Track generation method and device

Abstract

The invention provides a track generation method and device. The trajectory generation method of the embodiment of the invention comprises the steps of generating an initial trajectory of a moving object based on signaling data generated by communication between the moving object and a base station, the initial trajectory comprising a plurality of initial trajectory points; obtaining the base station density of each initial track point, wherein the base station density is used for representing the number of base stations capable of covering the initial track points in a unit area; for each initial trajectory point, determining an observation noise covariance and a process noise covariance of the initial trajectory point based on the base station density of the initial trajectory point; and performing Kalman filtering on the initial trajectory based on the observation noise covariance and the process noise covariance of each initial trajectory point to obtain an optimized trajectory of the moving object. According to the embodiment of the invention, the accuracy and reliability of the trajectory can be improved.

Description

Technical Field

[0001] This disclosure relates to the field of Internet of Things (IoT) technology, and in particular to a trajectory generation method and apparatus. Background Technology

[0002] Vehicle trajectory data plays a crucial role in traffic flow analysis, route planning, and safety management in intelligent transportation systems. With the rapid development of vehicle-to-everything (V2X) technology and the continuous improvement of intelligent transportation systems, the requirements for the accuracy and reliability of vehicle trajectory data are becoming increasingly stringent. Therefore, a highly accurate and reliable method for generating vehicle trajectories is urgently needed. Summary of the Invention

[0003] This disclosure provides a trajectory generation method and apparatus.

[0004] In a first aspect, embodiments of this disclosure provide a trajectory generation method, the method comprising:

[0005] Based on the signaling data generated by the communication between the moving object and the base station, an initial trajectory of the moving object is generated, and the initial trajectory includes multiple initial trajectory points;

[0006] Obtain the base station density for each initial trajectory point, whereby the base station density is used to characterize the number of base stations that can cover the initial trajectory point per unit area.

[0007] For each initial trajectory point, the observation noise covariance and process noise covariance of the initial trajectory point are determined based on the base station density of the initial trajectory point;

[0008] Based on the observation noise covariance and process noise covariance of each initial trajectory point, Kalman filtering is performed on the initial trajectory to obtain the optimized trajectory of the moving object.

[0009] Secondly, embodiments of this disclosure provide a trajectory generation apparatus, the apparatus comprising:

[0010] The generation module is used to generate an initial trajectory of the moving object based on signaling data generated by communication between the moving object and the base station. The initial trajectory includes multiple initial trajectory points.

[0011] The acquisition module is used to acquire the base station density of each initial trajectory point, wherein the base station density is used to characterize the number of base stations that can cover the initial trajectory point within a unit area.

[0012] The determination module is used to determine the observation noise covariance and process noise covariance of each initial trajectory point based on the base station density of the initial trajectory point;

[0013] The filtering module is used to perform Kalman filtering on the initial trajectory based on the observation noise covariance and process noise covariance of each initial trajectory point to obtain the optimized trajectory of the moving object.

[0014] Thirdly, embodiments of this disclosure provide an electronic device including a memory and a processor; the memory stores a computer program executable by the processor, and when the computer program is executed by the processor, it implements the trajectory generation method described in the first aspect.

[0015] Fourthly, embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the trajectory generation method described in the first aspect.

[0016] Fifthly, embodiments of this disclosure provide a computer program product, which includes a computer program that, when executed by a processor, implements the trajectory generation method described in the first aspect.

[0017] In this embodiment, an initial trajectory is first generated based on signaling data generated from communication between the moving object and the base station. Then, for each initial trajectory point, its base station density is obtained, and its observation noise covariance and process noise covariance are determined based on this density. Finally, Kalman filtering is applied to the initial trajectory based on the observation noise covariance and process noise covariance of each initial trajectory point to obtain an optimized trajectory for the moving object. This optimized trajectory can adapt to different base station densities, improving the accuracy and reliability of the moving object's trajectory in different base station density regions, eliminating trajectory distortion caused by uneven base station density, enhancing the scene generalization ability of trajectory generation, and avoiding the generation of abnormal trajectories. Attached Figure Description

[0018] In the accompanying drawings of the embodiments disclosed herein:

[0019] Figure 1 This is a flowchart of a trajectory generation method provided in an embodiment of the present disclosure.

[0020] Figure 2 This is a flowchart of a trajectory generation method provided in an embodiment of the present disclosure.

[0021] Figure 3 This is a block diagram of a trajectory generation device provided in an embodiment of the present disclosure.

[0022] Figure 4 This is a block diagram of an electronic device provided in an embodiment of the present disclosure.

[0023] Figure 5 This is a block diagram illustrating the composition of a computer-readable medium provided in an embodiment of the present disclosure. Detailed Implementation

[0024] To enable those skilled in the art to better understand the technical solutions of this disclosure, the embodiments of this disclosure will be described in detail below with reference to the accompanying drawings.

[0025] The present disclosure will be described more fully below with reference to the accompanying drawings; however, the embodiments shown may be embodied in different forms, and the present disclosure should not be construed as limited to the embodiments set forth below. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will enable those skilled in the art to fully understand the scope of the disclosure.

[0026] The accompanying drawings are provided to further illustrate this disclosure and form part of the specification. They are used together with the detailed embodiments to explain this disclosure and do not constitute a limitation thereof. These and other features and advantages will become more apparent to those skilled in the art from the description of detailed embodiments with reference to the accompanying drawings.

[0027] Unless otherwise specified, each embodiment and feature of this disclosure may be used individually or in combination with other embodiments and features thereof.

[0028] Those skilled in the art will understand that various changes in form and detail may be made to the embodiments of this disclosure without departing from the scope of this disclosure as set forth by the appended claims.

[0029] The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to limit the disclosure. The term "and / or" as used in this disclosure includes any and all combinations of one or more of the associated enumerated entries. The singular forms "a" and "the" as used in this disclosure are also intended to include the plural forms, unless the context clearly indicates otherwise. The terms "comprising," "made of," etc., as used in this disclosure specify the presence of the stated feature, integral, step, operation, element, and / or component, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof.

[0030] Unless otherwise specified, all terms used in this disclosure (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and this disclosure, and will not be interpreted as having an idealized or overly formal meaning, unless expressly so defined in this disclosure.

[0031] This disclosure is not limited to the embodiments shown in the accompanying drawings, but includes modifications to the configuration based on the manufacturing process. Therefore, the areas illustrated in the drawings are schematic, and the shapes of the areas shown illustrate specific shapes of the areas of an element, but are not intended to be limiting.

[0032] Trajectory fitting methods based on vehicle-to-everything (V2X) signaling data have become a hot topic in the field of intelligent transportation. However, V2X data itself has inherent defects such as sampling fluctuations, signal drift, and abnormal interruptions; at the same time, base stations are unevenly distributed across different regions, with dense base stations in urban areas and sparse base stations in rural areas and highways, which poses challenges to trajectory fitting based on signaling data.

