Vehicle conflict detection method, equipment, medium and product based on trajectory reconstruction

Abstract

The present disclosure relates to the field of artificial intelligence technology, and in particular provides a vehicle conflict detection method, device, medium and product based on trajectory reconstruction. The method can reduce the error in the original observation data by identifying abnormal acceleration in trajectory information and performing interpolation processing to complete the abnormal acceleration, thereby alleviating the problem of misjudgment in directly using the original observed trajectory information for conflict event detection.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to a vehicle conflict detection method, device, medium and product based on trajectory reconstruction. Background Art

[0002] Intersections have always been a focus of smart city traffic. The trajectories of different types of road users converge and intersect at intersections, making them one of the road space structures with the highest risk. Fortunately, the latest progress in intelligent transportation systems has significantly promoted the deep interaction between connected and automated vehicles (CAVs) and road intelligent facilities, showing great potential in intersection traffic safety control in the context of smart cities. Relying on the powerful computing power of technologies such as roadside units (RSUs) and cloud servers, cooperative vehicle infrastructure systems (CVIS) can integrate vehicle trajectories output by various sensors and publish the required traffic information to CAVs on the road in real time through vehicle-to-everything (V2X). Unlike single-vehicle intelligent technology that relies only on on-board sensors, CVIS is based on the tripartite cooperation of "vehicle-road-cloud". This approach can obtain a wider range of intersection environment information, and then achieve safer and smarter cooperative driving by minimizing the temporal and spatial overlap of vehicle trajectories.

[0003] Although CVIS has obvious advantages in intelligent traffic control, it also has certain limitations. Due to the influence of factors such as a large amount of data interaction, overlap of sensor detection ranges, and complexity of the external environment, there are inevitable errors in the raw observation data of various sensor fusions. Directly using noisy trajectories containing complex and non-stationary vibration signals to describe vehicle kinematics and various risk indicators may result in values ​​that do not match the actual situation, leading to misjudgment of the results. Summary of the invention

[0004] In view of the above problems, the present disclosure is proposed. The present disclosure provides a vehicle conflict detection method, device, medium and product based on trajectory reconstruction.

[0005] According to a first aspect of the present disclosure, a vehicle conflict detection method based on trajectory reconstruction is provided, comprising:

[0006] Acquiring the trajectory information of the vehicle within a currently set time range, the trajectory information including the acceleration and speed of the vehicle within the currently set time range;

[0007] Performing abnormal point detection on the acceleration within the currently set time range to obtain abnormal acceleration;

[0008] When confirming that the abnormal acceleration satisfies the interpolation condition, interpolation processing is performed on the abnormal acceleration to obtain an interpolation acceleration;

[0009] Using the interpolated acceleration to update the speed corresponding to the abnormal acceleration to obtain an updated speed;

[0010] Predicting predicted position information of the vehicle at a future set time based on the interpolated acceleration, the update speed, the acceleration and speed within the set time range except the abnormal acceleration and the speed corresponding to the abnormal acceleration, and the position information of the vehicle;

[0011] Based on the predicted position information and the contour information of the vehicle, collision conflict detection is performed on the vehicle.

[0012] In addition, according to the vehicle conflict detection method based on trajectory reconstruction described in the first aspect of the present disclosure, abnormal point detection is performed on the acceleration within the current set time range to obtain abnormal acceleration, including:

[0013] From the accelerations within the currently set time range, searching for a target acceleration that is greater than a preset maximum acceleration or less than a preset minimum acceleration;

[0014] The target acceleration is used as the abnormal acceleration.

[0015] In addition, according to the vehicle conflict detection method based on trajectory reconstruction described in the first aspect of the present disclosure, based on the interpolated acceleration, the update speed, the acceleration and speed other than the abnormal acceleration and the speed corresponding to the abnormal acceleration within the set time range, and the position information of the vehicle, predicting the predicted position information of the vehicle at a future set time includes:

[0016] Respectively extracting the high-frequency components of the acceleration other than the abnormal acceleration and the interpolation acceleration within the set time range to obtain the acceleration high-frequency component and the interpolation high-frequency component;

[0017] Predicted position information of the vehicle at a future set time is predicted based on the acceleration high-frequency component, the update speed, the speed within the set time range excluding the speed corresponding to the abnormal acceleration and the acceleration high-frequency component, and the vehicle position information.

[0018] In addition, according to the vehicle conflict detection method based on trajectory reconstruction described in the first aspect of the present disclosure, based on the predicted position information and the contour information of the vehicle, collision conflict detection is performed on the vehicle, including:

[0019] Acquire static profile information of the vehicle in a static state; predict the predicted heading angle of the vehicle within the set time using the heading angle of the vehicle within the current set time range;

[0020] Correcting the stationary profile information using the predicted heading angle to obtain driving profile information;

[0021] Based on the driving profile information and the predicted position information, collision conflict detection is performed on the vehicle.

[0022] In addition, according to the vehicle conflict detection method based on trajectory reconstruction described in the first aspect of the present disclosure, obtaining the static contour information of the vehicle in a static state includes:

[0023] Acquire the center point coordinates of the vehicle in the stationary state;

[0024] Based on the center point coordinates, calculating the coordinates of the four corner points of the vehicle;

[0025] A closed rectangle formed by connecting the coordinates of the four corner points is used as the static contour information of the vehicle.

[0026] In addition, according to the vehicle conflict detection method based on trajectory reconstruction described in the first aspect of the present disclosure, the predicted heading angle is used to correct the stationary profile information to obtain the driving profile information, including:

[0027] constructing a rotation matrix based on the predicted heading angle;

[0028] The driving profile information is obtained by multiplying the coordinates of each corner point included in the stationary profile information by the rotation matrix.

