Trajectory Point Filtering Method and Device, and Storage Medium

Abstract

The present application discloses a trajectory point filtering method, device, and storage medium, belonging to the field of intelligent transportation. The method includes: obtaining n types of target calculation data in the positioning data collected by m acquisition sources at the same target trajectory point, where the n types of target calculation data include at least one positioning calculation data in the positioning data collected by each acquisition source, the target trajectory point is any trajectory point during vehicle driving, n≥m≥2, and both m and n are integers; scoring the target trajectory point according to a data scoring model and the n types of target calculation data to obtain a scoring value of the target trajectory point; and filtering the target trajectory point according to a data filtering model and the scoring value of the target trajectory point. The present application helps to improve the filtering effect of trajectory points.

Description

Technical Field

[0001] The present application relates to the field of intelligent transportation, and particularly to a method and device for filtering trajectory points and a storage medium. Background Art

[0002] With the rapid development of Internet technology and communication technology, crowd sourcing technology has emerged. Crowd sourcing technology has the advantages of rich data, easy collection, and low cost, and is widely used in location-based services such as map drawing, navigation, and vehicle travel trajectory restoration.

[0003] Currently, in the service of restoring vehicle travel trajectories based on crowd sourcing technology, positioning data of a vehicle at multiple trajectory points can be collected by multiple collection sources, and eligible trajectory points can be filtered out from the multiple trajectory points according to the positioning data of the vehicle collected by each collection source. The trajectory points filtered out according to the positioning data collected by the multiple collection sources are fused as the finally filtered trajectory points, and the vehicle travel trajectory is restored based on the finally filtered trajectory points.

[0004] However, the current trajectory point filtering scheme is difficult to comprehensively consider filtering trajectory points from the perspectives of different collection sources, so the filtering effect is poor. Summary of the Invention

[0005] The present application provides a method and device for filtering trajectory points and a storage medium, which helps to improve the filtering effect of trajectory points.

[0006] The technical solution of the present application is as follows:

[0007] In a first aspect, a method for filtering trajectory points is provided. The method includes: obtaining n types of target calculation data in the positioning data collected by m collection sources at the same target trajectory point, where the n types of target calculation data include at least one positioning calculation data in the positioning data collected by each collection source, the target trajectory point is any trajectory point during the vehicle travel, n≥m≥2, and both m and n are integers; scoring the target trajectory point according to a data scoring model and the n types of target calculation data to obtain a scoring value of the target trajectory point; and filtering the target trajectory point according to a data filtering model and the scoring value of the target trajectory point.

[0008] In the trajectory point filtering scheme provided by the embodiment of the present application, the server filters the target trajectory point according to the target calculation data of the target trajectory point. The n types of target calculation data include at least one positioning calculation data in the positioning data collected by each of the m collection sources at the target trajectory point. Therefore, the trajectory point filtering scheme can comprehensively consider filtering trajectory points from the perspectives of different collection sources, which helps to improve the filtering effect.

[0009] Optionally, the data scoring model is obtained by updating the parameter values of n model parameters in the initial scoring model according to the target parameter values of the n model parameters. The initial scoring model includes n model variables and the n model parameters corresponding to the n model variables, and the target parameter values of the n model parameters are determined according to the sample data.

[0010] The trajectory point filtering solution provided by the embodiment of the present application determines the target parameter values of the n model parameters according to the sample data, and updates the parameter values of the n model parameters in the initial scoring model according to the target parameter values of the n model parameters, so as to facilitate the training of the data scoring model.

[0011] Optionally, the sample data includes n types of sample solution data corresponding to n types of target solution data; the target parameter values of the n model parameters are determined through the following steps: determining the conflict quantification value between the i-th type of sample solution data and the j-th type of sample solution data, where the conflict quantification value is used to characterize the conflict between the i-th type of sample solution data and the j-th type of sample solution data, 1 ≤ i ≤ n, 1 ≤ j ≤ n, i ≠ j, and both i and j are integers; according to the conflict quantification value between the i-th type of sample solution data and the n types of sample solution data, and the score variance corresponding to the i-th type of sample solution data, determining the discrimination quantification value of the i-th type of sample solution data, where the discrimination quantification value is used to characterize the discrimination of the i-th type of sample solution data; according to the discrimination quantification value of the i-th type of sample solution data and the discrimination quantification values of the n types of sample solution data, determining the target parameter value of the i-th model parameter among the n model parameters.

[0012] The trajectory point filtering solution provided by the embodiment of the present application determines the target parameter values of the model parameters according to the conflict and discrimination between the positioning solution data, and can determine the target parameter values of the model parameters without any prior knowledge, which is convenient for training the data scoring model.