[0033] In related technologies, trajectory fitting techniques can be mainly divided into three categories. The first category is fitting techniques based on spatiotemporal filtering and spline curves. This technique first processes noise and outliers in signaling data through spatiotemporal clustering and Gaussian filtering to construct a preliminary signaling trajectory. Then, it uses non-uniform rational B-spline curves for path fitting to adapt to the continuous driving characteristics of vehicles in dense urban road networks. The second category is handover data matching techniques based on Thiessen polygons and fuzzy decision-making. Its core is to model the base station coverage area using Thiessen polygons, transforming the base station handover trajectory into a cell trajectory. It then combines a global shortest distance algorithm or a fuzzy partial order relation multi-attribute decision-making algorithm to determine the optimal matching road segment, generating a complete travel trajectory. The third category is deep learning fitting techniques based on spatiotemporal features and attention mechanisms. This technique uses deep learning models to mine the spatiotemporal correlation features of signaling data and strengthens the correlation between base station interaction relationships and the semantic environment through attention mechanisms, reducing dependence on a single base station signal to improve the accuracy of location determination.

[0034] However, spline curve fitting techniques based on spatiotemporal filtering have poor adaptability to low sampling rate data. When there are too many signaling breakpoints, spline curves are prone to over-smoothing, deviating from actual road constraints. Furthermore, they do not deeply integrate road network data, and in complex intersections (such as overpasses and multi-way intersections), they are prone to generating trajectories that do not conform to traffic rules. In the handover data matching technology based on Thiessen polygons and fuzzy decision-making, the default base station of Thiessen polygons is uniform coverage. When the actual base station distribution is uneven (such as dense in urban areas and sparse in suburbs), the cell trajectory mapping is distorted. Moreover, the evaluation index weights of fuzzy decision-making depend on manual setting, requiring repeated debugging under different cities or road network types, resulting in high adaptation costs. In the deep learning fitting technology based on spatiotemporal features and attention mechanisms, model training requires massive amounts of labeled signaling and real trajectory data, resulting in high data acquisition costs. It is difficult to implement in small-area scenarios, and its robustness to abnormal signaling (such as temporary signal interruption or equipment offline) is weak, easily leading to trajectory breakage or misjudgment.

[0035] In this embodiment of the disclosure, an initial trajectory is first generated based on signaling data, and then an optimized trajectory is obtained by performing Kalman filtering on the initial trajectory according to the base station density. The optimized trajectory obtained in this way can adapt to different base station densities, improve the accuracy and reliability of the trajectory of the moving object in different base station density areas, eliminate the trajectory distortion caused by uneven base station density, enhance the scene generalization ability of trajectory generation, and avoid the generation of abnormal trajectories.

[0036] In related technologies, spline curve fitting is prone to deviating from the road network when there are too many signaling breakpoints. Although dynamic programming can handle low sampling rate data, probabilistic models are prone to path misjudgment in sparse base station areas due to insufficient data. In the embodiments of this disclosure, the fitting strategy is adjusted by utilizing base station density differences. For example, in sparse areas, neighboring base stations with density correlation are used to assist in inference, reducing trajectory breaks caused by breakpoints and making up for the adaptation defects of the original technology in extreme density scenarios.

[0037] In related technologies, road network base station matching technology is prone to abnormal detours due to incorrect road segment selection when base stations cover intersections; Thiessen polygon technology can also lead to distorted intersection trajectory mapping due to uneven base station distribution. This embodiment combines base station density and road network topology, refining base station-road segment matching rules for high-density intersection areas, while suppressing irrational paths through base station density correlation. Simultaneously, it uses base station density to divide road network calculation units, refining path nodes in high-density areas and simplifying calculations in low-density areas, thus ensuring the accuracy of ambiguous path identification while reducing computational load, balancing accuracy and efficiency.

[0038] Related technologies, such as fuzzy decision-making using Thiessen polygons and base station association in road network matching models, all require manual setting of indicator weights, resulting in high debugging costs when adapting to different regions. The embodiments disclosed in this disclosure can automatically optimize weights through base station density; for example, dense areas emphasize signal stability indicators, while sparse areas emphasize distance correlation indicators, reducing manual intervention and achieving low-cost multi-region adaptation.

[0039] Firstly, this disclosure provides a trajectory generation method. This trajectory generation method can be executed by an electronic device such as a terminal device or a server. The terminal device can be a user equipment (UE), mobile device, user terminal, terminal, cellular phone, cordless phone, personal digital assistant (PDA), handheld device, computing device, in-vehicle device, wearable device, etc. The trajectory generation method can be implemented by a processor calling computer-readable instructions stored in memory. Alternatively, the trajectory generation can be executed by a server. (See reference...) Figure 1 The trajectory generation method may include:

[0040] S101, Based on the signaling data generated by the communication between the moving object and the base station, an initial trajectory of the moving object is generated, the initial trajectory including multiple initial trajectory points.

[0041] The moving object can refer to an object whose location changes. For example, the moving object can be a vehicle or in-vehicle equipment. This embodiment uses a vehicle as an example for illustration. Signaling data can represent the interaction data generated when the moving object communicates with the base station through a vehicle networking module, etc., and can be extracted from the signaling system data according to certain rules. Signaling data includes, but is not limited to, sampling time, vehicle location (such as latitude and longitude), signal strength, and the identifier of the base station to which it belongs.

[0042] In this embodiment, the initial trajectory of a moving object can be generated based on signaling data. Considering that signaling data has characteristics such as unstable sampling intervals and accuracy greatly affected by the environment, this embodiment requires preprocessing the signaling data.

[0043] In one possible implementation, the preprocessing process may include converting multi-source heterogeneous data into a uniform format, outlier detection and filtering, and time series alignment.

[0044] In this embodiment of the disclosure, converting multi-source heterogeneous data into a unified format may include uniformly converting timestamps (i.e., sampling times) to UTC time zones and uniformly adopting coordinates in the WGS84 coordinate system for latitude and longitude. Outlier detection and filtering may include using [missing information - likely a specific method or technique] for continuous fields such as acquisition time, latitude and longitude, and speed. The criteria identify and filter outliers, and then use interpolation formulas for correction. Time series alignment may involve sorting the interpolated signaling data in ascending order of timestamps to obtain the initial trajectory, ensuring the temporal continuity of the initial trajectory.

[0045] In one example, the interpolation formula for longitude can be: .

[0046] in, This can represent the longitude in the k-th signaling data obtained through interpolation; These represent the longitudes in the (k-1)th and (k+1)th signaling data entries, respectively. In this embodiment of the disclosure, This represents the weight of the i-th signaling data. , This represents the signal strength of the i-th signaling data. It should be understood that the greater the signal strength, the greater the weight. In the above formula... and These represent the weights of the (k-1)th and (k+1)th signaling data, respectively. , .

[0047] In one example, the interpolation formula for latitude can be: .

[0048] in, This can represent the latitude in the k-th signaling data obtained through interpolation. The interpolation formula for latitude can be referenced from the interpolation formula for longitude, and will not be repeated here.

[0049] The interpolation formula for velocity can be: .

