Roadside unit trustworthiness ranking method and system based on data skeleton similarity

Abstract

This invention belongs to the field of intelligent transportation technology, specifically disclosing a method and system for roadside unit (Roadside Unit) credibility rating based on data skeleton similarity. The method includes the following steps: a requesting vehicle sends a request message to a Roadside Unit; the Roadside Unit extracts the position (x, y), speed v, and direction angle θ features from the requesting vehicle's message, constructs a feature matrix, calculates the spatial position difference, relative speed difference trend, and direction angle similarity between the requesting vehicle and surrounding target vehicles; uses the vector cross product method for geometric occlusion to determine the visibility of surrounding target vehicles to the requesting vehicle, and transmits this information back to the requesting vehicle's On-Board Unit (OBU); the OBU then calculates the credibility of the Roadside Unit data. Using this technical solution, the accuracy of calculating the credibility of Roadside Unit collected data is higher, thereby determining the application of the collected data in vehicle-side decision-making.

Description

Technical Field

[0001] This invention belongs to the field of intelligent transportation technology and relates to a roadside unit credibility rating method and system based on data skeleton similarity. Background Technology

[0002] With the rapid increase in the number of motor vehicles, urban traffic conditions are becoming increasingly complex. Ensuring traffic safety and improving traffic efficiency have become two core challenges in current urban development. Against this backdrop, the intelligent transportation industry has entered a period of rapid development, with numerous solutions based on vehicle intelligence and connected intelligence emerging.

[0003] However, given the complexity of urban road conditions, single-vehicle intelligence cannot provide a comprehensive solution.

[0004] In contrast, vehicle-to-infrastructure (V2I) systems integrate dynamic traffic information from people, vehicles, roads, and the cloud, and utilize Vehicle-to-Everything (V2X) communication technology to achieve information interconnection. This enables the sharing of traffic information such as vehicle and road conditions, effectively assisting vehicles and other traffic participants in making safer and more efficient decisions. In this context, vehicles in V2I systems receive and process data from roadside units via their onboard units (OBUs). This data is categorized into different levels based on its impact on vehicle control. However, due to differences in the time and spatial location of data collection between roadside units and vehicles, inconsistencies may arise between the data.

[0005] Existing technologies do not consider the computing power limitations of vehicle-mounted terminals. Filtering and screening the data transmitted from roadside units leads to significant processing delays on the vehicle side, affecting data real-time performance. Furthermore, they do not compare and calculate the content of data packets transmitted between roadside units and vehicle-mounted terminals, failing to account for the differences between data collected by roadside units and the vehicle itself, thus impacting the accuracy of decisions made by the vehicle-mounted terminal using data from roadside units. Summary of the Invention

[0006] The purpose of this invention is to provide a roadside unit credibility rating method and system based on data skeleton similarity, so as to obtain accurate roadside unit data credibility and solve the problem of discrepancies between roadside unit and vehicle data.

[0007] To achieve the above objectives, the basic solution of this invention is: a roadside unit credibility rating method based on data skeleton similarity, comprising the following steps:

[0008] The requesting vehicle sends a request message to the roadside unit, which selects a target vehicle within a radius R around the requesting vehicle based on the location of the requesting vehicle, and determines the target time point;

[0009] The roadside unit extracts the position (x,y), speed v, and direction angle θ features from the information sent by the requesting vehicle based on the target time point, and constructs a feature matrix.

[0010] Based on the feature matrix, the spatial position difference, relative speed difference trend and similarity of direction angle between the requesting vehicle and surrounding target vehicles are calculated respectively.

[0011] Based on the spatial position differences, relative speed differences, and similarity of direction angles between the requesting vehicle and surrounding target vehicles, the visibility of surrounding target vehicles to the requesting vehicle is determined by the vector cross product method of geometric occlusion.

[0012] The roadside unit transmits the visibility data of surrounding target vehicles to the requesting vehicle to the on-board unit (OBU) of the requesting vehicle. The OBU matches and compares the target vehicle information in the visibility data with the data of the target vehicle collected by the OBU itself, and calculates the credibility of the roadside unit data.

[0013] The working principle and beneficial effects of this basic solution are as follows: This technical solution first requires the vehicle's onboard unit (OBU) to proactively send a request to the roadside unit to obtain local information about its surrounding environment. Upon receiving the request, the roadside unit generates and transmits a highly customized data packet based on the vehicle's specific needs and the current environmental conditions. After receiving this data packet, the onboard unit extracts information correlated with data collected by the vehicle's own sensors, aligns and compares this information, and thus calculates the reliability of the roadside unit's data.

[0014] This not only enhances the vehicle's ability to perceive external information, but also, through multi-sensor fusion technology, improves the accuracy of overall data, thereby reducing the limitations of a single data source. It improves the reliability and stability of the vehicle-road cooperative system, enhances the vehicle's decision-making ability in complex traffic environments, and thus improves overall driving safety and traffic efficiency. It effectively solves the problem of data discrepancies between roadside units and vehicles, and also promotes the widespread application of vehicle-road cooperative technology in actual urban traffic management.

[0015] Furthermore, the requesting vehicle sends a request message to the roadside unit. The roadside unit selects a target vehicle within a radius R around the requesting vehicle based on the location of the requesting vehicle, and determines the target time point using the following method:

[0016] Request vehicle A to use T s It collects its own and surrounding vehicles' driving information at fixed time intervals, in T A Vehicle A is constantly requested to send its current location information, timestamp, and collection frequency information as a high-frequency measurement dataset to the roadside unit.