[0013] Optionally, the sample data includes n types of sample calculation data for each of the k sample trajectory points. The n types of sample calculation data for each sample trajectory point include at least one positioning calculation data collected by each of the m acquisition sources at the sample trajectory point. Each type of sample calculation data for each sample trajectory point corresponds to a sample score. The variance of the scores corresponding to the i-th type of sample calculation data is the variance of the normalized scores corresponding to the i-th type of sample calculation data of the k sample trajectory points. k≥2 and k is an integer. Determining the conflict quantification value between the i-th type of sample calculation data and the j-th type of sample calculation data includes: using the conflict quantification formula to determine the conflict quantification value between the i-th type of sample calculation data and the j-th type of sample calculation data. According to the conflict quantification value between the i-th type of sample calculation data and the n types of sample calculation data, and the variance of the scores corresponding to the i-th type of sample calculation data, determining the discrimination quantification value of the i-th type of sample calculation data includes: according to the conflict quantification value between the i-th type of sample calculation data and the n types of sample calculation data, and the variance of the normalized scores corresponding to the i-th type of sample calculation data, using the discrimination quantification formula to determine the discrimination quantification value of the i-th type of sample calculation data. According to the discrimination quantification value of the i-th type of sample calculation data and the discrimination quantification values of the n types of sample calculation data, determining the target parameter value of the i-th model parameter among the n model parameters includes: according to the discrimination quantification value of the i-th type of sample calculation data and the discrimination quantification values of the n types of sample calculation data, using the model parameter formula to determine the target parameter value of the i-th model parameter among the n model parameters;

[0014] Wherein, the conflict quantification formula is The discrimination quantification formula is The model parameter formula is: r ij represents the conflict quantification value between the i-th type of sample calculation data and the j-th type of sample calculation data, q = 1, 2, 3... k, d q represents the difference between the normalized score corresponding to the i-th type of sample calculation data of the q-th sample trajectory point among the k sample trajectory points and the normalized score corresponding to the j-th type of sample calculation data of the q-th sample trajectory point, C i represents the discrimination quantification value of the i-th type of sample calculation data, σ i represents the variance of the normalized scores corresponding to the i-th type of sample calculation data, w i represents the i-th model parameter, C j represents the discrimination quantification value of the j-th type of sample calculation data, j = 1, 2, 3... n.

[0015] Optionally, the normalized score corresponding to the i-th sample solution data is obtained by normalizing the sample score corresponding to the i-th sample solution data, and the normalized score corresponding to the j-th sample solution data is obtained by normalizing the sample score corresponding to the j-th sample solution data.

[0016] The trajectory point filtering solution provided by the embodiments of the present application can facilitate mapping the sample solution data to values within the range of [0, 1] by normalizing the sample scores corresponding to the sample solution data, thereby facilitating the calculation of the conflict quantification value between the sample solution data.

[0017] Optionally, the normalized score corresponding to the i-th sample solution data is obtained by normalizing the sample score corresponding to the i-th sample solution data using a normalization formula, and the normalized score corresponding to the j-th sample solution data is obtained by normalizing the sample score corresponding to the j-th sample solution data using the normalization formula;

[0018] where the normalization formula is x ap represents the normalized score of the a-th sample score among the k sample scores corresponding to the p-th sample solution data among the n sample solution data, f p (a) represents the a-th sample score among the k sample scores corresponding to the p-th sample solution data, f p * represents the optimal score among the k sample scores corresponding to the p-th sample solution data, f p* represents the worst score among the k sample scores corresponding to the p-th sample solution data, p = i or j, 1 ≤ a ≤ k, and a is an integer.

[0019] Optionally, the n model variables of the data scoring model are score variables corresponding to n types of target solution data; scoring the target trajectory point according to the data scoring model and the n types of target solution data in the positioning data collected by m acquisition sources at the target trajectory point to obtain the scoring score of the target trajectory point, including: determining the scores corresponding to the n types of target solution data; scoring the target trajectory point according to the data scoring model and the scores corresponding to the n types of target solution data to obtain the scoring score of the target trajectory point.

[0020] The trajectory point filtering solution provided by the embodiments of the present application enables the server to determine the scores corresponding to the target solution data, which can facilitate scoring the target trajectory point using the data scoring model according to the scores corresponding to the n types of target solution data of the target trajectory point.

[0021] Optionally, the data filtering model is obtained by updating the parameter value of the threshold parameter in the initial filtering model according to the target parameter value of the threshold parameter. The initial filtering model includes model variables and the threshold parameter corresponding to the model variables, and the target parameter value of the threshold parameter is determined according to the sample data.

[0022] The trajectory point filtering solution provided by the embodiments of the present application determines the target parameter value of the threshold parameter according to the sample data, and updates the parameter value of the threshold parameter in the initial filtering model according to the target parameter value of the threshold parameter, so as to facilitate the generation of the data filtering model.

[0023] Optionally, the sample data includes n types of sample calculation data of each of the k sample trajectory points. The n types of sample calculation data of each sample trajectory point include at least one positioning calculation data in the positioning data collected by each of the m acquisition sources at the sample trajectory point. k≥2, and k is an integer. The target parameter value of the threshold parameter is determined through the following steps: for each of the k sample trajectory points, score the sample trajectory point according to the data scoring model and the n types of sample calculation data of the sample trajectory point to obtain the scoring value of the sample trajectory point; according to the scoring values of the k sample trajectory points, determine the target parameter value of the threshold parameter.

[0024] The trajectory point filtering solution provided by the embodiments of the present application scores the sample trajectory point according to the data scoring model and the n types of sample calculation data of the sample trajectory point, which can facilitate determining the target parameter value of the threshold parameter according to the scoring values of the k sample trajectory points.