[0029] According to a first aspect of the present disclosure, a vehicle conflict detection device based on trajectory reconstruction is provided, comprising:

[0030] An acquisition module, used to acquire the trajectory information of the vehicle within a currently set time range, wherein the trajectory information includes the acceleration and speed of the vehicle within the currently set time range;

[0031] A first detection module is used to perform abnormal point detection on the acceleration within the current set time range to obtain abnormal acceleration;

[0032] An interpolation module, configured to perform interpolation processing on the abnormal acceleration to obtain an interpolated acceleration when confirming that the abnormal acceleration satisfies an interpolation condition;

[0033] An obtaining module, used to update the speed corresponding to the abnormal acceleration by using the interpolated acceleration to obtain an updated speed;

[0034] A prediction module, configured to predict predicted position information of the vehicle at a future set time based on the interpolated acceleration, the update speed, the acceleration and speed within the set time range except the abnormal acceleration and the speed corresponding to the abnormal acceleration, and the position information of the vehicle;

[0035] The second detection module is used to perform collision conflict detection on the vehicle based on the predicted position information and the contour information of the vehicle.

[0036] According to a third aspect of the present disclosure, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described in the first aspect.

[0037] According to a fourth aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program / instruction is stored, and when the computer program / instruction is executed by a processor, the steps of the method described in the first aspect are implemented.

[0038] According to a fifth aspect of the present disclosure, there is provided a computer program product, comprising a computer program / instruction, which implements the steps of the method described in the first aspect when executed by a processor.

[0039] As will be described in detail below, according to the vehicle conflict detection method based on trajectory reconstruction of the embodiment of the present disclosure, by identifying abnormal acceleration in the trajectory information and interpolating and completing the abnormal acceleration, the error existing in the original observation data can be reduced, thereby alleviating the problem of misjudgment that exists in directly using the original observed trajectory information for vehicle conflict detection based on trajectory reconstruction.

[0040] It is to be understood that both the foregoing general description and the following detailed description are exemplary, and are intended to provide further explanation of the technology as claimed. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The above and other purposes, features and advantages of the present disclosure will become more apparent by describing the embodiments of the present disclosure in more detail in conjunction with the accompanying drawings. The accompanying drawings are used to provide a further understanding of the embodiments of the present disclosure and constitute a part of the specification. Together with the embodiments of the present disclosure, they are used to explain the present disclosure and do not constitute a limitation of the present disclosure. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0042] Figure 1 is a flowchart illustrating application of a vehicle conflict detection method based on trajectory reconstruction according to an embodiment of the present disclosure.

[0043] Figure 2FIG. 4 is a schematic diagram illustrating a vehicle conflict in which outer contour information is considered according to an embodiment of the present disclosure.

[0044] Figure 3 is a schematic diagram illustrating a study area to which an embodiment of the present disclosure is applied.

[0045] FIG. 4( a ) is a schematic diagram illustrating an acceleration distribution of an original trajectory according to an embodiment of the present disclosure.

[0046] FIG. 4( b ) is a schematic diagram illustrating acceleration distribution of a reconstructed trajectory according to an embodiment of the present disclosure.

[0047] Figure 5 2 is a schematic diagram illustrating the distribution of the number of wavelet bases applied according to an embodiment of the present disclosure.

[0048] FIG. 6( a ) is a schematic diagram illustrating changes in an overall acceleration curve according to an embodiment of the present disclosure.

[0049] FIG6( b ) is a schematic diagram illustrating a detailed comparison of the two-step noise reduction effect according to an embodiment of the present disclosure.

[0050] FIG6( c ) is a schematic diagram illustrating the application of the cumulative distribution frequency according to an embodiment of the present disclosure.

[0051] FIG6(d) is a schematic diagram illustrating a spectrum diagram according to an embodiment of the present disclosure.

[0052] FIG. 7( a ) is a schematic diagram illustrating the change of the overall acceleration curve according to an embodiment of the present disclosure.

[0053] FIG. 7( b ) is a schematic diagram showing a detailed comparison of the two-step noise reduction effect according to an embodiment of the present disclosure.

[0054] FIG. 7( c ) is a schematic diagram illustrating the application of the cumulative distribution frequency according to an embodiment of the present disclosure.

[0055] FIG. 7( d ) is a schematic diagram illustrating a spectrum diagram applying an embodiment of the present disclosure.

[0056] Figure 8 is a schematic diagram illustrating a comparison of conflict detection results before and after trajectory reconstruction according to an embodiment of the present disclosure.

[0057] Fig. 9 is a schematic diagram illustrating a comparison of conflict detection results using a center point method and an outer contour method according to an embodiment of the present disclosure.

[0058] FIG. 10( a ) is a schematic diagram illustrating the application of an outer contour-based method according to an embodiment of the present disclosure.

[0059] FIG. 10( b ) is a schematic diagram illustrating the application of a center point based method according to an embodiment of the present disclosure.

[0060] FIG. 11( a ) is a schematic diagram illustrating a northbound vehicle to which an embodiment according to the present disclosure is applied.

[0061] FIG. 11( b ) is a schematic diagram illustrating a southbound vehicle to which an embodiment according to the present disclosure is applied.

[0062] FIG. 11( c ) is a schematic diagram illustrating a westbound vehicle to which an embodiment of the present disclosure is applied.

[0063] FIG. 11( d ) is a schematic diagram illustrating an eastbound vehicle to which an embodiment of the present disclosure is applied.

[0064] Fig.12 is a structural diagram illustrating the application of a vehicle conflict detection device based on trajectory reconstruction according to an embodiment of the present disclosure.

[0065] Fig.13 is a hardware block diagram illustrating an electronic device according to an embodiment of the present disclosure.

[0066] Fig.14 is a schematic diagram illustrating a computer-readable storage medium according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0067] In order to make the purpose, technical scheme and advantages of the embodiments of the present disclosure clearer, the technical scheme in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all of the embodiments. The components of the embodiments of the present disclosure generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present disclosure provided in the drawings is not intended to limit the scope of the present disclosure for protection, but merely represents the selected embodiments of the present disclosure. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present disclosure.

[0068] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0069] The term "and / or" herein only describes an association relationship, indicating that three relationships may exist. For example, A and / or B may represent the following three situations: A exists alone, A and B exist at the same time, and B exists alone. In addition, the term "at least one" herein represents any combination of at least two of any one or more of a plurality of. For example, including at least one of A, B, and C may represent including any one or more elements selected from the set consisting of A, B, and C.