[0017] The roadside unit queries N target vehicles within a radius R around the requesting vehicle A based on location information, and calculates the target time point T. 目标 :

[0018] T 目标 =T A +T s .

[0019] Obtaining the target time point is beneficial for subsequent use.

[0020] Furthermore, the roadside unit extracts the position (x, y), speed v, and direction angle θ features from the information sent by the requesting vehicle based on the target time point, and constructs the feature matrix using the following method:

[0021] The roadside unit selects the timestamp and T from the high-frequency measurement dataset of the requested vehicle. 目标 Extract position (x, y), velocity v, and orientation angle θ features from the same or closest set of data to construct a feature matrix X:

[0022]

[0023] Where, x N y N v N θ N These represent the x-coordinate, y-coordinate, speed, and direction angle of the Nth vehicle, respectively.

[0024] Construct a feature matrix and filter to obtain the required feature data for subsequent visibility filtering.

[0025] Furthermore, based on the feature matrix, the steps for calculating the spatial position difference between the requesting vehicle and surrounding target vehicles are as follows:

[0026] Perform min-max normalization on the eigenvalues ​​of the feature matrix X to scale each eigenvalue to the range of 0 to 1:

[0027]

[0028] Among them, X i ′ j The feature matrix after maxima-minus normalization; X ij Let X be the j-th feature value of the i-th vehicle. j Let j be the j-th eigenvalue of the eigenvalue matrix;

[0029] For the location feature value (x, y), the Mahalanobis distance is used to calculate the spatial position difference D between the requesting vehicle A and the surrounding target vehicles. 马氏距离(i,j) :

[0030]

[0031] Among them, P A It requests the location feature vector of vehicle A, P. 其他车辆 is the position feature vector of the surrounding target vehicles, and S is the covariance matrix of the position features.

[0032] For the location feature value (x,y), Mahalanobis distance is used to calculate the spatial position difference between the requesting vehicle A and the surrounding target vehicles, which is simple to operate.

[0033] Furthermore, based on the feature matrix, the changing trend Δv of the relative speed difference between the requesting vehicle and surrounding target vehicles is calculated. 趋势 ,for:

[0034]

[0035] Among them, v A (t) represents the speed of vehicle A at time t, v B (t) is the speed of the surrounding target vehicles at time t.

[0036] For the rate characteristic value, calculate the changing trend of the relative speed difference between the two vehicles for later use.

[0037] Furthermore, based on the feature matrix, the similarity of the heading angles between the requesting vehicle and surrounding target vehicles is calculated. The steps are as follows:

[0038] Calculate the direction from vehicle A to target vehicle B. i Direction vector

[0039]

[0040] in, For target vehicle B i Location information; x A y A To request the location information of vehicle A;

[0041] Calculate the direction vector and the target vehicle B i The angle cos(Δθ) between the driving directions is used to measure the consistency of direction through cosine similarity:

[0042]

[0043] in, Indicates target vehicle B i The direction of travel.

[0044] For the direction angle feature value, calculate the similarity of the direction angles of the requesting vehicle A and the surrounding vehicles, which is beneficial for use.

[0045] Furthermore, the visibility of surrounding target vehicles to the requesting vehicle is determined using the vector cross product method of geometric occlusion. The specific steps are as follows:

[0046] For requesting vehicle A and target vehicle B i Define a line-of-sight segment L: from the position (x, y) of the requesting vehicle A A ,y A ) to target vehicle B i Location

[0047] For vehicle B with obstacles j Its rectangular boundary consists of four vertices (x, y, z). j1 ,y j1 ),(x j2 ,y j2 ),(x j3 ,y j3 ),

[0048] (x j4 ,y j4 These four lines, forming the vehicle's four boundary lines, are used to determine whether the view is obstructed.

[0049] Based on the vector cross product method, the intersection of line segments is determined as follows:

[0050] Direction vector of line of sight segment L Defined as:

[0051]

[0052] in, For target vehicle B i Location information; x A y A To request the location information of vehicle A;

[0053] Obstacle boundary segment for:

[0054]

[0055] For these two vectors, we determine whether they are parallel by their cross product. If they are not parallel, we further determine whether they intersect.

[0056] Determine the endpoints of a line segment:

[0057] For non-parallel line-of-sight segment L and obstacle vehicle B j For boundary line segments, determine whether their endpoints are located on different sides of the other line segment. Assume the obstacle vehicle B needs to be judged. j The boundary line segment starts from point B. j1 =(x j1 ,yj1 ) to point B j2 =(x j2 ,y j2 ):

[0058]

[0059] If the signs of D1 and D2 are different, it indicates that the obstacle vehicle is B. j The two endpoints of the boundary line segment are located on different sides of the line of sight segment, which means that the line of sight segment and obstacle vehicle B j If the boundary segments may intersect, then mark target vehicle B. i It is invisible.

[0060] The vehicle filtering method based on vector cross product visibility determines the current visibility of surrounding vehicles to the requested vehicle by the vehicle's location, reducing the data computation of the vehicle terminal for invisible vehicles and alleviating the computational load of the vehicle terminal.

[0061] Furthermore, the steps for determining the future visibility of surrounding target vehicles to the requesting vehicle are as follows:

[0062] The future relative velocity v at a future time point t+Δt 相对 (t+Δt) passes through the current relative velocity v 相对 (t) and velocity trend Δv 趋势 (t) is used for prediction:

[0063] v 相对 (t+Δt)=v 相对 (t)+Δv 趋势 (t)·Δt;

[0064] For a certain car B, suppose its position at time t is (x B ,y B The velocity is v and the direction angle is θ. B At a future time t+Δt, the position (x′) of vehicle B B ,y′ B )for:

[0065] (x′ B ,y′ B )=(x B +v 相对 (t+Δt)·cos(θ B )·Δt,y B +v 相对 (t+Δt)·sin(θ B )·

[0066] Δt);

[0067] Based on this prediction, the requesting vehicle A and the target vehicle B are identified. i and obstacle vehicle B j The location at a future point in time is then determined using the vector cross product method to assess visibility.