[0025] Optionally, each type of sample calculation data of each sample trajectory point corresponds to a sample score. For each of the k sample trajectory points, scoring the sample trajectory point according to the data scoring model and the n types of sample calculation data of the sample trajectory point to obtain the scoring value of the sample trajectory point includes: for each of the k sample trajectory points, scoring the sample trajectory point according to the data scoring model and the sample scores corresponding to the n types of sample calculation data of the sample trajectory point to obtain the scoring value of the sample trajectory point.

[0026] The trajectory point filtering solution provided by the embodiments of the present application enables the server to score the sample trajectory point according to the data scoring model and the sample scores corresponding to the n types of sample calculation data of the sample trajectory point, which can facilitate the server to determine the target parameter value of the threshold parameter according to the scoring values of the k sample trajectory points.

[0027] In a second aspect, a trajectory point filtering device is provided, including: each module for executing the trajectory point filtering method provided in the first aspect or any optional implementation manner of the first aspect.

[0028] In a third aspect, a trajectory point filtering device is provided, including: a processor and a memory. A program is stored in the memory, and the processor is configured to call the program stored in the memory, so that the trajectory point filtering device executes the trajectory point filtering method provided in the first aspect or any optional implementation manner of the first aspect.

[0029] In a fourth aspect, a computer-readable storage medium is provided. A computer program is stored in the computer-readable storage medium. When the computer program runs on a computer, the computer is caused to execute the trajectory point filtering method provided in the first aspect or any optional implementation manner of the first aspect.

[0030] In a fifth aspect, a computer program product including instructions is provided. When the computer program product runs on a computer, the computer is caused to execute the trajectory point filtering method provided in the first aspect or any optional manner of the first aspect.

[0031] In a sixth aspect, a chip is provided. The chip includes a programmable logic circuit and / or program instructions, and is configured to implement the trajectory point filtering method provided in the first aspect or any optional manner of the first aspect when running.

[0032] In a seventh aspect, a method for training a data scoring model is provided. The method includes:

[0033] Obtaining sample data and an initial scoring model, where the initial scoring model includes n model variables and n model parameters corresponding to the n model variables, n≥2, and n is an integer;

[0034] Determining target parameter values of the n model parameters according to the sample data;

[0035] Updating parameter values of the n model parameters in the initial scoring model according to the target parameter values of the n model parameters to obtain a data scoring model.

[0036] In the training solution of the data scoring model provided by the embodiments of the present application, after obtaining sample data and an initial scoring model, determining target parameter values of n model parameters according to the sample data, and updating parameter values of the n model parameters in the initial scoring model according to the target parameter values of the n model parameters can facilitate obtaining a data scoring model.

[0037] Optionally, the sample data includes n kinds of sample solution data corresponding to n kinds of target solution data;

[0038] The determining the target parameter values of the n model parameters according to the sample data includes:

[0039] Determine the conflict quantification value between the i-th sample solution data and the j-th sample solution data, where the conflict quantification value is used to characterize the conflict between the i-th sample solution data and the j-th sample solution data, 1 ≤ i ≤ n, 1 ≤ j ≤ n, i ≠ j, and both i and j are integers;

[0040] According to the conflict quantification value between the i-th sample solution data and the n sample solution data, and the score variance corresponding to the i-th sample solution data, determine the discrimination quantification value of the i-th sample solution data, where the discrimination quantification value is used to characterize the discrimination of the i-th sample solution data;

[0041] According to the discrimination quantification value of the i-th sample solution data and the discrimination quantification values of the n sample solution data, determine the target parameter value of the i-th model parameter among the n model parameters.

[0042] The training scheme of the data scoring model provided by the embodiments of the present application determines the target parameter value of the model parameter according to the conflict and discrimination between the positioning solution data, and can determine the target parameter value of the model parameter without any prior knowledge, which is convenient for training the data scoring model.

[0043] Optionally, the sample data includes the n sample solution data of each of the k sample trajectory points, and the n sample solution data of each sample trajectory point includes at least one positioning solution data in the positioning data collected by each of the m acquisition sources at the sample trajectory point. Each sample solution data of each sample trajectory point corresponds to a sample score, and the score variance corresponding to the i-th sample solution data is the variance of the normalized scores corresponding to the i-th sample solution data of the k sample trajectory points, k ≥ 2, n ≥ m ≥ 2, and both k and m are integers;

[0044] The determination of the conflict quantification value between the i-th sample solution data and the j-th sample solution data includes:

[0045] Use the conflict quantification formula to determine the conflict quantification value between the i-th sample solution data and the j-th sample solution data;

[0046] The determination of the discrimination quantification value of the i-th sample solution data according to the conflict quantification value between the i-th sample solution data and the n sample solution data, and the score variance corresponding to the i-th sample solution data includes:

[0047] Determine the discrimination quantization value of the i-th sample solution data according to the conflict quantization value between the i-th sample solution data and the n sample solution data, and the variance of the normalized scores corresponding to the i-th sample solution data, using the discrimination quantization formula;

[0048] Determining the target parameter value of the i-th model parameter among the n model parameters according to the discrimination quantization value of the i-th sample solution data and the discrimination quantization values of the n sample solution data includes:

[0049] According to the discrimination quantization value of the i-th sample solution data and the discrimination quantization values of the n sample solution data, use the model parameter formula to determine the target parameter value of the i-th model parameter among the n model parameters;

[0050] Among them, the conflict quantization formula is

[0051] The discrimination quantization formula is

[0052] The model parameter formula is:

[0053] r ij represents the conflict quantization value between the i-th sample solution data and the j-th sample solution data, q = 1, 2, 3... k, d q represents the difference between the normalized score corresponding to the i-th sample solution data and the normalized score corresponding to the j-th sample solution data of the q-th sample trajectory point among the k sample trajectory points, C i represents the discrimination quantization value of the i-th sample solution data, σ i represents the variance of the normalized scores corresponding to the i-th sample solution data, w i represents the i-th model parameter, C j represents the discrimination quantization value of the j-th sample solution data, j = 1, 2, 3... n.

[0054] Optionally, before determining the conflict quantization value between the i-th sample solution data and the j-th sample solution data using the conflict quantization formula, the method further includes: respectively performing normalization processing on the sample scores corresponding to the i-th sample solution data and the sample scores corresponding to the j-th sample solution data to obtain the normalized scores corresponding to the i-th sample solution data and the normalized scores corresponding to the j-th sample solution data.

[0055] The training scheme of the data scoring model provided by the embodiments of the present application can facilitate mapping the sample solution data into values within the range of [0, 1] by normalizing the sample scores corresponding to the sample solution data, so as to facilitate calculating the conflict quantification values between the sample solution data.

[0056] Optionally, the normalizing the sample scores corresponding to the i-th type of sample solution data and the sample scores corresponding to the j-th type of sample solution data respectively includes: normalizing the sample scores corresponding to the i-th type of sample solution data and the sample scores corresponding to the j-th type of sample solution data respectively by using a normalization formula;

[0057] wherein, the normalization formula is x ap represents the normalized score of the a-th sample score among the k sample scores corresponding to the p-th sample solution data among the n types of sample solution data, f p (a) represents the a-th sample score among the k sample scores corresponding to the p-th sample solution data, f p * represents the optimal score among the k sample scores corresponding to the p-th sample solution data, f p* represents the worst score among the k sample scores corresponding to the p-th sample solution data, p = i or j, 1 ≤ a ≤ k, and a is an integer.

[0058] It should be noted that the training method of the data scoring model provided by the embodiments of the present application can be executed by a server, or by other devices external to the server. The server executing the training method of the data scoring model and the server executing the trajectory point filtering method provided in the first aspect above can be the same server or different servers, and the embodiments of the present application do not make any limitations in this regard.

[0059] In an eighth aspect, a training device for a data scoring model is provided, and the device includes:

[0060] An acquisition module, configured to acquire sample data and an initial scoring model, where the initial scoring model includes n model variables and n model parameters corresponding to the n model variables, n ≥ 2, and n is an integer;

[0061] A determination module, configured to determine target parameter values of the n model parameters according to the sample data;

[0062] An update module, configured to update parameter values of the n model parameters in the initial scoring model according to the target parameter values of the n model parameters to obtain a data scoring model.

[0063] Optionally, the sample data includes n types of sample solution data corresponding to n types of target solution data;

[0064] The determining module is configured to:

[0065] Determine a conflict quantification value between the i-th sample solution data and the j-th sample solution data, where the conflict quantification value is used to characterize the conflict between the i-th sample solution data and the j-th sample solution data, 1 ≤ i ≤ n, 1 ≤ j ≤ n, i ≠ j, and both i and j are integers;

[0066] According to the conflict quantification value between the i-th sample solution data and the n sample solution data, and the score variance corresponding to the i-th sample solution data, determine a discrimination quantification value of the i-th sample solution data, where the discrimination quantification value is used to characterize the discrimination of the i-th sample solution data;

[0067] According to the discrimination quantification value of the i-th sample solution data and the discrimination quantification values of the n sample solution data, determine a target parameter value of the i-th model parameter among the n model parameters.

[0068] Optionally, the sample data includes the n sample solution data of each of the k sample trajectory points, the n sample solution data of each sample trajectory point includes at least one positioning solution data in the positioning data collected by each of the m acquisition sources at the sample trajectory point, each sample solution data of each sample trajectory point corresponds to a sample score, and the score variance corresponding to the i-th sample solution data is the variance of the normalized scores corresponding to the i-th sample solution data of the k sample trajectory points, k ≥ 2, n ≥ m ≥ 2, and both k and m are integers;

[0069] The determining module is configured to:

[0070] Use a conflict quantification formula to determine the conflict quantification value between the i-th sample solution data and the j-th sample solution data;

[0071] According to the conflict quantification value between the i-th sample solution data and the n sample solution data, and the variance of the normalized scores corresponding to the i-th sample solution data, use a discrimination quantification formula to determine the discrimination quantification value of the i-th sample solution data;

[0072] According to the discrimination quantification value of the i-th sample solution data and the discrimination quantification values of the n sample solution data, use a model parameter formula to determine the target parameter value of the i-th model parameter among the n model parameters;

[0073] Wherein, the conflict quantification formula is

[0074] The discrimination force quantization formula is

[0075] The model parameter formula is:

[0076] r ij represents the conflict quantization value between the i-th sample solution data and the j-th sample solution data, q = 1, 2, 3... k, d q represents the difference between the normalized score corresponding to the i-th sample solution data of the q-th sample trajectory point among the k sample trajectory points and the normalized score corresponding to the j-th sample solution data of the q-th sample trajectory point, C i represents the discrimination force quantization value of the i-th sample solution data, σ i represents the variance of the normalized score corresponding to the i-th sample solution data, w i represents the i-th model parameter, C j represents the discrimination force quantization value of the j-th sample solution data, j = 1, 2, 3... n.