[0070] Among the related technologies, CVIS based on the tripartite cooperation of "vehicle-road-cloud" has obvious advantages in intelligent traffic control, but it also has certain limitations. Due to the influence of factors such as a large amount of data interaction, overlap of sensor detection ranges, and complexity of the external environment, there are inevitable errors in the original observation data fused by various sensors. Directly using noisy trajectories containing complex and non-stationary vibration signals to describe vehicle kinematics and various risk indicators may result in values ​​that do not match the actual situation, leading to misjudgment of the results.

[0071] In order to alleviate the technical problems existing in the related art, the embodiments of the present disclosure provide a vehicle conflict detection method, device, medium and product based on trajectory reconstruction. The method can reduce the error existing in the original observation data by identifying the abnormal acceleration in the trajectory information and performing interpolation processing to complete the abnormal acceleration, thereby alleviating the problem of misjudgment existing in vehicle conflict detection based on trajectory reconstruction directly using the original observed trajectory information.

[0072] To facilitate understanding of this embodiment, a vehicle conflict detection method based on trajectory reconstruction disclosed in the embodiment of the present disclosure is first introduced in detail. The execution subject of the vehicle conflict detection method based on trajectory reconstruction provided in the embodiment of the present disclosure is generally an electronic device with certain computing capabilities, and the electronic device includes, for example: a terminal device or a server or other processing device. In some possible implementations, the vehicle conflict detection method based on trajectory reconstruction can be implemented by a processor calling a computer-readable instruction stored in a memory.

[0073] See also Figure 1 FIG. 1 is a flow chart of a vehicle conflict detection method based on trajectory reconstruction provided by an embodiment of the present disclosure, wherein the method comprises the following steps:

[0074] Step 101: Acquire the trajectory information of the vehicle within the current set time range, where the trajectory information includes the acceleration and speed of the vehicle within the current set time range.

[0075] In this embodiment, the currently set time range includes the current time and the set historical time before the current time.

[0076] It should be understood that the trajectory information within the current set time range includes the acceleration and speed at each moment within the current set time range. It should also be understood that, according to actual needs, the trajectory information may include other information within the set time range, such as heading angle, in addition to the acceleration and speed within the set time range.

[0077] Step 102: Perform abnormal point detection on the acceleration within the currently set time range to obtain abnormal acceleration.

[0078] In one or more embodiments, step 102 may include the following steps:

[0079] From the accelerations within the currently set time range, searching for a target acceleration that is greater than a preset maximum acceleration or less than a preset minimum acceleration;

[0080] The target acceleration is used as the abnormal acceleration.

[0081] In the application, after obtaining the original trajectory information of the vehicle, the trajectory information needs to be reconstructed first. The performance of the reconstruction algorithm is usually highly dependent on the stability of the input data, and the input outliers may seriously interfere with the final result of the trajectory reconstruction. Before performing trajectory denoising, the outliers need to be deleted and interpolated. To this end, this embodiment designs a rule-based outlier detector for preliminary screening. The principle of the outlier detector can be found in the following formula (1):

[0082]

[0083] in, The acceleration within the current set time range for the trajectory information. The preset minimum and maximum accelerations.

[0084] The selection of the limit value range is crucial. A range that is too loose will weaken the outlier detection effect, while a range that is too strict will cause some normal trajectories to be misjudged as outliers. In general, the change in vehicle speed should be within the acceptable range of vehicle performance. This range is considered to be [a min = -8, a max = 5]m / s². Points outside this range are marked as abnormal points and need to be deleted and interpolated.

[0085] Step 103: When it is confirmed that the abnormal acceleration meets the interpolation condition, interpolation processing is performed on the abnormal acceleration to obtain the interpolated acceleration.

[0086] In this embodiment, the abnormal acceleration is interpolated using a cubic spline function, but is not limited to the cubic spline function. The cubic spline function can provide accurate estimates between known data points while maintaining data smoothness. When the cubic spline function interpolates the abnormal acceleration, it is necessary to refer to a reference point sequence including a set number of non-abnormal points. Therefore, before the abnormal acceleration is interpolated, if there are a set number of non-abnormal point accelerations before the acceleration corresponding to the abnormal acceleration, it is considered that the abnormal acceleration meets the interpolation condition, and the abnormal acceleration can be interpolated based on the cubic spline function. It should be understood that when the abnormal acceleration does not meet the interpolation condition, the abnormal acceleration is directly deleted from the trajectory information.

[0087] Assume that the outlier point is ( t i , a i ), then the interpolation reference point sequence is ( t i-10 , a i-10 ), …, ( t i-1 , a i-1 ),in t i Indicates i Time nodes, a i is t i The acceleration value at . For each adjacent time point subinterval [ t j , t j+1 ]( j = i -10, …, i -1), the cubic spline function can be defined as follows:

[0088]

[0089] in, w j is the substituted spline coefficient. In order to ensure the validity and smoothness of the interpolation result, the spline function of each subinterval must meet the following continuity conditions:

[0090]

[0091] in, and Represent the first and second order derivatives of the function respectively. Substituting the above constraints into the cubic spline function, the spline coefficients can be solved w j, we get a more realistic estimate of the abnormal acceleration value. This estimate is limited to [-8,5] m / s².

[0092] Step 104: Use the interpolated acceleration to update the speed corresponding to the abnormal acceleration to obtain an updated speed.

[0093] Step 105: predict the predicted position information of the vehicle at a future set time based on the interpolated acceleration, the update speed, the acceleration and speed within a set time range except the abnormal acceleration and the speed corresponding to the abnormal acceleration, and the position information of the vehicle.

[0094] In some embodiments, in the process of predicting the predicted position information of the vehicle at a set time in the future, noise filtering can be performed on the interpolated acceleration and the acceleration within the set time range except for abnormal acceleration. In specific implementation, step 105 may include the following steps:

[0095] The high-frequency components of the acceleration and the interpolation acceleration except the abnormal acceleration within the set time range are extracted respectively to obtain the acceleration high-frequency component and the interpolation high-frequency component;

[0096] Based on the high-frequency component of acceleration, the update speed, the speed and high-frequency component of acceleration except the speed corresponding to the abnormal acceleration within a set time range, and the position information of the vehicle, the predicted position information of the vehicle at a set time in the future is predicted.