[0050] in, This represents the speed in the k-th signaling data obtained through interpolation. This represents the sampling time interval between the k-th signaling data and the (k-1)-th signaling data. , This represents the sampling time interval between the (k+1)th signaling data and the kth signaling data. , This indicates the rate at which the (k-1)th signaling data was collected. This indicates the speed at which the (k+1)th signaling data is collected.

[0051] For example, suppose the kth signaling data is collected. signal strength This is a low-confidence point and requires interpolation. The collected (k-1)th signaling data... The sampling time, longitude, latitude, signal strength, and velocity are as follows: Its weight The (k+1)th signaling data collected. The sampling time, longitude, latitude, signal strength, and speed are respectively Its weight Using the above interpolation formula, the interpolation result for the k-th signaling data is as follows: .

[0052] In one example, the preprocessed signaling data is: Where D represents the initial trajectory, Let represent the k-th initial trajectory point of the initial trajectory D, and n represent the number of initial trajectory points in the initial trajectory. ,in, These represent the sampling time, longitude, latitude, signal strength, speed, and the identifier of the base station to which the k-th initial trajectory point belongs, respectively.

[0053] It should be understood that the embodiments disclosed herein also involve base station data and routing data.

[0054] Base station data represents the attributes of a base station. Base station data includes, but is not limited to, the location (e.g., latitude and longitude), coverage radius, and base station identifier. In this embodiment, base station density can be calculated based on the base station data, and the spatial relationship (e.g., distance, angle) between vehicles and base stations can be determined based on the base station data. Road network data describes the spatial location, topology, attribute information, and traffic rules of the road network. Road network data is fundamental data supporting applications such as Geographic Information Systems (GIS) and navigation. The preprocessing process for base station data and routing data can refer to the preprocessing process for signaling data; this embodiment does not impose any limitations.

[0055] In one example, the preprocessed base station data is Where B represents base station data, This represents the base station data of the m-th base station, where M represents the number of base stations. ,in, These represent the longitude, latitude, and coverage radius of the m-th base station, respectively.

[0056] In one example, the preprocessed road network data is Where G represents road network data, V represents road segment identification, and E represents the latitude and longitude, road grade, road shape, and road type of the road segment.

[0057] S102, Obtain base station density information for each initial trajectory point.

[0058] The base station density information can be used to characterize the number of base stations that can cover the initial trajectory point per unit area. A higher base station density indicates a greater number of base stations near the initial trajectory point, potentially resulting in a smaller positional error for the initial trajectory; conversely, a lower base station density indicates less base station data near the initial trajectory point, potentially leading to a larger positional error for the initial trajectory point.

[0059] In one possible implementation, step S102 may include:

[0060] S1021, using the position of the initial trajectory point as the center and the base station query radius as the radius, determine the base station query range.

[0061] With the initial trajectory point Location, i.e., latitude and longitude Using the base station as the center, the query radius The radius is used to determine the base station query range. The base station query radius is pre-set and can cover most or even all base stations within the current movement range of the moving target. For example, the base station query radius can be set to 2000 meters.

[0062] S1022, for each base station within the query range of the base station, if the straight-line distance between the initial trajectory point and the base station is less than the coverage radius of the base station, the base station is determined as a valid base station.

[0063] Get the set of base stations within the base station query range For base station sets base stations in (i.e., the m-th base station), calculate the initial trajectory point With base station The straight-line distance. In one example, the straight-line distance... The unit for straight-line distance is meters, and 111319.5 represents the number of meters corresponding to 1 degree of longitude. These represent the initial trajectory points. latitude and longitude Representing base stations The latitude and longitude.

[0064] When straight distance At that time, base station Marked as a valid base station, i.e. Otherwise, the base station Marked as an invalid base station, i.e. .

[0065] For example, initial trajectory point of Base station set There are 30 base stations in the area, of which 20 base stations have a coverage radius of... The distance between the base station and the initial trajectory point is less than or equal to 500 meters. Then the number of effective base stations is... .

[0066] S1023, determine the statistical radius based on the coverage radius of each effective base station.

[0067] In one example, the minimum coverage radius among all effective base stations is determined as the statistical radius, i.e., the statistical radius. It should be understood that the above is only an example of the statistical radius. The statistical radius can also be determined in other ways, such as using the average coverage radius of each effective base station as the statistical radius. It should be noted that in extreme scenarios where there are no effective base stations, a default value (such as 1000 meters) can be used as the statistical radius r.

[0068] S1024, using the position of the initial trajectory as the center and the statistical radius as the radius, determine the statistical range of the base station.

[0069] S1025, determine the base station density of the initial trajectory point based on the number of effective base stations and the area of ​​the statistical range of the base stations.

[0070] Among them, the area of ​​the base station's statistical scope. The unit is square kilometers ( ).

[0071] In this embodiment of the disclosure, the base station density is obtained based on the number of effective base stations around the initial trajectory point and the dynamic statistical radius. Compared with a fixed statistical radius, this improves the accuracy of base station density calculation in low-density base station areas.

[0072] In one example, step S1025 may include: dividing the number of effective base stations by the area of ​​the statistical range of the base stations to obtain the base station density of the initial trajectory points.

[0073] Initial trajectory point The base station density is: .

[0074] For example, suppose an initial trajectory point in a densely populated urban area has 22 effective base stations, a statistical radius of 500 meters, and an area of ​​approximately 0.785. Therefore, the base station density of this initial trajectory point is approximately 22 / 0.785, or 28 per point. Assuming the initial trajectory point in a rural area has one effective base station, a statistical radius of 2000 meters, and an area of ​​approximately 12.566 square kilometers... Therefore, the base station density of this initial trajectory point is approximately 1 / 12.566, or 0.08 per [unit / location]. .

[0075] In another example, step S1025 may include: dividing the number of effective base stations by the area of ​​the statistical range of the base stations to obtain the average density of the initial trajectory points; determining the mean of the average density of the first trajectory points, the average density of the second trajectory points, and the average density of the initial trajectory points; and determining the mean as the base station density of the initial trajectory points.

[0076] Wherein, the first trajectory point represents the point in the initial trajectory that is located in the initial trajectory. One or more previous initial trajectory points, such as the initial trajectory point and initial trajectory point The second trajectory point represents one or more initial trajectory points that are located after the initial trajectory point in the initial trajectory, such as the initial trajectory point. and initial trajectory point The method for determining the average density of the first and second trajectory points can be the same as that for determining the initial trajectory points. The method for determining the average density will not be elaborated here.

[0077] In this embodiment, the base station density of the initial trajectory points can be smoothed using a sliding window. This eliminates fluctuations in base station density at single points, such as sudden changes in the number of effective base stations caused by momentary signal blockage. The window size can be set to 2w+1, where w represents the smoothing range of the base station density before and after each initial trajectory point. Assuming w is 5, the window size is 11, meaning the smoothing range of the base station density for each initial trajectory point is the base station density of the five initial trajectory points before and after it. The smoothing formula is as follows: .