[0077] Optionally, the device further includes: a normalization module, configured to perform normalization processing on the sample score corresponding to the i-th sample solution data and the sample score corresponding to the j-th sample solution data respectively, to obtain the normalized score corresponding to the i-th sample solution data and the normalized score corresponding to the j-th sample solution data.

[0078] Optionally, the normalization module is configured to perform normalization processing on the sample score corresponding to the i-th sample solution data and the corresponding sample score of the j-th sample solution data respectively by using a normalization formula;

[0079] wherein, the normalization formula is x ap represents the normalized score of the a-th sample score among the k sample scores corresponding to the p-th sample solution data among the n sample solution data, f p (a) represents the a-th sample score among the k sample scores corresponding to the p-th sample solution data, f p * represents the optimal score among the k sample scores corresponding to the p-th sample solution data, f p* represents the worst score among the k sample scores corresponding to the p-th sample solution data, p = i or j, 1 ≤ a ≤ k, and a is an integer.

[0080] It should be noted that the technical effects of the eighth aspect can refer to the seventh aspect, and are not elaborated herein in the embodiments of the present application.

[0081] In a ninth aspect, a training device for a data scoring model is provided, including: a processor and a memory. A program is stored in the memory, and the processor is configured to call the program stored in the memory, so that the training device for the data scoring model executes the training method for the data scoring model provided in the seventh aspect or any optional implementation manner of the seventh aspect.

[0082] In a tenth aspect, a computer-readable storage medium is provided. A computer program is stored in the computer-readable storage medium. When the computer program runs on a computer, the computer is caused to execute the training method for the data scoring model provided in the seventh aspect or any optional implementation manner of the seventh aspect.

[0083] In an eleventh aspect, a computer program product including instructions is provided. When the computer program product runs on a computer, the computer is caused to execute the training method for the data scoring model provided in the seventh aspect or any optional manner of the seventh aspect.

[0084] In a twelfth aspect, a chip is provided. The chip includes a programmable logic circuit and / or program instructions, and is configured to implement the training method for the data scoring model provided in the seventh aspect or any optional manner of the seventh aspect when the chip runs.

[0085] In a thirteenth aspect, a training method for a data filtering model is provided. The method includes:

[0086] Obtaining sample data and an initial filtering model, where the initial filtering model includes model variables and threshold parameters corresponding to the model variables;

[0087] Determining a target parameter value of the threshold parameter according to the sample data;

[0088] Updating a parameter value of the threshold parameter in the initial filtering model according to the target parameter value of the threshold parameter to obtain a data filtering model.

[0089] In the training solution for the data filtering model provided by the embodiments of the present application, after obtaining the sample data and the initial filtering model, determining the target parameter value of the threshold parameter according to the sample data, and updating the parameter value of the threshold parameter in the initial filtering model according to the target parameter value of the threshold parameter can facilitate obtaining the data filtering model.

[0090] Optionally, the sample data includes n types of sample solution data of each sample trajectory point among k sample trajectory points. The n types of sample solution data of each sample trajectory point include at least one positioning solution data in the positioning data collected by each of m acquisition sources at the sample trajectory point. k≥2, n≥m≥2, and k, n, and m are all integers;

[0091] Determining a target parameter value of the threshold parameter according to the sample data includes:

[0092] For each of the k sample trajectory points, score the sample trajectory point according to a data scoring model and the n types of sample solution data of the sample trajectory point to obtain a scoring value of the sample trajectory point;

[0093] Determine the target parameter value of the threshold parameter according to the scoring values of the k sample trajectory points.

[0094] The training scheme of the data filtering model provided by the embodiments of the present application can score the sample trajectory point according to the data scoring model and the n types of sample solution data of the sample trajectory point, which is convenient for determining the target parameter value of the threshold parameter according to the scoring values of the k sample trajectory points.

[0095] Optionally, each type of sample solution data of each sample trajectory point corresponds to a sample score value. For each of the k sample trajectory points, scoring the sample trajectory point according to the data scoring model and the n types of sample solution data of the sample trajectory point to obtain the scoring value of the sample trajectory point includes: for each of the k sample trajectory points, score the sample trajectory point according to the sample score values corresponding to the n types of sample solution data of the sample trajectory point to obtain the scoring value of the sample trajectory point.

[0096] The training scheme of the data filtering model provided by the embodiments of the present application can score the sample trajectory point according to the sample score values corresponding to the n types of sample solution data of the sample trajectory point, which is convenient for determining the target parameter value of the threshold parameter according to the scoring values of the k sample trajectory points.