[0068] The vehicle filtering method based on vector cross product visibility determines the current and future visibility of surrounding vehicles to the requesting vehicle by using the vehicle's location. This reduces the data computation of the vehicle terminal for invisible vehicles and alleviates the computational load on the vehicle terminal.

[0069] Furthermore, a comprehensive assessment of future visibility:

[0070] Whether target vehicle B will be visible to requesting vehicle A at a future time depends on whether the following conditions are met:

[0071] Current visibility: Target vehicle B is currently visible to requesting vehicle A;

[0072] Future occlusion conditions are visible: at a future time point, the line of sight of target vehicle B and requesting vehicle A will not be obstructed;

[0073] The directions are consistent or the differences are small: cos(Δθ)≈1, that is, the directions of the target vehicle B and the requesting vehicle A are highly consistent;

[0074] The speed trend shows a close approximation of: v 相对 (t+Δt)<0, the speed trend of target vehicle B shows that it is gradually approaching the requesting vehicle A;

[0075] When all conditions are met, it is considered that the target vehicle B is visible to the requesting vehicle A in the current moment and in the future, which means it passes the visibility screening.

[0076] Iterate through the feature matrix, filter, and return the data:

[0077] Iterate through all vehicles in the feature matrix and perform visibility filtering on them;

[0078] For vehicles that pass the visibility screening, the roadside unit sends multiple sets of measurement data of the vehicle at different times to the requesting vehicle A for continuous calculation of credibility.

[0079] The raw vehicle driving data collected by the roadside unit was filtered based on current and future visibility, and vehicle nodes with strong correlation to the requesting vehicle were selected and returned. In dynamic traffic scenarios with high traffic volume, this can effectively reduce the computing load of the vehicle terminal, while ensuring the accuracy of the credibility calculation.

[0080] Furthermore, the method for calculating the reliability of roadside unit data by the on-board unit (OBU) is as follows:

[0081] Data reception and matching: After receiving the data returned from the roadside unit, vehicle A identifies the corresponding vehicle from the vehicles it has detected based on the vehicle information;

[0082] Data validity check: After identifying the corresponding vehicle, check whether the feature values ​​in the returned data are valid data, specifically:

[0083] If a key feature is invalid data, the data for that target vehicle will be discarded.

[0084] Data can be filled when non-critical features are null, or only at the moment of partial measurement.

[0085] Based on the graph structure formed by the roadside unit for the requesting vehicle A and its surrounding vehicles that have passed the screening, and the graph structure generated by the requesting vehicle A and its surrounding vehicles with the same characteristics, the similarity score of edges and nodes is calculated through matrix iteration to evaluate the overall similarity. Specifically:

[0086] Based on the measurement and screening of roadside units, an undirected complete graph G is generated for the requesting vehicle A and the surrounding vehicles that have passed the screening. A Based on the measurements of the requesting vehicle A and other vehicles in the same neighborhood selected by roadside units, map G is generated in the same manner. B ;

[0087] The node similarity score S is calculated using matrix iteration. N S and edge similarity score E Finally, the overall similarity score S is calculated. G :

[0088] S G =S N +S E

[0089] Compare S G With τ G , τ G For a pre-set threshold, when S G >τ G When the confidence level is considered low, proceed to the next step of calculating the confidence level of the speed measurement value.

[0090] Calculate the mean square error (MSE) of the speed measurements of the roadside unit and the requesting vehicle A relative to the speeds of other vehicles in the surrounding area. v :

[0091]

[0092] Among them, v R,t v is the measured velocity of the target vehicle by the roadside unit at time t. A,tTo request the measurement value of vehicle A's speed relative to the target vehicle at time t, N is the total number of surrounding vehicles;

[0093] Comparison of MSE v With τ v , τ v For a pre-set threshold, when MSE v <τ v When the confidence level is reached, the next step of calculating the confidence level of the orientation angle measurement value is initiated.

[0094] Calculate the mean square error (MSE) of the heading angle measurements of the roadside unit and the requesting vehicle A relative to the same surrounding vehicles;

[0095] Consider the periodicity of the orientation angle:

[0096] Δθ t =min(|θ R,t -θ A,t |,360°-|θ R,t -θ A,t |)

[0097] Where, θ R,t Let θ be the direction angle measured by the roadside unit for the target vehicle at time t. A,t To request the direction angle Δθ that vehicle A measures relative to the target vehicle at time t. t The difference in the direction angle measurement between the roadside unit and the requesting vehicle A relative to the target vehicle at time t, ranging from [0°, 180°];

[0098] Mean Square Error of Direction Angle (MSE) θ for:

[0099]

[0100] Comparison of MSE θ With τ θ , τ θ For a pre-set threshold, when MSE θ <τ θ At that time, it was considered to have reached a high level of credibility.

[0101] For the data skeletons constructed based on the data returned by roadside units and the data detected by the requesting vehicle itself, the graph similarity is first calculated to assess the overall similarity. Then, the differences in other feature values ​​are calculated to gradually classify the data. This allows for effective credibility assessment and classification of roadside units with different data collection differences, thereby determining the application of their collected data in vehicle-side decision-making.