[0078] It should be noted that for initial trajectory points whose beginning and end are less than the window length, a boundary replication strategy is adopted. For example, when k is 1 and w is 5, the value of i ranges from 0 to 10.

[0079] For example, suppose the base station density of seven consecutive initial trajectory points on a rural road section is... They are respectively: ; window The smoothing density of the 4th initial trajectory point / km². In this embodiment of the disclosure, the smoothed base station density (referred to as smoothed density) can be determined as the base station density of the initial trajectory points.

[0080] In this way, the impact of instantaneous fluctuations in base station signals is eliminated by smoothing through a sliding window of base station density.

[0081] In another example, step S1025 may include: dividing the number of effective base stations by the area of ​​the statistical range of the base stations to obtain the initial value of the initial trajectory point; determining the average density of the first trajectory point, the average density of the second trajectory point, and the mean of the average density of the initial trajectory point; searching for the density level corresponding to the mean value in a preset level mapping table, the preset level mapping table including multiple density levels and the value range corresponding to each density level; and determining the found density level as the base station density of the initial trajectory point.

[0082] In this embodiment of the disclosure, the smoothed density can also be quantified and mapped to a density level. For example, if the smoothed base station density is greater than or equal to a first threshold, the density level is determined to be a dense urban area; if the smoothed base station density is greater than or equal to a second threshold and less than the first threshold, the density level is determined to be a general urban area; if the smoothed base station density is greater than or equal to a third threshold and less than the second threshold, the density level is determined to be a suburban area; and if the smoothed base station density is less than the third threshold, the density level is determined to be a rural area (or highway). The first threshold is greater than the second threshold, and the second threshold is greater than the third threshold. For example, the first threshold is 10, the second threshold is 5, and the third threshold is 2.

[0083] It should be understood that a level value can be set for densely populated urban areas, general urban areas, suburbs, and rural areas (or highways), such as 4, 3, 2, 1, etc. In this way, the smooth density is quantified into a discrete number with a finite number of values, which can reduce the amount of computation when using base station density for subsequent calculations.

[0084] S103, for each initial trajectory point, determine the observation noise covariance and process noise covariance of the initial trajectory point based on the base station density of the initial trajectory point.

[0085] In this embodiment of the disclosure, after determining the base station density of each initial trajectory point in the initial trajectory, adaptive filtering (such as Kalman filtering) can be performed on the initial trajectory based on the base station density of each initial trajectory point to obtain an optimized trajectory. Specifically, when performing adaptive filtering based on the base station density of the initial trajectory points, the observation noise covariance and process noise covariance of the initial trajectory points can be determined based on the base station density. Then, adaptive filtering is performed based on these observation noise covariance and process noise covariance.

[0086] To facilitate understanding, we will first explain the state vector, state transition equation, and observation equation used in the adaptive filtering process.

[0087] For the motion characteristics of a moving object, the state vector at time k is defined as follows:

[0088] .

[0089] in, This represents the longitude of the moving object in the WGS84 coordinate system at time k; This represents the latitude of the moving object in the WGS84 coordinate system at time k; This represents the speed of the moving object at time k, in km / h. This represents the direction angle of the moving object at time k, in radians, with a value range of... Assume that the value of due north is 0, and the values ​​increase clockwise.

[0090] Considering the curvature of the Earth and the continuity of motion of the moving object, the state transition matrix is ​​defined as:

[0091] .

[0092] in, , The sampling time interval is expressed in hours, equivalent to converting speed (km / h) to hourly displacement. 111319.5 is the Earth's equatorial circumference conversion factor; 1 degree of latitude corresponds to 111319.5 meters. The higher the latitude, the shorter the actual distance corresponding to 1 degree of longitude. The Earth's equatorial circumference conversion factor can be used to correct for changes in longitude with latitude. In this embodiment, an Earth curvature correction term is introduced into the state transition matrix, reducing the conversion error of latitude and longitude distances in high-latitude regions.

[0093] The matrix form of the state transition equation is: .

[0094] in, For process noise, That is, the process noise follows a pattern with a mean of 0 and a covariance of . The state vector follows a normal distribution. The state transition equation characterizes the state vector from time k-1. The state vector at time k The state transition process.

[0095] The observation equation is defined as: .

[0096] Where H is the observation matrix, ; To observe the noise, That is, the observed noise follows a pattern with a mean of 0 and a covariance of 0. It follows a normal distribution.

[0097] In this embodiment, the process noise covariance can be adjusted in real time according to the base station density. and observation noise covariance For ease of understanding, the following text uses smooth base station density. This example of base station density is used for illustration and is not intended to limit base station density.

[0098] The process of determining the observation noise covariance and the process noise covariance is explained below.

[0099] In one possible implementation, determining the observation noise covariance of the initial trajectory point based on the base station density of the initial trajectory point in step S103 may include: determining the observation data noise coefficient based on the base station density of the initial trajectory point, wherein the signaling noise coefficient is negatively correlated with the base station density; determining the noise variance of the observation data measurement based on the signal strength of the initial trajectory point; and determining the observation noise covariance based on the product of the observation data noise coefficient and the noise variance of the observation data measurement.

[0100] Considering that lower base station density leads to larger signaling positioning errors, it is necessary to reduce the confidence level of the observation data, i.e., increase the observation noise covariance. .

[0101] In one example, the observation noise covariance is determined by the following formula:

[0102] .

[0103] Among them, a, b and These are empirical parameters, and their values ​​can be determined using a large amount of sample data (e.g., 100,000 data points). For example, the value of 'a' could be 0.3, and the value of 'b' could be 2.5. The value can be 0.1; The fundamental noise variance is based on signal strength. , This represents the signal strength at time k (i.e., the kth initial trajectory point). It is evident that the stronger the signal strength, the smaller the fundamental noise variance. The units for both the fundamental noise variance and the observation noise covariance are... .

[0104] For example, the initial trajectory point in a densely populated urban area Smooth base station density The variance of the fundamental noise can be obtained. Therefore, the observation noise covariance can be obtained. (unit: Initial trajectory points in the rural area. Smooth base station density The variance of the fundamental noise can be obtained. Therefore, the observation noise covariance can be obtained. (unit ).

[0105] In one possible implementation, the process noise covariance determination based on the base station density of the initial trajectory points in step S103 may include: determining the process noise variance of the location based on the base station density of the initial trajectory points, wherein the process noise variance of the location is negatively correlated with the base station density; and determining the process noise covariance based on the process noise variance of the location, the process noise variance of the preset velocity, and the process noise variance of the preset orientation angle.

[0106] Considering that the lower the base station density, the greater the uncertainty in trajectory prediction, it is necessary to enhance the filtering correction, i.e., increase the process noise covariance. The position noise term in the data.

[0107] In one example, the process noise covariance is determined by the following formula:

[0108] .

[0109] Where c represents the location noise figure, and c can take values ​​such as 0.02; d represents the velocity or azimuth noise figure (i.e., the preset process noise variance of velocity and the preset process noise variance of azimuth), and d can take values ​​such as 0.1. In this embodiment of the present disclosure, the location noise system can smooth the base station density. Dynamic adjustments are made to enhance the filtering correction in real time; speed or directional noise remains stable. This represents a diagonal matrix, where the first two terms are location noise terms, which increase as the density of smooth base stations decreases.