[0097] It should be noted that the training method of the data filtering model provided by the embodiments of the present application can be executed by a server or by other devices outside the server. The server executing the training method of the data filtering model and the server executing the trajectory point filtering method provided in the first aspect above and the server executing the training method of the data scoring model provided in the seventh aspect above can be the same server or different servers. The embodiments of the present application do not limit this.

[0098] In a fourteenth aspect, a training device for a data filtering model is provided. The device includes:

[0099] An acquisition module, configured to acquire sample data and an initial filtering model, where the initial filtering model includes model variables and threshold parameters corresponding to the model variables;

[0100] A determination module, configured to determine a target parameter value of the threshold parameter according to the sample data;

[0101] An update module, configured to update a parameter value of the threshold parameter in the initial filtering model according to the target parameter value of the threshold parameter, so as to obtain a data filtering model.

[0102] Optionally, the sample data includes n types of sample solution data of each of the k sample trajectory points, and the n types of sample solution data of each sample trajectory point include at least one positioning solution data in the positioning data collected by each of the m acquisition sources at the sample trajectory point, k≥2, n≥m≥2, and k, n, and m are all integers;

[0103] The determination module is configured to:

[0104] For each of the k sample trajectory points, score the sample trajectory point according to a data scoring model and the n types of sample solution data of the sample trajectory point, so as to obtain a scoring value of the sample trajectory point;

[0105] Determine the target parameter value of the threshold parameter according to the scoring values of the k sample trajectory points.

[0106] Optionally, each type of sample solution data of each sample trajectory point corresponds to a sample score value,

[0107] The determination module is configured to score each of the k sample trajectory points according to the data scoring model and the sample score values corresponding to the n types of sample solution data of the sample trajectory point, so as to obtain a scoring value of the sample trajectory point.

[0108] It should be noted that the technical effects of the fourteenth aspect can refer to the thirteenth aspect, and the embodiments of the present application will not be elaborated herein.

[0109] In a fifteenth aspect, there is provided a training device for a data filtering model, including: a processor and a memory, where a program is stored in the memory, and the processor is configured to call the program stored in the memory, so that the training device for the data filtering model executes the data filtering model training method provided in the thirteenth aspect or any optional implementation manner of the thirteenth aspect.

[0110] In a sixteenth aspect, there is provided a computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, and when the computer program runs on a computer, the computer is enabled to execute the data filtering model training method provided in the thirteenth aspect or any optional implementation manner of the thirteenth aspect.

[0111] In a seventeenth aspect, there is provided a computer program product containing instructions, which, when the computer program product runs on a computer, cause the computer to execute the method for training a data filtering model provided in the thirteenth aspect or any optional implementation manner of the thirteenth aspect.

[0112] In an eighteenth aspect, there is provided a chip, which includes a programmable logic circuit and / or program instructions, and is used to implement the method for training a data filtering model provided in the thirteenth aspect or any optional implementation manner of the thirteenth aspect when the chip runs.

[0113] The beneficial effects brought by the technical solution provided in the embodiments of this application are as follows:

[0114] For the trajectory point filtering method, device, and storage medium provided in the embodiments of this application, after the server obtains n types of target solution data in the positioning data collected by m collection sources at the same target trajectory point, it scores the target trajectory point according to the data scoring model and the n types of target solution data to obtain the scoring value of the target trajectory point, and filters the target trajectory point according to the data filtering model and the scoring value of the target trajectory point. The n types of target solution data include at least one positioning solution data in the positioning data collected by each collection source. This trajectory point filtering solution can comprehensively consider filtering trajectory points from the perspectives of different collection sources, which helps to improve the filtering effect. Description of the Drawings

[0115] Figure 1 It is a schematic diagram of an implementation environment related to each embodiment of this application;

[0116] Figure 2 It is a flowchart of a method for training a data scoring model provided in the embodiments of this application;

[0117] Figure 3 It is a flowchart of a method for determining the target parameter value of the i-th model parameter according to sample data provided in the embodiments of this application;

[0118] Figure 4 It is a flowchart of a method for training a data filtering model provided in the embodiments of this application;

[0119] Figure 5 It is a flowchart of a method for determining the target parameter value of a threshold parameter according to sample data provided in the embodiments of this application;

[0120] Figure 6 It is a flowchart of a method for filtering trajectory points provided in the embodiments of this application;

[0121] Figure 7 It is a flowchart of a method for scoring a target trajectory point according to a data scoring model provided in the embodiments of this application;

[0122] Figure 8 It is a schematic logical structure diagram of a trajectory point filtering device provided by an embodiment of the present application;

[0123] Figure 9 It is a schematic logical structure diagram of a training device for a data scoring model provided by an embodiment of the present application;

[0124] Figure 10 It is a schematic logical structure diagram of a training device for a data filtering model provided by an embodiment of the present application;

[0125] Figure 11 It is a schematic hardware structure diagram of a processing device provided by an embodiment of the present application;

[0126] Figure 12 It is a schematic hardware structure diagram of another processing device provided by an embodiment of the present application. Detailed implementation manners

[0127] To make the principles, technical solutions, and advantages of the present application clearer, the embodiments of the present application will be further described in detail below with reference to the accompanying drawings.