[0102] Furthermore, according to the pre-set τ 位置 , τ v , τ θAnd the similarity S to the overall graph G and MSE v and MSE θ The calculation results are used to assign different confidence levels C. 等级 :

[0103]

[0104] Based on the trust level, a decision is made as to whether requesting vehicle A should use the data provided by the roadside unit for subsequent decision-making.

[0105] The credibility grading calculation method based on overall graph similarity and item-by-item differences constructs a data skeleton based on roadside unit data and vehicle terminal data, and performs similarity analysis and calculation step by step from the overall level to specific feature values ​​to evaluate the credibility level of roadside unit data.

[0106] The present invention also provides a roadside unit credibility rating system based on the method described in the present invention, including a roadside unit and an on-board unit (OBU). The on-board unit (OBU) is used to collect the current requesting vehicle's own location information, timestamp, and collection frequency information, and send them to the roadside unit.

[0107] The roadside unit is used to acquire visibility data of surrounding target vehicles to the requesting vehicle and transmit it back to the on-board unit (OBU) of the requesting vehicle. The OBU then calculates the reliability of the roadside unit data.

[0108] This system processes the raw vehicle driving data collected by roadside units, adds the timestamps and locations of the requested vehicles, and returns the vehicle data, filtered by visibility, to the requesting vehicle. The onboard terminal then constructs a data skeleton using the returned vehicle feature values ​​and compares it with its own detected data to calculate similarity and evaluate the reliability of the roadside units. Attached Figure Description

[0109] Figure 1 This is a flowchart illustrating the roadside unit credibility rating method based on data skeleton similarity according to the present invention. Detailed Implementation

[0110] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0111] In the description of this invention, it should be understood that the terms "longitudinal", "lateral", "up", "down", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0112] In the description of this invention, unless otherwise specified and limited, it should be noted that the terms "installation", "connection" and "linking" should be interpreted broadly. For example, they can refer to mechanical or electrical connections, or internal connections between two components. They can be direct connections or indirect connections through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms according to the specific circumstances.

[0113] This invention discloses a roadside unit credibility rating method based on data skeleton similarity, which achieves data verification and credibility assessment between vehicles and roadside units through an interactive data request and response mechanism. Figure 1 As shown, the roadside unit credibility rating method based on data skeleton similarity includes the following steps:

[0114] The vehicle requests to send a request to the roadside unit (the vehicle's on-board unit (OBU) actively sends a request to the roadside unit). The roadside unit selects a target vehicle within a radius R around the requesting vehicle based on the location of the requesting vehicle and determines the target time point.

[0115] The roadside unit extracts the position (x,y), speed v, and direction angle θ features from the information sent by the requesting vehicle based on the target time point, and constructs a feature matrix.

[0116] Based on the feature matrix, the spatial position difference, relative speed difference trend and similarity of direction angle between the requesting vehicle and surrounding target vehicles are calculated respectively.

[0117] Based on the spatial positional differences, relative speed differences, and similarity of azimuth angles between the requesting vehicle and surrounding target vehicles, the visibility of surrounding target vehicles to the requesting vehicle is determined using the vector cross product method of geometric occlusion. Target vehicles that meet the detection range of the requesting vehicle are selected from the multi-vehicle data collected from roadside units. Taking into account the current and future visibility of target vehicles to the requesting vehicle in dynamic traffic scenarios, the vehicle-mounted terminal of the requesting vehicle can calculate a temporally continuous roadside unit credibility score.

[0118] The roadside unit (Roadside Unit) transmits visibility data of surrounding target vehicles to the requesting vehicle's On-Board Unit (OBU). The OBU then matches and compares the target vehicle information from the visibility data with its own collected data for that target vehicle, calculating the reliability of the Roadside Unit data. This enhances the vehicle's ability to perceive external information, and multi-sensor fusion technology further improves the accuracy of the overall data, thereby reducing the limitations of a single data source.

[0119] In a preferred embodiment of the present invention, the requesting vehicle sends a request message to the roadside unit, and the roadside unit selects a target vehicle within a radius R around the requesting vehicle based on the location of the requesting vehicle, and determines the target time point using the following method:

[0120] Request vehicle A to use T s It collects its own and surrounding vehicles' driving information at fixed time intervals, in T A Vehicle A is constantly requested to send its current location information, timestamp, and collection frequency as a high-frequency measurement dataset to the roadside unit.

[0121] The roadside unit queries N target vehicles within a radius R around the requesting vehicle A based on location information, and calculates the target time point T. 目标 :

[0122] T 目标 =T A +T s .

[0123] In a preferred embodiment of the present invention, the roadside unit extracts the position (x, y), speed v, and direction angle θ features from the information sent by the requesting vehicle based on the target time point, and constructs the feature matrix as follows:

[0124] The roadside unit selects the timestamp and T from the high-frequency measurement dataset of the requested vehicle. 目标 Extract position (x, y), velocity v, and orientation angle θ features from the same or closest set of data to construct a feature matrix X:

[0125]

[0126] Where, x N y N v N θ N These represent the x-coordinate, y-coordinate, speed, and direction angle of the Nth vehicle, respectively.

[0127] In a preferred embodiment of the present invention, the step of calculating the spatial position difference between the requesting vehicle and surrounding target vehicles based on the feature matrix is ​​as follows:

[0128] Perform min-max normalization on the eigenvalues ​​of the feature matrix X to scale each eigenvalue to the range of 0 to 1:

[0129]

[0130] Among them, X i ′ j The feature matrix after maxima-minus normalization; X ij Let X be the j-th feature value of the i-th vehicle. j X is the j-th eigenvalue of the eigenvalue matrix; in the eigenvalue matrix, the j-th eigenvalue X in the first row is... 1j and the j-th feature value X in the second row 2j , representing the j-th eigenvalue of the first vehicle and the j-th eigenvalue of the second vehicle, respectively. Therefore, from the perspective of matrix rows and columns, X j It refers to the j-th column, which is the j-th feature value of all vehicles.