[0097] After eliminating the interference of abnormal acceleration, the noise in the trajectory can be removed, thereby restoring the internal consistency of the trajectory data. Compared with Fourier transform, wavelet transform can provide information in both time domain and frequency domain, and has good time-frequency localization ability. In practical applications, discrete wavelets can effectively solve the problem of low computational efficiency of continuous wavelets and show better robustness. First, the discrete wavelet transform can be used to decompose the signal at time t (i.e., vehicle trajectory) into components in different frequency bands:

[0098]

[0099] in, J is the number of discrete wavelet decomposition layers, It is a scale j The approximate component below represents the low-frequency component of the signal and reflects the overall trend of the signal; It is a scale j The detail components under represent the high-frequency components of the signal, reflecting the details and instantaneous changes of the signal. They can be calculated by the following formula:

[0100]

[0101]

[0102] in, c j,k is in scale j and pan k The approximate coefficients of d j,k is in scale j and pan k The detail coefficient is φ is the scaling function, ψ is a wavelet function. c j,k and d j,k It can be obtained by integrating the signal and the wavelet basis function:

[0103]

[0104]

[0105] After clarifying the discrete wavelet decomposition process, the following steps are required to achieve signal reconstruction:

[0106] For the noise reduction effect of wavelet transform, it is very important to choose the appropriate wavelet basis function. Different wavelet basis functions differ in time-frequency localization ability, smoothness and symmetry. Due to differences in the driving purpose and driving behavior of the driver, the trajectory of each vehicle may present a different signal structure. This embodiment matches an optimal wavelet basis for the signal of each vehicle by calculating the maximum energy of the signal and the Shannon entropy ratio. This criterion balances energy concentration and information entropy, ensuring that the selected wavelet basis minimizes redundancy while maximizing signal representation. Scale j The energy and Shannon entropy ratio under can be calculated by the following formula:

[0107]

[0108] The numerator is the scale j The energy of the wavelet coefficients, the denominator is the scale j The Shannon entropy of the wavelet coefficient is calculated as follows:

[0109]

[0110]

[0111] in, C ( j , i ) is the scale j The next i wavelet coefficients.

[0112] Noise usually exists in the form of high-frequency components, mainly concentrated in the detail components of the wavelet transform. We can calculate the threshold based on nonlinear multi-resolution analysis and modify the wavelet coefficients of the detail components. Commonly used shrinkage (threshold) functions include hard threshold and soft threshold:

[0113]

[0114] for λ VisuShrink is a concise and general method for calculating the global threshold value:

[0115]

[0116] in, N is the length of the signal (i.e. the number of data points); σ is the variance of the noise, which can be estimated by the median of the high-frequency subband coefficients after wavelet transformation:

[0117]

[0118] As can be seen from Eq.10, the hard threshold directly sets the coefficients below the threshold to 0, thereby obtaining a smoothing effect on the signal, but it is easy to produce discontinuous points in the signal. The soft threshold smoothes the signal by attenuating the coefficient, and the denoising effect is better, but it will cause a certain signal loss. In contrast, the semi-soft threshold achieves a better balance between denoising and signal preservation, and can smooth the coefficients close to the threshold, reducing artifacts and offsets:

[0119]

[0120] in, is a scaling factor with a value range of [0, 1]. = 0.5 to obtain a compromise between the hard and soft thresholds. This choice retains the strong signal fidelity of the hard threshold and combines the good noise suppression effect of the soft threshold.

[0121] After modifying the wavelet coefficients of the detail components, the original trajectory signal can be reconstructed. This process effectively removes noise without changing the basic structure and main features of the signal by processing the detail components and retaining the approximate components. The reconstructed signal can be expressed as:

[0122]

[0123] Step 106: Perform collision detection on the vehicle based on the predicted position information and the contour information of the vehicle.

[0124] In some embodiments, step 106 may include the following steps:

[0125] Obtain the static profile information of the vehicle in a static state; use the heading angle of the vehicle within the current set time range to predict the predicted heading angle of the vehicle within the set time;

[0126] The static profile information is corrected by using the predicted heading angle to obtain the driving profile information;

[0127] Based on the driving profile information and the predicted position information, the vehicle collision conflict detection is performed.

[0128] The static contour information of the vehicle in a static state can be obtained based on the coordinates of the center point of the vehicle in a static state. When implemented specifically, the following steps may be included:

[0129] Obtain the static profile information of the vehicle in a static state; use the heading angle of the vehicle within the current set time range to predict the predicted heading angle of the vehicle within the set time;

[0130] The static profile information is corrected by using the predicted heading angle to obtain the driving profile information;

[0131] Based on the driving profile information and the predicted position information, the vehicle collision conflict detection is performed.

[0132] The outer contour bounding box of the vehicle can be described by geometric theory. First, define the coordinates of the four corner points of the vehicle relative to the coordinates of the center point ( x 0, y o ) are:

[0133]

[0134]

[0135]

[0136]

[0137] in, L l represents the length of the vehicle, L w is the width of the vehicle. Respectively represent the lower left corner, lower right corner, upper right corner, and upper left corner of the vehicle ( i =1, 2, 3, 4) are the coordinates after translation relative to the center point.

[0138] Sequentially The corner point coordinates are connected to form a closed rectangle that includes the size information of the vehicle to a certain extent. However, this simple bounding box still lacks a description of the vehicle's heading angle. To this end, a rotation matrix is ​​constructed to transform the bounding box:

[0139]

[0140]

[0141] Among them, θ is the clockwise angle between the vehicle head direction and the true north direction; θ rad The vehicle heading is expressed in radians. Multiply them by the rotation matrix R, and then add the center point coordinates ( x 0, y o ), you can get the global bbox coordinates of the vehicle's outer contour level:

[0142]

[0143]

[0144] Substitution After that, formula (18) can be converted to:

[0145]

[0146]

[0147]

[0148]

[0149] So far, by The closed rectangular frame formed by the package perfectly fits the outer contour of the vehicle.