[0110] For example, the initial trajectory point in a densely populated urban area Smooth base station density The process noise covariance can be obtained. Location noise term Initial trajectory points in rural areas Smooth base station density The process noise covariance can be obtained. Location noise term It can be seen that the location noise term in rural areas is about 11 times that in densely populated urban areas, which enhances the correction power for predictions in areas with low base station density.

[0111] In this embodiment of the disclosure, by constructing the correlation between base station density and observation noise covariance and process noise covariance, dynamic adjustment and real-time adaptation of observation noise covariance and process noise covariance are achieved.

[0112] S104, based on the observation noise covariance and process noise covariance of each initial trajectory point, perform Kalman filtering on the initial trajectory to obtain the optimized trajectory of the moving object.

[0113] The process of performing Kalman filtering on the initial trajectory includes calculating the initial state vector. and the initial error covariance matrix Iterative prediction step and update step.

[0114] Wherein, the initial state vector and the initial error covariance matrix It can be calculated based on the first few (e.g., 3) initial trajectory points in the sorted sequence. For example, the initial state vector. The longitude and latitude are the average longitude and latitude of the first three initial trajectory points, respectively. The speed can be the average speed calculated based on the spatiotemporal difference of the first three initial trajectory points, and the direction angle can be the driving direction calculated from the longitude and latitude of the first two initial trajectory points. Initial error covariance matrix. The diagonal elements are the initial error variances of latitude and longitude, velocity, and direction angle, which can be estimated based on the dispersion of the first three initial trajectory points.

[0115] It should be noted that the trajectory data does not directly carry the direction angle. This can be determined by the difference in latitude and longitude between adjacent trajectory points (converted to meters). Calculate the driving direction angle. In subsequent iterations, and by combining observation data (latitude and longitude) for indirect correction, the accuracy is gradually improved.

[0116] State prediction can be expressed as .in, Indicates the predicted state at time k. This represents the optimal estimated state at time k-1. The error covariance prediction can be expressed as... .in, State transition matrix Jacobian matrix, state transition matrix For nonlinear models, the Jacobian matrix... Nonlinear models can be linearized. Jacobian matrix. The formula is as follows:

[0117] .

[0118] The update step refers to the process of correcting observations by fusing observation data. Kalman gain calculation can be expressed as... Kalman gain can be used to balance the confidence levels of predictions and observations.

[0119] The optimal state estimate can be expressed as .

[0120] Error covariance update can be expressed as Where I is the identity matrix.

[0121] After the iteration is complete, the optimal state estimate at time k is output. This refers to the k-th optimized trajectory point. All optimized trajectory points together form the optimized trajectory.

[0122] An optimized trajectory can include multiple optimized trajectory points, which correspond to the optimal state estimate. It integrates the observation data at the current moment and obtains the final result through Kalman gain correction. It is a filtered trajectory point with high reliability.

[0123] In this embodiment, an initial trajectory is first generated based on signaling data generated from communication between the moving object and the base station. Then, for each initial trajectory point, its base station density is obtained, and its observation noise covariance and process noise covariance are determined based on this density. Finally, Kalman filtering is applied to the initial trajectory based on the observation noise covariance and process noise covariance of each initial trajectory point to obtain an optimized trajectory for the moving object. This optimized trajectory can adapt to different base station densities, improving the accuracy and reliability of the moving object's trajectory in different base station density regions, eliminating trajectory distortion caused by uneven base station density, enhancing the scene generalization ability of trajectory generation, and avoiding the generation of abnormal trajectories.

[0124] In some embodiments, the trajectory generation method provided in this disclosure may further include:

[0125] S105, determine the road network search radius based on the base station density corresponding to the optimized trajectory point and the speed of the optimized trajectory point. The road network search radius is negatively correlated with the base station density and positively correlated with the speed.

[0126] In one example, the road network search radius can be determined using the following formula:

[0127] .

[0128] Where e can represent the base radius coefficient, which can take the value of 80, and the unit is meters; f can represent the density compensation coefficient, which can take the value of 300. It can represent a smoothing term, and its value can be 0.1; This can represent the filtered vehicle speed, i.e., the speed of the optimized trajectory point, in km / h. Higher speeds result in a larger road network search radius, thus better adapting to trajectory prediction of fast-moving objects. Lower smoothing base station density results in a larger road network search radius, thus better adapting to trajectory prediction in low-density areas. Based on data calibration and scenario verification, the density compensation coefficient and smoothing term are determined by statistically analyzing the spacing between typical road segments in the urban road network. Based on actual measurements of positioning errors in scenarios with different base station densities, to avoid a denominator of 0 in scenarios without base stations, a [missing information - likely a setting or parameter] is used. =0.1, which can smooth out the impact of small fluctuations in base station density on the road network search radius and avoid sudden changes in the road network search radius.

[0129] In this embodiment, the road network search radius is determined by combining base station density and vehicle speed, thereby achieving adaptive search that expands the search range in low-density areas and reserves buffers for high-speed scenarios.

[0130] It should be understood that in this embodiment of the disclosure, constraints can also be set for the road network search radius, limiting it to a range of values ​​to avoid missed matches due to an excessively small search radius or increased computational load due to an excessively large search radius. For example, the road network search radius can be set to... The radius of the road network search is defined as 50 meters or more and 1000 meters or less. If the road network search radius calculated using the above formula is less than 50 meters, then the road network search radius is set to 50; if the road network search radius calculated using the above formula is greater than 1000 meters, then the road network search radius is set to 1000 meters.

[0131] For example, the smoothing base station density for optimized trajectory points in densely populated urban areas. =12 pieces / Optimize speed =40km / h, then the road network search radius The meter is rounded to 147 meters. Optimized trajectory points in rural areas. =1 / Optimize speed =80km / h, then the road network search radius The distance is 635 meters, rounded to the nearest meter. This allows for a smaller road network search radius in densely populated base station areas to reduce computation, while a larger radius in sparsely populated areas can cover a sufficient number of candidate road segments.

[0132] S106, using the location of the optimized trajectory point as the center and the road network search radius as the radius, determine the road network search range.

[0133] S107, road segments within the road network search range are identified as candidate road segments.

[0134] In this embodiment of the disclosure, the location of trajectory points can be optimized by querying using an R-tree spatial index. Center and road network search radius Generate a candidate road segment set for all road segments within the specified range. .in, This represents the j-th candidate road segment.

[0135] S108 determines the distance factor, speed factor, and direction factor for each candidate road segment.

[0136] The distance factor can be used to characterize the vertical distance from the location of the optimized trajectory point to the candidate road segment. The distance factor is negatively correlated with this vertical distance. The distance factor describes the spatial distance consistency from the optimized trajectory point to the candidate road segment. In one example, the distance factor can be expressed as:

[0137] .

