of the vehicle, a second coordinate system set of a right end point of the vehicle, a third coordinate system set of a front end point of the vehicle, and a fourth coordinate system set of a rear end point of the vehicle; obtaining satellite navigation data of the vehicle to determine a traffic route of the vehicle; fusing the geometric coordinate system to the traffic route to generate a dynamic geometric coordinate system based on a traffic trajectory; generating a dynamic coordinate system set based on the geometric end points through the dynamic geometric coordinate system; updating the laser radar data in real time by the user and the dynamic coordinate system set; the method further comprises: receiving traffic image data of the camera, and inputting the corresponding traffic hidden danger identification model respectively; determining a target hidden danger factor corresponding to the target traffic hidden danger identification model from a plurality of traffic images; wherein, each traffic image stores hidden danger comparison benchmark features corresponding to traffic hidden dangers; comparing at least one hidden danger comparison benchmark feature in the traffic image with the traffic hidden danger model to obtain at least one target comparison result; determining a target identification result according to the at least one target comparison result to determine hidden danger positioning data of the traffic hidden danger; the method further comprises: determining coordinate data of the corresponding position on the dynamic coordinate system through the laser radar according to the hidden danger positioning data; calculating normal vectors of each coordinate point in the coordinate data, and setting a coordinate measurement grid of the hidden danger position; generating a clustering cluster by clustering the coordinate points and the normal vectors on each side of the vehicle through the coordinate measurement grid; calculating the consistency of the clustering cluster according to the clustering cluster; generating a 3D contour of the hidden danger positioning position of the vehicle according to the consistency of the clustering cluster; determining real-time coordinate data of the traffic hidden danger according to the 3D contour.

2. The method of claim 1, wherein the method comprises: the method further comprises: dividing the sensing accuracy of the laser radar into rings according to the sensing range of the laser radar to generate a plurality of ring-shaped sensing areas; wherein, each ring-shaped sensing area corresponds to a sensing accuracy of the laser radar; determining a deviation ring-shaped sensing area lower than a preset sensing accuracy according to the sensing accuracy; setting an ultrasonic auxiliary measurement sensing device according to the deviation ring-shaped sensing area; synchronously sensing obstacle data in the deviation ring-shaped sensing area according to the ultrasonic auxiliary measurement sensing device; performing sensing compensation according to the synchronous sensing.

3. The method of claim 1, wherein the method comprises: the traffic image data comprises: obtaining a plurality of traffic images collected by a camera, inertial measurement component measured inertial navigation data, and positioning data obtained by a positioning system, wherein, the camera, the inertial measurement component, and the positioning system are loaded on the same traffic image collection vehicle; Determine whether the travel speed of the traffic image collection vehicle is lower than a preset speed threshold based on the traffic image, inertial navigation data, and positioning data; If it is determined that the travel speed of the traffic image collection vehicle is lower than the preset speed threshold, the traffic image is regarded as invalid data; If it is determined that the travel speed of the traffic image collection vehicle is higher than the preset speed threshold, the traffic image is regarded as valid data.

4. The method of claim 1, wherein the method comprises: The traffic hazard identification model includes a weather-based meteorological hazard model, a traffic volume-based traffic hazard model, a geographic environment-based geographic hazard model, an obstacle-based collision hazard model, a road-based road hazard model, and a navigation signal-based communication hazard model; wherein The meteorological hazard model is used to determine the driving hazards of the vehicle under different weather conditions; The traffic hazard model is used to determine the accident hazards of the vehicle under different traffic density scenarios; The geographic hazard model is used to determine the travel hazards of the vehicle under different geographic environments; The collision hazard model is used to determine the collision hazards of the vehicle under different obstacles; The road hazard model is used to determine the travel speed hazards of the vehicle under different road conditions; The communication hazard model is used to determine the communication hazards of the vehicle under different communication conditions and communication strengths.

5. The method of claim 1, wherein the method comprises: The relative position data includes: According to the real-time coordinate data of the traffic hazard, determine the relative angle information of the traffic hazard position relative to the different endpoints of the vehicle; According to the relative angle information, determine the direction angle information of the traffic hazard relative to the vehicle; According to the direction angle information, determine the distribution data of the traffic hazard relative to the vehicle in the traffic image; According to the distribution data, construct a three-dimensional scene coordinate system based on the vehicle as the coordinate origin; According to the three-dimensional scene coordinate system, mark the relative position information of different traffic hazards relative to the vehicle and perform relative position modeling.

6. The method of claim 1, wherein the method comprises: The safety distance data includes: Obtain a target ranging image of different endpoints of the traffic hazard relative to the vehicle; Set a first measurement point and a second measurement point for the safety hazard position in the target ranging image, and set a reference line in the target ranging image; From the actual shape and size of the traffic hazard and its corresponding pixel distance, develop a calibration parameter of the actual distance in the image and its corresponding pixel distance; The angle between the ranging line formed by the first measurement point and the second measurement point and the reference line is used as the selection basis for the calibration parameter, and the product of the pixel distance of the ranging line and the selected calibration parameter is the measured distance of the traffic hazard.

7. The method of claim 1, wherein the method comprises: The method further includes: Obtain radar sensing data in the target sensing area of the laser radar, wherein Each radar sensing data corresponds to a group of video pictures taken by the camera at the corresponding viewing angle, and the group of video pictures includes pictures of the safety hazard at each viewing angle; In the case that the safety hazard appears in multiple video pictures in the group of video pictures, the ranging data of the safety hazard in the multiple video pictures is accumulated; Alarm according to the ranging data.

Images