[0150] In this embodiment, in order to ensure the detection accuracy, the vehicle's motion state needs to be accurately described first. Compared with some data-driven models that rely on complex machine learning algorithms, physics-driven models can significantly reduce computational complexity. In a short period of time, the vehicle's motion can be approximately regarded as a straight line motion under constant acceleration. t In general, the vehicle's motion information can be modeled as follows:

[0151]

[0152] Among them, X = [ x 0, y 0, v x =v ∗cos θ , v y = v ∗sin θ ]' is the vehicle state vector; U=[ a x = a ∗cos θ , a y = a ∗sin θ ]' is the vehicle control vector; v is the instantaneous speed of the vehicle; F is the state transfer matrix; B is the control input matrix. t The state transition equation at time can be written as:

[0153]

[0154] in, Δt is the unit interval length, set to 0.1s.

[0155] After completing the above preparations, the vehicle conflict detection algorithm can be described as a time series modeling problem. This process is shown in Figure 2 In. By time t For example, the steps are briefly summarized as follows:

[0156] (1) Obtain various kinematic indicators of the ego vehicle and its surroundings from the trajectory reconstruction module.

[0157] (2) Calculate the vehicle’s state vector and outer contour bounding box information at time t+n×Δt.

[0158] (3) Determine whether the bounding boxes overlap. If so, it means that a conflict event has occurred at this moment, and a warning message is sent to the CAV through the RSU.

[0159] (4) Repeat the above steps until the detection time reaches the maximum threshold (i.e. T conf ), start calculating the conflict event at the next time t+1.

[0160] In the solution provided in this embodiment, by identifying abnormal acceleration in trajectory information and performing interpolation processing to complete the abnormal acceleration, the error in the original observation data can be reduced, thereby alleviating the problem of misjudgment in vehicle conflict detection based on trajectory reconstruction directly using the original observed trajectory information.

[0161] In addition, this embodiment also performs noise reduction processing on the trajectory by discrete wavelet transform. By calculating the ratio of energy and Shannon entropy, the optimal wavelet basis is automatically assigned to each trajectory to solve the signal difference of different trajectories.

[0162] Finally, a rotation matrix is ​​constructed based on the center point of the vehicle trajectory to reconstruct the outer contour bounding box of the vehicle, accurately reflecting the actual position and posture of the vehicle. Through a concise and efficient physical model to describe the state of the vehicle, a vehicle collision detection algorithm based on trajectory reconstruction for multi-dimensional scenarios is designed to achieve real-time warning of risk events.

[0163] Taking a vehicle trajectory reconstruction, risk assessment and analysis study as an example, the specific implementation steps are as follows:

[0164] 1. Research area and data

[0165] The intersection selected as the research case in this embodiment has various types of areas near bus stops, subway stations, large shopping malls, office buildings, etc., which attract a large number of traffic participants to compete for road resources, with complex and changeable trajectories and frequent conflicts. In order to realize the visualization function of the risk map, this embodiment draws a high-precision map of the intersection based on ArcGIS software at a 1:1 ratio. Figure 3 It can be seen that the north and south entrance roads are composed of 4 straight lanes, a left turn lane and a right turn lane. The east and west entrance roads include one straight lane, one left turn lane and one right turn lane. This embodiment further limits the research area to the inner area of ​​the intersection surrounded by the stop lines of each entrance road. This is done in consideration of the fact that there are no lane markings and lane rules inside the intersection, which will bring greater trajectory complexity and conflict severity, making the research more challenging. It should be noted that since the right turn lane is separately divided into the auxiliary road section separated by the green belt, it is not considered in this embodiment.

[0166] The intersection demonstrated in this embodiment supports the large-scale operation of L4 and above CAVs, and is backward compatible with the test operation and vehicle networking application scenarios of low-level CAVs. Relying on the integrated advanced equipment in the demonstration area, the data collection method of this embodiment is of significant technological advancement. Through the RSU integrated camera, lidar and millisecond wave radar sensor, various dynamic information of the vehicle is obtained in real time. This information is fused and analyzed by Baidu's edge computing node to generate a high-precision continuous tracking vehicle trajectory. Subsequently, the RSU uploads the vehicle trajectory to the cloud server at a fixed frequency of 10Hz. This embodiment directly extracts relevant data through the cloud interface for research. This vehicle trajectory collection and processing process not only improves the data quality, but also provides a guarantee for the large-scale deployment and expansion of the method of this embodiment in the future, and can meet the diverse needs of modern smart city transportation systems.

[0167] In order to ensure a large sample size, data collection was carried out during the evening rush hour (18:00 to 19:00) on weekdays, and a total of 2,825 vehicle trajectories were collected for subsequent research. The extracted vehicle-related parameters include timestamp, unique identifier, location, speed, heading angle, vehicle length, vehicle width, etc.

[0168] 2. Performance Analysis of the Proposed Trajectory Reconstruction Method

[0169] In order to verify the performance of the proposed two-step reconstruction method at each stage, this embodiment counted the acceleration distribution before and after reconstruction, and the results are shown in Figure 4(a)-Figure 4(b). In the original trajectory, the acceleration distribution range is [-580 m / s², 377 m / s²], which obviously does not meet the actual physical conditions. After the two-step trajectory reconstruction, the acceleration range becomes [-7.62m / s², 4.77 m / s²], which is reduced by 98.71%. This range is reasonable and acceptable for vehicle performance.

[0170] Based on the energy and Shannon entropy ratio of each vehicle trajectory, the number of automatically matched wavelet bases is distributed as follows: Figure 5 As shown. It can be clearly seen from the figure that bior1.3, haar and coif5 wavelets are used most frequently, indicating that they have significant advantages in processing vehicle trajectories inside intersections. There are two main possible reasons for this phenomenon: the maneuvers of vehicles inside intersections usually contain relatively obvious instantaneous fluctuations or mutations, such as sudden braking, acceleration, turning, etc. The compact time domain support and localization characteristics of bior1.3 wavelet and Haar wavelet make them particularly suitable for capturing rapid changes in vehicle trajectories. In addition to rapid changes, there are also stable driving stages in intersections, such as slow acceleration and deceleration or uniform motion. These changes belong to lower-frequency dynamics. As a high-order orthogonal wavelet, coif5 can better handle this type of low-frequency dynamics while ensuring a smooth transition of trajectory signals.

[0171] To further verify the potential of wavelet transform in vehicle trajectory reconstruction, the signal-to-noise ratio (SNR) and root mean square error (RMSE) are used to evaluate the denoising performance and the consistency of the trajectories before and after reconstruction. A higher SNR indicates a good denoising effect, while a lower RMSE indicates that the trajectories before and after reconstruction are more similar.