[0138] in, To optimize the position of trajectory points To candidate road sections The vertical distance (unit: meters), the smaller this distance, the higher the distance factor. The closer it is to 1, the larger the distance factor.

[0139] The speed factor can be used to characterize the speed difference between the optimized trajectory point's speed and the candidate road segment's speed limit. The speed factor is negatively correlated with this speed difference. The speed factor describes the consistency between the optimized trajectory point's speed and the candidate road segment's speed limit. In one example, the speed factor can be expressed as:

[0140] .

[0141] in, Candidate road sections Speed ​​limit (unit: km / h), optimize the speed of trajectory points With speed limit The closer they are, the higher the velocity factor. The closer it is to 1, the larger the velocity factor.

[0142] The direction factor can be used to characterize the difference between the driving direction of the optimized trajectory point and the direction of the candidate road segment. The direction factor is negatively correlated with this difference. The direction factor describes the consistency between the driving direction of the optimized trajectory point and the direction of the candidate road segment. In one example, the direction factor can be expressed as:

[0143] .

[0144] in, Candidate road sections The direction angle (unit: radians, calculated from the starting and ending coordinates of the candidate road segment) is used to optimize the direction angle of the trajectory point. and candidate segment orientation angle The smaller the angle difference between them, the greater the direction factor. The closer it is to 1, the larger the direction factor.

[0145] S109: Each candidate road segment is scored based on its distance factor, speed factor, and direction factor.

[0146] In one possible implementation, step S109 may include:

[0147] S1091, if the base station density corresponding to the optimal trajectory point is greater than a preset threshold, the preset value is determined as the weight adjustment parameter; if the base station density corresponding to the optimal trajectory point is less than or equal to the preset threshold, the weight adjustment parameter is determined according to the base station density, and the weight adjustment parameter is positively correlated with the base station density.

[0148] The weighting parameter of the matching factor based on base station density is defined as follows: It can be determined by the following formula:

[0149] .

[0150] Where g represents the reference density, i.e., the preset threshold, in units of 1 / 2000. The value of g can be 5. When the base station density... When this occurs, it indicates that the optimized trajectory points are in a high-density region, and the weight adjustment parameters... At this point, the positioning error is small, so the distance factor is trusted first. When the distance factor is low, it indicates that the optimized trajectory points are in a low-density area. The weight adjustment parameter decreases as the base station density decreases. At this time, the positioning error is large, and the consistency between vehicle speed and road speed limit is more reliable. Therefore, the distance factor can be weakened and the weight of speed and direction factors can be strengthened.

[0151] For example, suppose the base station density in a densely populated urban area... Weight adjustment parameters Assuming a base station in a rural area. Weight adjustment parameters .

[0152] S1092, Based on the weight adjustment parameter, determine the distance factor weight, speed factor weight, and direction factor weight, wherein the weight adjustment parameter is positively correlated with the distance factor weight, and negatively correlated with the speed factor weight and the direction factor weight.

[0153] S1093, according to the distance factor weight, speed factor weight and direction factor weight, the distance factor, speed factor and direction factor of the candidate road segment are weighted and summed to obtain the score of the candidate road segment.

[0154] For candidate road sections The score is calculated by combining distance, speed and direction factors, and the weight of each factor is adjusted by a weight adjustment parameter during the calculation process.

[0155] In one example, the score for a candidate road segment can be calculated using the following formula: .

[0156] Among them, h can be used to adjust the weights of the direction factor and the velocity factor, and is independent of the weight of the distance factor. Considering that the stability of velocity is higher than that of direction, the value of h can be set to a value greater than 0.5 and less than 1. For example, the value of h can be 0.6. It can be used to adjust the weights of distance factor, direction factor, and velocity factor, among which, The adjustment ranges for the weights of the direction factor and the velocity factor are consistent.

[0157] For example, suppose there are candidate road segments in a rural scene. Weight adjustment parameters =0.2, h takes a value of 0.6. Optimize the position of trajectory points. To candidate road sections vertical distance Given a road network search radius R(k) = 635 meters, the distance factor can be obtained. Optimize the speed of trajectory points. km / h, candidate road segment Speed ​​limit km / h, the speed factor can be obtained Optimize the direction angle of the trajectory points. Curvature, candidate road segment The direction angle Radius, from which the direction factor can be obtained. In summary, the scores for the candidate road sections are as follows: .

[0158] In this embodiment of the disclosure, a multi-factor weight dynamic allocation mechanism is adopted to achieve strategy switching between high density focusing on distance and low density focusing on velocity direction.

[0159] S110, the candidate road segment with the highest score is determined as the optimal road segment of the optimized trajectory point.

[0160] Optimal route segment for optimizing trajectory points .

[0161] S111, the position of the optimized trajectory point is projected onto the optimal road segment, and the position of the perpendicular foot of the projection is determined as the position of the road network fitting point corresponding to the optimized trajectory point.

[0162] Optimize the position of trajectory points Projected to the optimal road segment The position of the foot of the perpendicular is used to obtain the road network fitting point. Its location is This ensures that the road network fitting point falls on the optimal road segment.

[0163] S112, Generate the road network fitting trajectory based on the fitting points of each road network.

[0164] Fitting points of each road network According to their respective sampling times Arranging the data in ascending order yields the road network fitting trajectory. This ensures the temporal continuity of the trajectory. The road network fitting trajectory can be represented as follows: .

[0165] In this embodiment, a score is calculated for each candidate road segment of each optimized trajectory point. The optimal road segment is selected as the candidate road segment with the highest score. The optimized trajectory point is projected onto the optimal road segment to obtain the road network fitting point. The fitting points of adjacent optimized trajectory points are connected through road segments to avoid trajectory breakage or deviation from the road, and finally form a continuous trajectory that fits the road network.

[0166] In some embodiments, the trajectory generation method provided in this disclosure may further include: smoothing the road network fitted trajectory to obtain a smooth trajectory.

[0167] In this embodiment of the disclosure, a third-order B-spline curve can be used to eliminate abrupt changes in road segment transitions. In one example, the following formula can be used for smoothing: .in, Using B-spline basis functions, the final output is a smooth trajectory. The format is structured data containing timestamps, latitude and longitude, and the identifier of the road segment to which it belongs.

[0168] For example, three consecutive fitting points for a rural curve: , Supplement virtual points ,right part, hour: ; Similarly, output a continuous and smooth curve trajectory.

[0169] The acquisition of latitude and longitude of supplementary virtual points follows the principle of spatiotemporal continuity extension, prioritizing linear extension based on adjacent fitted points. If the linearly extended virtual points exceed the road network range (such as extending to non-road areas), the direction is adjusted in conjunction with the optimal road segment.

[0170] In this embodiment, base station density is used as the core optimization parameter throughout the entire trajectory fitting process. Compared with fixed parameters for adapting to all scenarios, this embodiment dynamically adjusts the filter noise covariance (including observation noise covariance and process noise covariance), road network search radius, and matching factor weights through base station density. This enables the switching of strategies for relocating longitude in high-density areas and imposing scenario constraints in low-density areas, reducing the probability of trajectory drift and mismatch in rural areas, highways, and other base station coefficient areas, and improving accuracy.