[0128] The trajectory point filtering solution provided by the embodiments of the present application can be applied to location-based services and is a basic solution for location-based services. Such location-based services include, for example, but are not limited to, map drawing, navigation, vehicle travel trajectory restoration, etc. The trajectory point filtering solution can be applied to a trajectory point filtering system, which can include a server and at least one vehicle. The at least one vehicle can be communicatively connected to the server, and at least two acquisition sources can be deployed on each vehicle. Each acquisition source can collect the positioning data of the vehicle where it is located. During the driving process of each vehicle, the at least two acquisition sources on the vehicle can collect the positioning data of the trajectory points of the vehicle, and the vehicle can report the positioning data of the trajectory points collected by the at least two acquisition sources to the server. The server filters the trajectory points of the vehicle according to the positioning data reported by the vehicle.

[0129] Among them, the above-mentioned server can be a single server, or a server cluster composed of several servers, or a cloud computing service center. Optionally, the above-mentioned server can include a data center server and at least one edge server. Each edge server can be communicatively connected to the data center server. Each edge server is responsible for a region. Each vehicle is communicatively connected to the edge server responsible for its region. Each vehicle can report the positioning data of the trajectory points collected by at least two collection sources of the vehicle to the edge server responsible for its region. After the edge server converges the positioning data of the trajectory points reported by the vehicles in the region, it reports them to the data center server uniformly, which can reduce the pressure on the data center server. Especially in the crowdsourcing mode, the pressure on the data center server can be reduced.

[0130] Exemplarily, please refer to Figure 1, which shows a schematic diagram of an implementation environment involved in various embodiments of the present application. The trajectory point filtering system provided by this implementation environment includes a data center server 001, edge servers 011 - 013 (i.e., edge server 011, edge server 012, and edge server 013), and vehicles 021 - 026 (i.e., vehicle 021, vehicle 022, vehicle 023, vehicle 024, vehicle 025, and vehicle 026). The edge servers 011 - 013 are respectively communicatively connected to the data center server 001, and each of the edge servers 011 - 013 is responsible for a region. Vehicles 021 - 022 are located in the region responsible for by edge server 011, and vehicles 021 - 022 are respectively communicatively connected to edge server 011. Each vehicle among vehicles 021 - 022 can send the positioning data of the trajectory points of this vehicle collected by at least two collection sources on this vehicle to edge server 011. After the edge server 011 aggregates the positioning data sent by vehicles 021 - 022, it uniformly reports to the data center server 001. Vehicles 023 - 024 are located in the region responsible for by edge server 012, and vehicles 023 - 024 are respectively communicatively connected to edge server 012. Each vehicle among vehicles 023 - 024 can send the positioning data of the trajectory points of this vehicle collected by at least two collection sources on this vehicle to edge server 012. After the edge server 012 aggregates the positioning data sent by vehicles 023 - 024, it uniformly reports to the data center server 001. Vehicles 025 - 026 are located in the region responsible for by edge server 013, and vehicles 025 - 026 are respectively communicatively connected to edge server 013. Each vehicle among vehicles 025 - 026 can send the positioning data of the trajectory points of this vehicle collected by at least two collection sources on this vehicle to edge server 013. After the edge server 013 aggregates the positioning data sent by vehicles 025 - 026, it uniformly reports to the data center server 001. The data center server 001 can filter the trajectory points of each vehicle according to the positioning data reported by vehicles 021 - 026.

[0131] Among them, the above communication connection may include a wired connection or a wireless connection. The wired connection may include, but is not limited to, a universal serial bus (USB) connection. The wireless connection may include, but is not limited to, a wireless fidelity (WIFI) connection, a Bluetooth connection, an infrared connection, a ZigBee connection, etc. Exemplarily, the communication connection between the edge server and the data center server may be a wired connection or a wireless connection. The communication connection between the vehicle and the edge server may be a wireless connection, for example, a WIFI connection. The above at least two acquisition sources may include any positioning function-equipped devices such as a global positioning system (GPS) device, a camera, and a laser positioning device. There is heterogeneity in the positioning data collected by the at least two acquisition sources, that is, the structures (or formats) of the positioning data collected by the at least two acquisition sources are different.

[0132] Due to the heterogeneity of the positioning data collected by different acquisition sources, in the current trajectory point filtering scheme, for the positioning data of the trajectory points collected by at least two acquisition sources reported by each vehicle, the server filters the trajectory points of the vehicle according to the positioning data collected by each acquisition source, and then fuses the trajectory points filtered according to the positioning data collected by the at least two acquisition sources as the finally filtered trajectory points. For example, the data center server 001 filters the trajectory points 1 to 10 (that is, trajectory point 1, trajectory point 2, trajectory point 3, trajectory point 4, trajectory point 5, trajectory point 6, trajectory point 7, trajectory point 8, trajectory point 9, and trajectory point 10) collected by the GPS device of vehicle 021 according to the positioning data of these trajectory points 1 to 10, and filters out the qualified trajectory points 1, 2, 3, 7, 8, and 10. According to the positioning data of the trajectory points 1 to 10 collected by the camera of vehicle 021, the trajectory points 1 to 10 are filtered, and the qualified trajectory points 1, 2, 3, 5, 8, and 9 are filtered out. The data center server 001 fuses the trajectory points filtered according to the positioning data collected by the GPS device of vehicle 021 and the trajectory points filtered according to the positioning data collected by the camera of vehicle 021 to obtain the finally filtered trajectory points as: trajectory point 1, trajectory point 2, trajectory point 3, trajectory point 5, trajectory point 7, trajectory point 8, trajectory point 9, and trajectory point 10. However, this trajectory point filtering scheme is difficult to comprehensively consider filtering trajectory points from the perspective of different acquisition sources, with low filtering efficiency and poor filtering effect.