[0172]

[0173]

[0174] in, represents the original signal, The method proposed in this embodiment is compared with the Kalman filter and Savitzky-Golay filter designed by previous researchers, and the results are shown in Table 1.

[0175] Table 1. Comparison of different trajectory reconstruction methods

[0176]

[0177] The mean SNR of the proposed method is 12.43, which is significantly better than the two baseline methods of 6.04 and 3.73. This shows that the proposed method can more effectively enhance the signal quality during the denoising process and more accurately recover the true trajectory from the noise. In addition, the mean RMSE of the proposed method is 0.67, which is acceptable for the interior of the intersection with serious stop-and-go phenomenon. This shows that the proposed method can effectively reduce the trajectory noise while also maintaining the consistency requirement of trajectory reconstruction. The lower standard deviation further shows that the proposed method exhibits stable fluctuations during data reconstruction. In contrast, the SG filter exhibits the lowest SNR and the highest RMSE, indicating that the method may have the phenomenon of over-denoising, which is consistent with the description in the relevant literature. For different thresholds, the semi-soft threshold shows the best effect.

[0178] In order to demonstrate the role of each step in the trajectory reconstruction method proposed in this embodiment, a straight-moving vehicle and a left-turning vehicle are used as examples for explanation. As can be seen from Figure 6 (ab), the trajectory of the left-turning vehicle has a typical stop-and-go phenomenon, and the acceleration fluctuates greatly, which is challenging for the trajectory reconstruction task. The acceleration range of the original trajectory of the vehicle is [-38.39, 65.87], which is obviously an unrealistic range. After setp1: outlier deletion, this range is reduced to [-7.82, 4.96] respectively. After further step2 wavelet transform denoising, the acceleration distribution is within [-3.82, 2.05], and the range is reduced by 94.4%. Figure 6 (c) shows the cumulative percentage of acceleration values. The cumulative distribution of the original acceleration data is wide, and there are extreme discrete points. After two steps of reconstruction, the acceleration values ​​become more concentrated to varying degrees, and the trajectory stability is significantly enhanced. Figure 6 (d) shows that the original trajectory has a violent fluctuation feature in the high frequency band, and the maximum amplitude is close to 2.2. The amplitude fluctuation of the reconstructed trajectory is significantly reduced, and the low-frequency components are retained to a certain extent.

[0179] Similarly, as shown in Figure 7(ab), the acceleration range of the original trajectory of the straight vehicle is [-17.55, 18.15], which is reduced to [-4.48, 4.33] after step 1, and further reduced to [-2.21, 3.55] after step 2, with a range reduction of 83.9%. As can be seen from Figure 7(c), the reconstructed acceleration cumulative distribution has a steep rise near 0, indicating that most values ​​are close to 0, showing stable fluctuations. Figure 7(d) further shows that the method proposed in this embodiment retains the basic structure of the original data in the low-frequency band while eliminating high-frequency noise.

[0180] 3. Evaluation of the effectiveness of real-time conflict detection algorithm

[0181] For the setting of the conflict detection threshold, existing research suggests that the value range is between [1, 4]s. T conf =1.5s, in order to show and analyze the more serious part of the conflict event. Although the RMSE of the velocity after trajectory reconstruction is only 0.13 m / s, there is a significant difference after applying it to the conflict event monitoring algorithm. Figure 8 It shows that the total number of conflict trajectory points detected after reconstruction is reduced by 20,514, and most of these reduced trajectory points are concentrated in areas with higher conflict thresholds. This shows that abnormal points and noise points in the original trajectory may lead to a large number of false detections of conflict events that do not exist, and the error will be amplified as the detection time increases. After further comparing the composition structure of the conflict events detected by the original trajectory and the reconstructed trajectory (including the unique identifier of the self-vehicle, the unique identifier of the surrounding vehicles, the time of conflict occurrence and the conflict value), it was found that only 62.58% of the data overlapped. Therefore, there are large differences in the detected conflict events between the original trajectory and the reconstructed trajectory. Directly using the original observed trajectory to provide risk warnings for CAVs may cause the automated driving assistance system or the driver to make wrong judgments, which in turn increases the possibility of accidents.

[0182] Fig. 9 The necessity of vehicle outer contour size information for vehicle conflict detection based on trajectory reconstruction is illustrated. It can be clearly seen that the vehicle center point-based method shows low sensitivity in the vehicle conflict detection task based on trajectory reconstruction, and the number of conflicts detected by the center point-based method accounts for only 44.07% of the outer contour-based method. This is because the vehicle particle form adopted by this method ignores the size differences between different types of traffic participants and cannot accurately reflect the relative positions between vehicles and their real competition relationship for road space resources.

[0183] Figures 10 (ab) and 11 (ad) demonstrate the potential of using the method proposed in this embodiment for risk map visualization. This is achieved based on the coordinate points of the vehicle conflict event and the kernel density estimation method. It can be clearly seen from Figures 10 (ab) that in some areas, the conflict hotspot density extracted based on the center point is significantly lower than that based on the outer contour. This may convey incorrect risk information to the autonomous driving system and the driver.

[0184] Figure 11(ad) shows the risk distribution pattern for different driving directions. It is worth noting that the risk is greater in the first half of the journey after the vehicle enters the intersection, while the risk hotspots are relatively sparse in the second half. There are two reasons for this phenomenon: (1) When the vehicle starts from the stop line, the initial acceleration fluctuates greatly, which is more risky than the smooth driving in the second half. (2) Due to the lack of lane line guidance at the intersection, the driver's perception of direction is affected when he first enters the area, which may cause abnormal trajectory changes. As time goes by, the driver begins to gradually adapt to this environment and eventually drives out of the intersection smoothly.