[0171] Figure 2 A schematic diagram of the trajectory generation method provided in an embodiment of this disclosure is shown. (Refer to...) Figure 2 The trajectory generation method provided in this disclosure embodiment may include:

[0172] S201, Data Acquisition and Preprocessing.

[0173] The data collected in this embodiment of the disclosure may include at least signaling data, base station data, and road network data.

[0174] The signaling data includes, but is not limited to, the International Mobile Subscriber Identity (IMSI), current location time (curr_time), longitude, latitude, and the identifier of the base station (cell_id). The IMSI can be used to uniquely identify moving objects. The current location time is the data collection time.

[0175] Base station data includes, but is not limited to, base station identifiers, longitude, latitude, and coverage radius. Each base station connected to by the moving object during its journey is statistically analyzed, resulting in both basic data (i.e., base station data) collection and real-time connection statistics. The total number of effective base stations across the entire region is fixed during data preprocessing; for example, the total number of all normally operating base stations in a city or region is fixed. Base stations connected to by the moving object during its journey are matched with their base station representations in the base station data using signaling data to determine the location and coverage area of ​​the currently connected base station, thus participating in the selection of effective base stations.

[0176] Road network data includes, but is not limited to, road identifiers (road_id), the latitude and longitude of the road's starting point, ending point, and midpoint, road shape (road_shape), road type (road_type), and road level (road_level). Road identifiers are used to distinguish different road segments and can be strings or numbers. The road starting point indicates the beginning of a road segment, the road ending point indicates the end of a road segment, and the road midpoint indicates the middle point of a road segment. The road starting and ending points can be used to determine the direction and length of a road segment, while the road midpoint can be used for quick location of the road segment. Road shape describes the geometry of a road segment, such as straight sections, right-turn lanes, roundabouts, and uphill ramps. Road type indicates the purpose or form of a road segment, such as motor vehicle lanes, bus lanes, non-motorized vehicle lanes, tunnels, interchange ramps, bridges, and one-way streets. Road level indicates the functional level or standards involved in a road segment, such as highways, urban arterial roads, expressways, and rural roads.

[0177] To facilitate unified processing of the aforementioned data, preprocessing is required in this embodiment. For example, noise points are removed using the 3σ criterion, and low-intensity execution points are interpolated. The final output is the preprocessed data, such as the initial trajectory.

[0178] S202, perform base station density quantization on the initial trajectory points.

[0179] First, calculate the initial trajectory points. The base station density, which is the density of base stations that can cover the initial trajectory point per unit area. The number of base stations. In one example, the initial trajectory point The base station density is .

[0180] Then, the base station density is smoothed to eliminate fluctuations. In one example, smoothing is performed using a sliding window. Initial trajectory points Smoothed base station density .

[0181] Finally, the smoothed base station density is quantified to obtain the density level. For example, if the smoothed base station density is greater than or equal to the first threshold, the density level is determined to be a dense urban area; if the smoothed base station density is greater than or equal to the second threshold but less than the first threshold, the density level is determined to be a general urban area; if the smoothed base station density is greater than or equal to the third threshold but less than the second threshold, the density level is determined to be a suburban area; and if the smoothed base station density is less than the third threshold, the density level is determined to be a rural area or a highway area.

[0182] S203, based on the smoothed base station density, adaptive filtering is performed on the initial trajectory to obtain the optimized trajectory.

[0183] In this embodiment of the disclosure, an Adaptive Extended Kalman Filter (AEKF) can be used to dynamically adjust noise parameters based on the smoothed base station density.

[0184] Constructing the state vector: .in, Let represent the state vector at time k (i.e., the kth initial trajectory point). These represent the longitude, latitude, velocity, and direction angle at the k-th moment, respectively.

[0185] Construct the observation equation: .

[0186] Constructing the observation noise covariance: .

[0187] Construction process noise covariance: .

[0188] Based on the aforementioned state vector, observation method, observation noise covariance, and process noise covariance, an adaptive extended Kalman filter is performed to output the optimized trajectory: .

[0189] S204, density-aware network matching based on base station density and optimized trajectory.

[0190] First, candidate road segments are searched. The search radius is dynamically adjusted based on base station density and vehicle speed in the optimized trajectory. .

[0191] Then, the matching degree is calculated. Density weights are introduced. Based on density weights Calculate the score for each candidate road segment The candidate road segment with the highest score is determined as the optimal matching road segment.

[0192] Finally, the optimized trajectory points in the optimized trajectory are mapped to the optimal matching road segments to obtain the optimal matching trajectory.

[0193] S205, smooth the optimal matching trajectory to obtain a smoothed trajectory.

[0194] A third-order B-spline curve is used to eliminate abrupt changes in road segment transitions, outputting a smooth trajectory. In one example, the smooth trajectory... ,in, These are B-spline basis functions.

[0195] In this embodiment, a closed-loop optimization structure is formed by four stages: preprocessing, filtering, matching, and smoothing, integrating constraint information from three types of data: signaling data, base station density, and road network data. At the data layer, weighted interpolation corrects low-confidence signaling points, eliminating missing data and noise issues in the original data. At the model layer, adaptive extended Kalman filtering is used, combined with dynamic adjustment of noise parameters based on base station density, to achieve accurate estimation of nonlinear trajectories. At the application layer, density-aware road network matching and third-order B-spline smoothing ensure that the trajectory conforms to the road network topology while maintaining physical motion continuity.

[0196] Secondly, referring to Figure 3 This disclosure provides a trajectory generation device, which may include:

[0197] The generation module is used to generate the initial trajectory of the moving object based on the signaling data generated by the moving object communicating with the base station during the movement. The initial trajectory includes multiple initial trajectory points.

[0198] The acquisition module is used to acquire the base station density of each initial trajectory point, wherein the base station density is used to characterize the number of base stations that can cover the initial trajectory point per unit area.

[0199] The determination module is used to determine the observation noise covariance and process noise covariance of each initial trajectory point based on the base station density of the initial trajectory point.

[0200] The filtering module is used to perform Kalman filtering on the initial trajectory based on the observation noise covariance and process noise covariance of each initial trajectory point to obtain the optimized trajectory of the moving object.

[0201] Thirdly, referring to Figure 4 This disclosure provides an electronic device, which includes a memory and a processor; the memory stores a computer program that can be executed by the processor, and when the computer program is executed by the processor, it implements any one of the trajectory generation methods of this disclosure.

[0202] Fourthly, refer to Figure 5 This disclosure provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements any of the trajectory generation methods of this disclosure.

[0203] Fifthly, embodiments of this disclosure provide a computer program product, which includes a computer program that, when executed by a processor, implements any one of the trajectory generation methods of this disclosure.