[0131] For the location feature value (x, y), the Mahalanobis distance is used to calculate the spatial position difference D between the requesting vehicle A and the surrounding target vehicles. 马氏距离(i,j) :

[0132]

[0133] Among them, P A It requests the location feature vector of vehicle A, P. 其他车辆 is the position feature vector of the surrounding target vehicles, and S is the covariance matrix of the position features.

[0134] In a preferred embodiment of the present invention, the changing trend Δv of the relative speed difference between the requesting vehicle and surrounding target vehicles is calculated based on the feature matrix. 趋势 ,for:

[0135]

[0136] Among them, v A (t) represents the speed of vehicle A at time t, v B (t) is the speed of the surrounding target vehicles at time t.

[0137] In a preferred embodiment of the present invention, the similarity of the heading angles of the requesting vehicle and surrounding target vehicles is calculated based on the feature matrix, and the steps are as follows:

[0138] Calculate the direction from vehicle A to target vehicle B. i Direction vector

[0139]

[0140] in, For target vehicle B i Location information; x A y A To request the location information of vehicle A;

[0141] Calculate the direction vector and the target vehicle B i The angle cos(Δθ) between the driving directions is used to measure the consistency of direction through cosine similarity:

[0142]

[0143] in, Indicates target vehicle B i The direction of travel.

[0144] In a preferred embodiment of the present invention, the current visibility of surrounding target vehicles to the requesting vehicle is determined by the vector cross product method of geometric occlusion. The specific steps are as follows:

[0145] For requesting vehicle A and target vehicle B i Define a line-of-sight segment L: from the position (x, y) of the requesting vehicle A A ,y A ) to target vehicle B i Location

[0146] For vehicle B with obstacles j Its rectangular boundary consists of four vertices (x, y, z). j1 ,y j1 ),(x j2 ,y j2 ),(x j3 ,y j3 ),

[0147] (x j4 ,y j4 These four lines, forming the vehicle's four boundary lines, are used to determine whether the view is obstructed.

[0148] Based on the vector cross product method, the intersection of line segments is determined as follows:

[0149] Direction vector of line of sight segment L Defined as:

[0150]

[0151] in, For target vehicle B i Location information; x A y A To request the location information of vehicle A;

[0152] Obstacle boundary segment (e.g., vehicle B) j A certain boundary is:

[0153]

[0154] For these two vectors, we determine whether they are parallel by their cross product. If they are not parallel, we further determine whether they intersect.

[0155] Determine the endpoints of a line segment:

[0156] For non-parallel line-of-sight segment L and obstacle vehicle B j For boundary line segments, determine whether their endpoints are located on different sides of the other line segment. Assume the obstacle vehicle B needs to be judged. j The boundary line segment starts from point B. j1 =(x j1 ,y j1 ) to point B j2 =(x j2 ,y j2 ):

[0157]

[0158] If the signs of D1 and D2 are different, it indicates that the obstacle vehicle is B. j The two endpoints of the boundary line segment are located on different sides of the line of sight segment, which means that the line of sight segment and obstacle vehicle B j If the boundary segments may intersect, then mark target vehicle B. i It is invisible.

[0159] In a preferred embodiment of the present invention, the step of determining the future visibility of surrounding target vehicles to the requesting vehicle is as follows:

[0160] The future relative velocity v at a future time point t+Δt 相对 (t+Δt) passes through the current relative velocity v 相对 (t) and velocity trend Δv 趋势 (t) is used for prediction:

[0161] v 相对 (t+Δt)=v 相对 (t)+Δv 趋势 (t)·Δt;

[0162] Future position prediction: For a certain car B, suppose its position at time t is (x B ,y B The velocity is v and the direction angle is θ. B At a future time t+Δt, the position (x′) of vehicle B B ,y′ B )for:

[0163] (x′ B ,y′ B )=(x B +v 相对 (t+Δt)·cos(θ B )·Δt,y B +v 相对 (t+Δt)·sin(θ B )·

[0164] Δt);

[0165] Based on this prediction, the requesting vehicle A and the target vehicle B are identified. i and obstacle vehicle B j Location at a future point in time (B) j For vehicles that are obstacles, it is necessary to determine whether they obstruct the path from vehicle A to target vehicle B. i Line of sight, thus determining B i (The visibility of A in the future) is then determined using the vector cross product method.

[0166] In a preferred embodiment of the present invention, future visibility is comprehensively assessed:

[0167] Whether target vehicle B will be visible to requesting vehicle A at a future time depends on whether the following conditions are met:

[0168] Current visibility: Target vehicle B is currently visible to requesting vehicle A;

[0169] Future occlusion conditions are visible: at a future time point, the line of sight of target vehicle B and requesting vehicle A will not be obstructed;

[0170] The directions are consistent or the differences are small: cos(Δθ)≈1, that is, the directions of the target vehicle B and the requesting vehicle A are highly consistent;

[0171] The speed trend shows a close approximation of: v 相对 (t+Δt)<0, the speed trend of target vehicle B shows that it is gradually approaching the requesting vehicle A;

[0172] When all conditions are met, it is considered that the target vehicle B is visible to the requesting vehicle A in the current moment and in the future, which means it passes the visibility screening.