[0185] Finally, the real-time performance of the method proposed in this embodiment is verified. Generally speaking, a complex traffic scene is considered to be composed of a self-vehicle and 6 surrounding vehicles, including the front and rear vehicles of the self-vehicle, and the front and rear vehicles of the adjacent two-side traffic. The conflict detection thresholds are set to 3s (representing general conflicts) and 1.5s (representing severe conflicts), respectively, and the results are shown in Table 2. For general conflicts, considering 6 surrounding vehicles, it only takes 7.04 milliseconds to complete the detection of one frame of data. In the severe conflict detection scenario, this time is further shortened to 3.56 milliseconds, and the computational complexity is extremely low. For CVIS equipped with RSU and cloud servers, it is fully capable of easily meeting the real-time requirements of risk warning.

[0186] Table 2. Algorithm computational complexity verification

[0187]

[0188] The embodiment of the present disclosure also provides a vehicle conflict detection device based on trajectory reconstruction, which is used to execute the vehicle conflict detection method based on trajectory reconstruction provided by any of the above embodiments. Fig.12 As shown, the device comprises:

[0189] An acquisition module 121 is used to acquire the trajectory information of the vehicle within a currently set time range, wherein the trajectory information includes the acceleration and speed of the vehicle within the currently set time range;

[0190] A first detection module 122 is used to perform abnormal point detection on the acceleration within the current set time range to obtain abnormal acceleration;

[0191] An interpolation module 123, configured to perform interpolation processing on the abnormal acceleration to obtain an interpolated acceleration when confirming that the abnormal acceleration satisfies an interpolation condition;

[0192] An obtaining module 124 is used to update the speed corresponding to the abnormal acceleration using the interpolated acceleration to obtain an updated speed;

[0193] A prediction module 125, configured to predict predicted position information of the vehicle at a future set time based on the interpolated acceleration, the update speed, the acceleration and speed within the set time range except the abnormal acceleration and the speed corresponding to the abnormal acceleration, and the position information of the vehicle;

[0194] The second detection module 126 is used to perform collision detection on the vehicle based on the predicted position information and the contour information of the vehicle.

[0195] In some optional embodiments, the first detection module 122 is used to:

[0196] From the accelerations within the currently set time range, searching for a target acceleration that is greater than a preset maximum acceleration or less than a preset minimum acceleration;

[0197] The target acceleration is used as the abnormal acceleration.

[0198] In some optional embodiments, the prediction module 125 is used to:

[0199] Respectively extracting the high-frequency components of the acceleration other than the abnormal acceleration and the interpolation acceleration within the set time range to obtain the acceleration high-frequency component and the interpolation high-frequency component;

[0200] Predicted position information of the vehicle at a future set time is predicted based on the acceleration high-frequency component, the update speed, the speed within the set time range excluding the speed corresponding to the abnormal acceleration and the acceleration high-frequency component, and the vehicle position information.

[0201] In some optional embodiments, the second detection module 126 is used to:

[0202] Acquire static profile information of the vehicle in a static state; predict the predicted heading angle of the vehicle within the set time using the heading angle of the vehicle within the current set time range;

[0203] Correcting the stationary profile information using the predicted heading angle to obtain driving profile information;

[0204] Based on the driving profile information and the predicted position information, collision conflict detection is performed on the vehicle.

[0205] In some optional embodiments, the second detection module 126 is used to:

[0206] Acquire the center point coordinates of the vehicle in the stationary state;

[0207] Based on the center point coordinates, calculating the coordinates of the four corner points of the vehicle;

[0208] A closed rectangle formed by connecting the coordinates of the four corner points is used as the static contour information of the vehicle.

[0209] In some optional embodiments, the second detection module 126 is used to:

[0210] constructing a rotation matrix based on the predicted heading angle;

[0211] The driving profile information is obtained by multiplying the coordinates of each corner point included in the stationary profile information by the rotation matrix.

[0212] The vehicle conflict detection device based on trajectory reconstruction provided by the embodiment of the present disclosure and the vehicle conflict detection method based on trajectory reconstruction provided by the embodiment of the present disclosure are based on the same inventive concept and have the same beneficial effects as the methods adopted, operated or implemented by them.

[0213] The disclosed embodiment also provides an electronic device to execute the above-mentioned vehicle conflict detection method based on trajectory reconstruction. Fig.13 It shows a schematic diagram of an electronic device provided by some embodiments of the present disclosure. Fig.13 As shown, the electronic device 8 includes: a processor 800, a memory 801, a bus 802 and a communication interface 803, wherein the processor 800, the communication interface 803 and the memory 801 are connected via the bus 802; the memory 801 stores a computer program that can be run on the processor 800, and when the processor 800 runs the computer program, the vehicle conflict detection method based on trajectory reconstruction provided by any of the aforementioned embodiments of the present disclosure is executed.

[0214] The memory 801 may include a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk storage. The communication connection between the device network element and at least one other network element is realized through at least one communication interface 803 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used.

[0215] The bus 802 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The memory 801 is used to store a program, and the processor 800 executes the program after receiving an execution instruction. The vehicle conflict detection method based on trajectory reconstruction disclosed in any implementation of the aforementioned embodiment of the present disclosure may be applied to the processor 800, or implemented by the processor 800.

[0216] The processor 800 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit in the processor 800 or the instruction in the form of software. The above processor 800 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The disclosed methods, steps and logic block diagrams in the embodiments of the present disclosure can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in conjunction with the embodiments of the present disclosure can be directly embodied as a hardware decoding processor to be executed, or the hardware and software modules in the decoding processor can be executed. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 801, and the processor 800 reads the information in the memory 801 and completes the steps of the above method in combination with its hardware.

[0217] The electronic device provided by the embodiment of the present disclosure and the vehicle conflict detection method based on trajectory reconstruction provided by the embodiment of the present disclosure are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by them.

[0218] The present disclosure also provides a computer-readable storage medium corresponding to the vehicle conflict detection method based on trajectory reconstruction provided in the above-mentioned embodiment. Fig.14 The computer-readable storage medium shown is a CD 30 on which a computer program (ie, a program product) is stored. When the computer program is run by a processor, the vehicle conflict detection method based on trajectory reconstruction provided in any of the aforementioned embodiments is executed.

[0219] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical or magnetic storage media, which are not listed here one by one.

[0220] The computer-readable storage medium provided by the above-mentioned embodiments of the present disclosure and the vehicle conflict detection method based on trajectory reconstruction provided by the embodiments of the present disclosure are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the application programs stored therein.