[0204] Among them, the processor is a device with data processing capabilities, including but not limited to the central processing unit (CPU); the memory is a device with data storage capabilities, including but not limited to random access memory (RAM), more specifically such as SDRAM, DDR, etc., read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory (FLASH); I / O interface, or read-write interface, is connected between the processor and the memory, enabling information exchange between the memory and the processor, including but not limited to the data bus (Bus).

[0205] Those skilled in the art will understand that all or some of the steps, systems, and devices disclosed above, as functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0206] In hardware implementations, the division between functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be executed by several physical components working together.

[0207] Some or all of the physical components may be implemented as software executed by a processor, such as a central processing unit (CPU), digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software may be distributed on a computer-readable medium, which may include computer storage media and communication media. In embodiments of this disclosure, computer storage media include volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information, and any other media that can be used to store desired information and can be accessed by a computer. In embodiments of this disclosure, communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

Claims

1. A trajectory generation method, characterized in that, The method includes: Based on the signaling data generated by the communication between the moving object and the base station, an initial trajectory of the moving object is generated, and the initial trajectory includes multiple initial trajectory points; Obtain the base station density for each initial trajectory point, whereby the base station density is used to characterize the number of base stations that can cover the initial trajectory point per unit area. For each initial trajectory point, the observation noise covariance and process noise covariance of the initial trajectory point are determined based on the base station density of the initial trajectory point; Based on the observation noise covariance and process noise covariance of each initial trajectory point, Kalman filtering is performed on the initial trajectory to obtain the optimized trajectory of the moving object.

2. The method according to claim 1, characterized in that, The process of obtaining the base station density for each initial trajectory point includes: For each initial trajectory point: The base station query range is determined by taking the position of the initial trajectory point as the center and the base station query radius as the radius; For each base station within the query range, if the straight-line distance between the initial trajectory point and the base station is less than the coverage radius of the base station, the base station is determined as a valid base station. The statistical radius is determined based on the coverage radius of each effective base station; The statistical range of the base station is determined by taking the position of the initial trajectory as the center and the statistical radius as the radius; The base station density of the initial trajectory point is determined based on the number of effective base stations and the area of ​​the statistical range of the base stations.

3. The method according to claim 2, characterized in that, Determining the base station density of the initial trajectory point based on the number of effective base stations and the area of ​​the statistical range of the base stations includes: The base station density of the initial trajectory point is obtained by dividing the number of effective base stations by the area of ​​the statistical range of the base stations.

4. The method according to claim 2, characterized in that, Determining the base station density of the initial trajectory point based on the number of effective base stations and the area of ​​the statistical range of the base stations includes: The average density of the initial trajectory points is obtained by dividing the number of effective base stations by the area of ​​the statistical range of the base stations. The average density of the first trajectory point, the average density of the second trajectory point, and the mean of the average density of the initial trajectory point are determined. The first trajectory point represents one or more initial trajectory points in the initial trajectory that are preceding the initial trajectory point, and the second trajectory point represents one or more initial trajectory points in the initial trajectory that are following the initial trajectory point. The mean value is determined as the base station density of the initial trajectory points.

5. The method according to claim 4, characterized in that, The step of determining the base station density of the initial trajectory point based on the number of effective base stations and the area of ​​the statistical range of the base stations further includes: In a preset level mapping table, the density level corresponding to the mean is found. The preset level mapping table includes multiple density levels and the value range corresponding to each density level. The density level found is determined as the base station density of the initial trajectory point.

6. The method according to claim 1, characterized in that, Determining the observation noise covariance of the initial trajectory points based on the base station density includes: The noise coefficient of the observation data is determined based on the base station density of the initial trajectory points, and the signaling noise coefficient is negatively correlated with the base station density; Based on the signal strength of the initial trajectory points, determine the noise variance of the observed data measurements; The observation noise covariance is determined by multiplying the noise coefficient of the observation data by the noise variance of the observation data measurement.

7. The method according to claim 1, characterized in that, The process noise covariance of the initial trajectory points is determined based on the base station density of the initial trajectory points, including: Based on the base station density of the initial trajectory points, the process noise variance of the location is determined, and the process noise variance of the location is negatively correlated with the base station density. The process noise covariance is determined based on the process noise variance at the stated position, the process noise variance at the preset speed, and the process noise variance at the preset direction angle.

8. The method according to claim 1, characterized in that, The optimized trajectory includes multiple optimized trajectory points, each optimized trajectory point corresponding to an initial trajectory point, and the method further includes: For each optimized trajectory point: The road network search radius is determined based on the base station density corresponding to the optimized trajectory point and the speed of the optimized trajectory point. The road network search radius is negatively correlated with the base station density and positively correlated with the speed. The road network search range is determined by taking the location of the optimized trajectory point as the center and the road network search radius as the radius; Road segments within the search area of ​​the road network are identified as candidate road segments; Determine the distance factor, speed factor, and direction factor for each candidate road segment. The distance factor is used to characterize the vertical distance from the position of the optimized trajectory point to the candidate road segment. The speed factor is used to characterize the speed difference between the speed of the optimized trajectory point and the speed limit of the candidate road segment. The direction factor is used to characterize the difference between the driving direction of the optimized trajectory point and the direction of the candidate road segment. Each candidate road segment is scored based on its distance factor, speed factor, and direction factor. The candidate road segment with the highest score is determined as the optimal road segment for the optimized trajectory point; The position of the optimized trajectory point is projected onto the optimal road segment, and the position of the perpendicular foot of the projection is determined as the position of the road network fitting point corresponding to the optimized trajectory point; Based on the fitting points of each road network, a road network fitting trajectory is generated.

9. The method according to claim 8, characterized in that, The process of scoring each candidate road segment based on its distance factor, speed factor, and direction factor includes: For each candidate road segment: If the base station density corresponding to the optimal trajectory point is greater than a preset threshold, the preset value is determined as the weight adjustment parameter; if the base station density corresponding to the optimal trajectory point is less than or equal to the preset threshold, the weight adjustment parameter is determined according to the base station density, and the weight adjustment parameter is positively correlated with the base station density. Based on the weight adjustment parameter, the distance factor weight, speed factor weight, and direction factor weight are determined, wherein the weight adjustment parameter is positively correlated with the distance factor weight, and negatively correlated with the speed factor weight and the direction factor weight. The distance factor, speed factor, and direction factor of the candidate road segment are weighted and summed according to the distance factor weight, speed factor weight, and direction factor weight to obtain the score of the candidate road segment.

10. A trajectory generation device, characterized in that, The device includes: The generation module is used to generate an initial trajectory of the moving object based on signaling data generated by communication between the moving object and the base station. The initial trajectory includes multiple initial trajectory points. The acquisition module is used to acquire the base station density of each initial trajectory point, wherein the base station density is used to characterize the number of base stations that can cover the initial trajectory point within a unit area. The determination module is used to determine the observation noise covariance and process noise covariance of each initial trajectory point based on the base station density of the initial trajectory point; The filtering module is used to perform Kalman filtering on the initial trajectory based on the observation noise covariance and process noise covariance of each initial trajectory point to obtain the optimized trajectory of the moving object.

Images