[0173] Iterate through the feature matrix, filter, and return the data:

[0174] Iterate through all vehicles in the feature matrix and perform visibility filtering on them;

[0175] For vehicles that pass the visibility screening, the roadside unit sends multiple sets of measurement data of the vehicle at different times to the requesting vehicle A for continuous calculation of credibility.

[0176] In a preferred embodiment of the present invention, the method for calculating the reliability of roadside unit data by the on-board unit (OBU) is as follows:

[0177] Data reception and matching: After receiving the data returned from the roadside unit, vehicle A identifies the corresponding vehicle from the vehicles it has detected based on vehicle information (vehicle ID, etc.);

[0178] Data validity check: After identifying the corresponding vehicle, check whether the feature values ​​in the returned data are valid data, specifically:

[0179] If a critical feature (such as position or speed) is invalid, it will significantly affect the accuracy of the credibility assessment; therefore, the data for that target vehicle should be discarded. When non-critical features (such as heading angle) are null, or only measured at the current moment, data imputation can be performed: Data imputation from the previous measurement: If a feature is missing at the current moment, it can be filled using the same feature value field from the previous measurement. If the missing feature is a continuous feature, it can be filled by linear interpolation or estimation based on data from multiple preceding and following time points.

[0180] Low-confidence calculation based on graph structure similarity:

[0181] Based on the graph structure formed by the roadside unit for the requesting vehicle A and its surrounding vehicles that have passed the screening, and the graph structure generated by the requesting vehicle A and its surrounding vehicles with the same characteristics, the similarity score of edges and nodes is calculated through matrix iteration to evaluate the overall similarity. Specifically:

[0182] Based on the measurement and screening of roadside units, an undirected complete graph G is generated for the requesting vehicle A and the surrounding vehicles that have passed the screening. A Based on the measurements of the requesting vehicle A and other vehicles in the same neighborhood selected by roadside units, map G is generated in the same manner. B ;

[0183] The node similarity score S is calculated using matrix iteration. N S and edge similarity score E Finally, the overall similarity score S is calculated. G :

[0184] S G =S N +S E

[0185] Compare S G With τ G , τ G For a pre-set threshold, when S G >τ GWhen the confidence level is considered low, proceed to the next step of calculating the confidence level of the speed measurement value.

[0186] Medium confidence calculation based on rate eigenvalues:

[0187] Calculate the mean square error (MSE) of the speed measurements of the roadside unit and the requesting vehicle A relative to the speeds of other vehicles in the surrounding area. v :

[0188]

[0189] Among them, v R,t v is the measured velocity of the target vehicle by the roadside unit at time t. A,t To request the measurement value of vehicle A's speed relative to the target vehicle at time t, N is the total number of surrounding vehicles;

[0190] Comparison of MSE v With τ v , τ v For a pre-set threshold, when MSE v <τ v When the confidence level is reached, the next step of calculating the confidence level of the orientation angle measurement value is initiated.

[0191] High-reliability calculation based on orientation angle eigenvalues:

[0192] Calculate the mean square error (MSE) of the heading angle measurements of the roadside unit and the requesting vehicle A relative to the same surrounding vehicles;

[0193] Consider the periodicity of the orientation angle:

[0194] Δθ t =min(|θ R,t -θ A,t |,360°-|θ R,t -θ A,t |)

[0195] Where, θ R,t Let θ be the direction angle measured by the roadside unit for the target vehicle at time t. A,t To request the direction angle Δθ that vehicle A measures relative to the target vehicle at time t. t The difference in the direction angle measurement between the roadside unit and the requesting vehicle A relative to the target vehicle at time t, ranging from [0°, 180°];

[0196] Mean Square Error of Direction Angle (MSE) θ for:

[0197]

[0198] Comparison of MSE θ With τ θ , τθ For a pre-set threshold, when MSE θ <τ θ At that time, it was considered to have reached a high level of credibility.

[0199] More preferably, according to a pre-set τ 位置 , τ v , τ θ And the similarity S to the overall graph G and MSE v and MSE θ The calculation results are used to assign different confidence levels C. 等级 :

[0200]

[0201] Based on the credibility level, it is determined whether requesting vehicle A should use the data provided by the roadside unit for subsequent decision-making. Through comprehensive analysis, differential analysis and calculation are performed step by step from the overall data to specific feature values, further classifying and evaluating the credibility of the roadside unit data, and providing a reliable reference for vehicle intelligent decision-making.

[0202] The present invention also provides a roadside unit credibility rating system based on the method described in the present invention, including a roadside unit and an on-board unit (OBU). The on-board unit (OBU) is used to collect its current location information, timestamp, and collection frequency information, and send them to the roadside unit.

[0203] The roadside unit is used to acquire visibility data of surrounding target vehicles to the requesting vehicle and send it back to the on-board unit (OBU) of the requesting vehicle. The OBU then calculates the reliability of the roadside unit data.

[0204] This invention processes the raw vehicle driving data collected by roadside units, adds the timestamp and location of the requested vehicle, and returns the vehicle data, filtered by visibility, to the requesting vehicle. The onboard terminal then constructs a data skeleton using the returned vehicle feature value data, compares it with its own detected data to calculate similarity, and evaluates the reliability of the roadside unit.