[0221] It should be noted that:

[0222] In the description provided herein, a large number of specific details are described. However, it is understood that the embodiments of the present disclosure can be practiced without these specific details. In some instances, well-known structures and technologies are not shown in detail so as not to obscure the understanding of this description.

[0223] Similarly, it should be understood that in order to streamline the present disclosure and aid in understanding one or more of the various inventive aspects, in the above description of the exemplary embodiments of the present disclosure, the various features of the present disclosure are sometimes grouped together into a single embodiment, figure, or description thereof. However, the disclosed method should not be interpreted as reflecting the following schematic diagram: the claimed disclosure requires more features than the features explicitly recited in each claim. More specifically, as reflected in the claims below, the inventive aspects lie in less than all the features of the single embodiment disclosed above. Therefore, the claims that follow the specific embodiment are hereby expressly incorporated into the specific embodiment, with each claim itself serving as a separate embodiment of the present disclosure.

[0224] In addition, those skilled in the art will appreciate that, although some embodiments described herein include certain features included in other embodiments but not other features, the combination of features of different embodiments is meant to be within the scope of this disclosure and form different embodiments. For example, in the claims below, any one of the claimed embodiments may be used in any combination.

[0225] The above is only a preferred specific implementation of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with the technical field within the technical scope disclosed in the present disclosure should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A vehicle conflict detection method based on trajectory reconstruction, characterized in that: include: Acquire the trajectory information of the vehicle in the intersection area within the currently set time range, wherein the trajectory information includes the acceleration and speed of the vehicle within the currently set time range; Performing abnormal point detection on the acceleration within the currently set time range to obtain abnormal acceleration; When confirming that the abnormal acceleration satisfies the interpolation condition, interpolation processing is performed on the abnormal acceleration to obtain an interpolation acceleration; Using the interpolated acceleration to update the speed corresponding to the abnormal acceleration to obtain an updated speed; Predicting predicted position information of the vehicle at a future set time based on the interpolated acceleration, the update speed, the acceleration and speed within the set time range except the abnormal acceleration and the speed corresponding to the abnormal acceleration, and the position information of the vehicle; Based on the predicted position information and the contour information of the vehicle, performing collision conflict detection on the vehicle; If there are a set number of non-abnormal acceleration points before the acceleration corresponding to the abnormal acceleration, the abnormal acceleration is considered to meet the interpolation condition; Predicting predicted position information of the vehicle at a future set time based on the interpolated acceleration, the update speed, the acceleration and speed within the set time range except the abnormal acceleration and the speed corresponding to the abnormal acceleration, and the position information of the vehicle, including: Using an optimal wavelet basis that matches the trajectory signal structure of the vehicle, respectively extracting the high-frequency components of the acceleration other than the abnormal acceleration and the interpolated acceleration within the set time range, to obtain the acceleration high-frequency component and the interpolated acceleration high-frequency component; Based on the interpolated acceleration high-frequency component, the update speed, the speed within the set time range except the speed corresponding to the abnormal acceleration and the acceleration high-frequency component, and the position information of the vehicle, the predicted position information of the vehicle at a future set time is predicted.

2. The method according to claim 1, characterized in that Performing abnormal point detection on the acceleration within the currently set time range to obtain abnormal acceleration includes: From the accelerations within the currently set time range, searching for a target acceleration that is greater than a preset maximum acceleration or less than a preset minimum acceleration; The target acceleration is used as the abnormal acceleration.

3. The method according to claim 1, characterized in that Based on the predicted position information and the contour information of the vehicle, collision detection is performed on the vehicle, including: Acquire static profile information of the vehicle in a static state; predict the predicted heading angle of the vehicle within the set time using the heading angle of the vehicle within the current set time range; Correcting the stationary profile information using the predicted heading angle to obtain driving profile information; Based on the driving profile information and the predicted position information, collision conflict detection is performed on the vehicle.

4. The method according to claim 3, characterized in that Acquiring static profile information of the vehicle in a static state, including: Acquire the center point coordinates of the vehicle in the stationary state; Based on the center point coordinates, calculating the coordinates of the four corner points of the vehicle; A closed rectangle formed by connecting the coordinates of the four corner points is used as the static contour information of the vehicle.

5. The method according to claim 3, characterized in that: Correcting the stationary profile information using the predicted heading angle to obtain driving profile information includes: constructing a rotation matrix based on the predicted heading angle; The driving profile information is obtained by multiplying the coordinates of each corner point included in the stationary profile information by the rotation matrix.

6. A vehicle conflict detection device based on trajectory reconstruction, characterized in that: include: An acquisition module, used to acquire the trajectory information of the vehicle in the intersection area within the currently set time range, wherein the trajectory information includes the acceleration and speed of the vehicle within the currently set time range; A first detection module is used to perform abnormal point detection on the acceleration within the current set time range to obtain abnormal acceleration; An interpolation module, configured to perform interpolation processing on the abnormal acceleration to obtain an interpolated acceleration when confirming that the abnormal acceleration satisfies an interpolation condition; An obtaining module, used to update the speed corresponding to the abnormal acceleration by using the interpolated acceleration to obtain an updated speed; A prediction module, configured to predict predicted position information of the vehicle at a future set time based on the interpolated acceleration, the update speed, the acceleration and speed within the set time range except the abnormal acceleration and the speed corresponding to the abnormal acceleration, and the position information of the vehicle; A second detection module, configured to perform collision detection on the vehicle based on the predicted position information and the contour information of the vehicle; If there are a set number of non-abnormal acceleration points before the acceleration corresponding to the abnormal acceleration, the abnormal acceleration is considered to meet the interpolation condition; The prediction module is used to: Using an optimal wavelet basis that matches the trajectory signal structure of the vehicle, respectively extracting the high-frequency components of the acceleration other than the abnormal acceleration and the interpolated acceleration within the set time range, to obtain the acceleration high-frequency component and the interpolated acceleration high-frequency component; Based on the interpolated acceleration high-frequency component, the update speed, the speed within the set time range except the speed corresponding to the abnormal acceleration and the acceleration high-frequency component, and the position information of the vehicle, the predicted position information of the vehicle at a future set time is predicted.

7. An electronic device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

9. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images