[0205] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0206] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A roadside unit credibility rating method based on data skeleton similarity, characterized in that, Includes the following steps: The requesting vehicle sends a request message to the roadside unit, which selects a target vehicle within a radius R around the requesting vehicle based on the location of the requesting vehicle, and determines the target time point; The roadside unit extracts the position (x,y), speed v, and direction angle θ features from the information sent by the requesting vehicle based on the target time point, and constructs a feature matrix. Based on the feature matrix, the spatial position difference, relative speed difference trend and similarity of direction angle between the requesting vehicle and surrounding target vehicles are calculated respectively. Based on the spatial position differences, relative speed differences, and similarity of direction angles between the requesting vehicle and surrounding target vehicles, the visibility of surrounding target vehicles to the requesting vehicle is determined by the vector cross product method of geometric occlusion. The roadside unit transmits the visibility data of surrounding target vehicles to the requesting vehicle to the on-board unit (OBU) of the requesting vehicle. The OBU matches and compares the target vehicle information in the visibility data with the data of the target vehicle collected by the OBU itself, and calculates the credibility of the roadside unit data.

2. The roadside unit credibility rating method based on data skeleton similarity as described in claim 1, characterized in that, The requesting vehicle sends a request message to the roadside unit. The roadside unit selects target vehicles within a radius R around the requesting vehicle based on the location of the requesting vehicle, and determines the target time point using the following method: Request vehicle A to use T s It collects driving information of itself and surrounding vehicles at fixed time intervals, in T A Vehicle A is constantly requested to send its current location information, timestamp, and collection frequency information as a high-frequency measurement dataset to the roadside unit. The roadside unit queries N target vehicles within a radius R around the requesting vehicle A based on location information, and calculates the target time point T. 目标 : T 目标 =T A +T s 。 3. The roadside unit credibility rating method based on data skeleton similarity as described in claim 1, characterized in that, The roadside unit extracts the position (x, y), speed v, and direction angle θ features from the information sent by the requesting vehicle based on the target time point, and constructs the feature matrix using the following method: The roadside unit selects the timestamp and T from the high-frequency measurement dataset of the requested vehicle. 目标 Extract position (x, y), velocity v, and orientation angle θ features from the same or closest set of data to construct a feature matrix X: Where, x N y N v N θ N These represent the x-coordinate, y-coordinate, speed, and direction angle of the Nth vehicle, respectively.

4. The roadside unit credibility rating method based on data skeleton similarity as described in claim 1, characterized in that, Based on the feature matrix, the steps to calculate the spatial position difference between the requesting vehicle and surrounding target vehicles are as follows: Perform min-max normalization on the eigenvalues ​​of the feature matrix X to scale each eigenvalue to the range of 0 to 1: Among them, X i ′ j The feature matrix after maxima-minus normalization; X ij Let X be the j-th feature value of the i-th vehicle. j Let j be the j-th eigenvalue of the eigenvalue matrix; For the location feature value (x, y), the Mahalanobis distance is used to calculate the spatial position difference D between the requesting vehicle A and the surrounding target vehicles. 马氏距离(i,j) : Among them, P A It requests the location feature vector of vehicle A, P. 其他车辆 is the position feature vector of the surrounding target vehicles, and S is the covariance matrix of the position features.

5. The roadside unit credibility rating method based on data skeleton similarity as described in claim 1, characterized in that, Based on the feature matrix, the changing trend Δv of the relative speed difference between the requesting vehicle and surrounding target vehicles is calculated. 趋势 ,for: Among them, v A (t) represents the speed of vehicle A at time t, v B (t) is the speed of the surrounding target vehicles at time t.

6. The roadside unit credibility rating method based on data skeleton similarity as described in claim 1, characterized in that, Based on the feature matrix, the similarity of the heading angles between the requesting vehicle and surrounding target vehicles is calculated. The steps are as follows: Calculate the direction from vehicle A to target vehicle B. i Direction vector in, For target vehicle B i Location information; x A y A To request the location information of vehicle A; Calculate the direction vector and the target vehicle B i The angle cos(Δθ) between the driving directions is used to measure the consistency of direction through cosine similarity: in, Indicates target vehicle B i The direction of travel.

7. The roadside unit credibility rating method based on data skeleton similarity as described in claim 1, characterized in that, The visibility of surrounding target vehicles to the requesting vehicle is determined using the vector cross product method of geometric occlusion. The specific steps are as follows: For requesting vehicle A and target vehicle B i Define a line-of-sight segment L: from the position (x, y) of the requesting vehicle A A ,y A ) to target vehicle B i Location For vehicle B with obstacles j Its rectangular boundary consists of four vertices (x, y, z). j1 ,y j1 ),(x j2 ,y j2 ),(x j3 ,y j3 ), (x j4 ,y j4 These four lines, forming the vehicle's four boundary lines, are used to determine whether the view is obstructed. Based on the vector cross product method, the intersection of line segments is determined as follows: Direction vector of line of sight segment L Defined as: in, For target vehicle B i Location information; x A y A To request the location information of vehicle A; Obstacle boundary segment for: For these two vectors, we determine whether they are parallel by their cross product. If they are not parallel, we further determine whether they intersect. Determine the endpoints of a line segment: For non-parallel line-of-sight segment L and obstacle vehicle B j For boundary line segments, determine whether their endpoints are located on different sides of the other line segment. Assume the obstacle vehicle B needs to be judged. j The boundary line segment starts from point B. j1 =(x j1 ,y j1 ) to point B j2 =(x j2 ,y j2 ): If the signs of D1 and D2 are different, it indicates that the obstacle vehicle is B. j The two endpoints of the boundary line segment are located on different sides of the line of sight segment, which means that the line of sight segment and obstacle vehicle B j If the boundary segments may intersect, then mark target vehicle B. i It is invisible.

8. The roadside unit credibility rating method based on data skeleton similarity as described in claim 7, characterized in that, The steps to determine the future visibility of surrounding target vehicles to the requesting vehicle are as follows: The future relative velocity v at a future time point t+Δt 相对 (t+Δt) passes through the current relative velocity v 相对 (t) and velocity trend Δv 趋势 (t) is used for prediction: v 相对 (t+Δt)=v 相对 (t)+Δv 趋势 (t)·Δt; For a certain car B, suppose its position at time t is (x B ,y B The velocity is v and the direction angle is θ. B At a future time t+Δt, the position (x′) of vehicle B B ,y′ B )for: (x′ B ,y′ B )=(x B +v 相对 (t+Δt)·cos(θ B )·Δt,y B +v 相对 (t+Δt)·sin(θ B )·Δt); Based on this prediction, the requesting vehicle A and the target vehicle B are identified. i and obstacle vehicle B j The location at a future point in time is then determined using the vector cross product method to assess visibility.

9. The roadside unit credibility rating method based on data skeleton similarity as described in claim 8, characterized in that, Comprehensive assessment of future visibility: Whether target vehicle B will be visible to requesting vehicle A at a future time depends on whether the following conditions are met: Current visibility: Target vehicle B is currently visible to requesting vehicle A; Future occlusion conditions are visible: at a future time point, the line of sight of target vehicle B and requesting vehicle A will not be obstructed; The directions are consistent or the differences are small: cos(Δθ)≈1, that is, the directions of the target vehicle B and the requesting vehicle A are highly consistent; The speed trend shows a close approximation of: v 相对 (t+Δt)<0, the speed trend of target vehicle B shows that it is gradually approaching the requesting vehicle A; When all conditions are met, it is considered that the target vehicle B is visible to the requesting vehicle A in the current moment and in the future, which means it passes the visibility screening. Iterate through the feature matrix, filter, and return the data: Iterate through all vehicles in the feature matrix and perform visibility filtering on them; For vehicles that pass the visibility screening, the roadside unit sends multiple sets of measurement data of the vehicle at different times to the requesting vehicle A for continuous calculation of credibility.

10. The roadside unit credibility rating method based on data skeleton similarity as described in claim 1, characterized in that, The method for calculating the reliability of roadside unit data using the on-board unit (OBU) is as follows: Data reception and matching: After receiving the data returned from the roadside unit, vehicle A identifies the corresponding vehicle from the vehicles it has detected based on the vehicle information; Data validity check: After identifying the corresponding vehicle, check whether the feature values ​​in the returned data are valid data, specifically: If a key feature is invalid data, the data for that target vehicle will be discarded. Data can be filled when non-critical features are null, or only at the moment of partial measurement. Based on the graph structure formed by the roadside unit for the requesting vehicle A and its surrounding vehicles that have passed the screening, and the graph structure generated by the requesting vehicle A and its surrounding vehicles with the same characteristics, the similarity score of edges and nodes is calculated through matrix iteration to evaluate the overall similarity. Specifically: Based on the measurement and screening of roadside units, an undirected complete graph G is generated for the requesting vehicle A and the surrounding vehicles that have passed the screening. A Based on the measurements of the requesting vehicle A and other vehicles in the same neighborhood selected by roadside units, map G is generated in the same manner. B ; The node similarity score S is calculated using matrix iteration. N S and edge similarity score E Finally, the overall similarity score S is calculated. G : S G =S N +S E Compare S G With τ G , τ G For a pre-set threshold, when S G >τ G When the confidence level is considered low, proceed to the next step of calculating the confidence level of the speed measurement value. Calculate the mean square error (MSE) of the speed measurements of the roadside unit and the requesting vehicle A relative to the speeds of other vehicles in the surrounding area. v : Among them, v R,t v is the measured velocity of the target vehicle by the roadside unit at time t. A,t To request the measurement value of vehicle A's speed relative to the target vehicle at time t, N is the total number of surrounding vehicles; Comparison of MSE v With τ v , τ v For a pre-set threshold, when MSE v <τ v When the confidence level is reached, the next step of calculating the confidence level of the orientation angle measurement value is initiated. Calculate the mean square error (MSE) of the heading angle measurements of the roadside unit and the requesting vehicle A relative to the same surrounding vehicles; Consider the periodicity of the orientation angle: Dth t =min(|θ R,t -θ A,t |,360°-|θ R,t -θ A,t |) Where, θ R,t Let θ be the direction angle measured by the roadside unit for the target vehicle at time t. A,t To request the direction angle Δθ that vehicle A measures relative to the target vehicle at time t. t The difference in the direction angle measurement between the roadside unit and the requesting vehicle A relative to the target vehicle at time t, ranging from [0°, 180°]; Mean Square Error of Direction Angle (MSE) θ for: Comparison of MSE θ With τ θ , τ θ For a pre-set threshold, when MSE θ <τ θ At that time, it was considered to have reached a high level of credibility.

11. The roadside unit credibility rating method based on data skeleton similarity as described in claim 10, characterized in that, According to the pre-set τ 位置 , τ v , τ θ , and the similarity S to the overall graph G and MSE v and MSE θ The calculation results are used to assign different confidence levels C. 等级 : Based on the trust level, a decision is made as to whether requesting vehicle A should use the data provided by the roadside unit for subsequent decision-making.

12. A roadside unit reliability rating system based on the method of any one of claims 1-11, characterized in that, It includes a roadside unit and an on-board unit (OBU). The on-board unit (OBU) is used to collect the current requesting vehicle's own location information, timestamp, and collection frequency information, and send them to the roadside unit. The roadside unit is used to acquire visibility data of surrounding target vehicles to the requesting vehicle and transmit it back to the on-board unit (OBU) of the requesting vehicle. The OBU then calculates the reliability of the roadside unit data.

Images