[0133] In the trajectory point filtering solution provided by the embodiments of the present application, the positioning data collected by each collection source includes at least one type of positioning calculation data. For each vehicle, the server can obtain at least two types of target calculation data in the positioning data collected by at least two collection sources of the vehicle at any trajectory point, score the trajectory point according to the data scoring model and the at least two types of target calculation data in the positioning data collected by the at least two collection sources at the trajectory point to obtain the scoring value of the trajectory point, and filter the trajectory point according to the data filtering model and the scoring value of the trajectory point. Among them, the at least two types of target calculation data include at least one type of positioning calculation data in the positioning data collected by each of the at least two collection sources. The trajectory point filtering solution can comprehensively consider filtering trajectory points from the perspective of different collection sources, with higher filtering efficiency and better filtering effect.

[0134] It is easy for those skilled in the art to understand that Figure 1 the illustrated implementation environment is only for example and is not used to limit the technical solutions of the embodiments of the present application. During implementation, the number of edge servers, the number of data center servers, and the number of vehicles can be configured according to needs. Moreover, the vehicles can be communicatively connected to the data center servers, so that the edge servers can also not be configured. The embodiments of the present application do not make any limitations in this regard.

[0135] It should be noted that when filtering trajectory points using the trajectory point filtering solution provided by the embodiments of the present application, a data scoring model and a data filtering model are required. The data scoring model and the data filtering model can be pre-trained and transplanted into the server that executes the trajectory point filtering solution, or can be trained by the server when executing the trajectory point filtering solution. Whether the data scoring model and the data filtering model are pre-trained and transplanted into the server that executes the trajectory point filtering solution, or are trained by the server when executing the trajectory point filtering solution, it can be understood that the embodiments of the present application involve the trajectory point filtering process, the training process of the data scoring model, and the training process of the data filtering model. The trajectory point filtering process, the training process of the data scoring model, and the training process of the data filtering model can be executed by the same device or different devices. For example, the training process of the data scoring model, the training process of the data filtering model, and the trajectory point filtering process are all executed by Figure 1 the data center server 001 in. The embodiments of the present application take the training process of the data scoring model, the training process of the data filtering model, and the trajectory point filtering process as examples for Figure 1 the illustrated implementation environment, and introduce the training process of the data scoring model, the training process of the data filtering model, and the trajectory point filtering process provided by the embodiments of the present application in three embodiments respectively.

[0136] Exemplarily, please refer to Figure 2, which shows a flowchart of a method for training a data scoring model provided by an embodiment of the present application. This method can be applied to Figure 1 the implementation environment shown. Refer to Figure 2 , this method may include the following steps:

[0137] Step 201, obtain sample data.

[0138] Optionally, the sample data includes n types of sample solution data, and each type of sample solution data is the positioning solution data in the positioning data collected by one acquisition source. Optionally, the sample data may include n types of sample solution data for each of the k sample trajectory points. The n types of sample solution data for each sample trajectory point may include at least one type of positioning solution data in the positioning data collected by each of the m acquisition sources at this sample trajectory point, where n≥m≥2, k≥2, and n, m, and k are all integers. Optionally, among the n types of sample solution data for each sample trajectory point, each type of sample solution data may correspond to a sample score. Among them, the sample solution data may be quantitative data or non-quantitative data. Among the n types of sample solution data, the sample score corresponding to the quantitative sample solution data may be the sample solution data itself (that is, the sample score corresponding to the quantitative sample solution data may be equal to the sample solution data), and the sample score corresponding to the non-quantitative sample solution data may be determined according to the score mapping relationship.

[0139] Optionally, a server (such as data center server 001) may obtain the positioning data of a vehicle (such as vehicle 021) at k sample trajectory points, and obtain the sample data from the positioning data of the vehicle at these k sample trajectory points. Among them, the k sample trajectory points may be located on the same historical driving trajectory of the vehicle, and the historical driving trajectory refers to the driving trajectory of the vehicle during a historical time period. Among them, the positioning data of the vehicle at each of the k sample trajectory points includes m positioning data corresponding to the m acquisition sources of the vehicle (that is, the positioning data of the vehicle at each sample trajectory point is a set of data composed of m positioning data, and the m positioning data correspond to the m acquisition sources). Each of the m positioning data is collected by the corresponding acquisition source at this sample trajectory point. Each of the m positioning data may include at least one type of positioning solution data, and the positioning solution data included in the m positioning data are not all the same. The server obtaining the sample data from the positioning data of the vehicle at the k sample trajectory points may include: for each of the k sample trajectory points, the server extracts n types of sample solution data from the positioning data of this sample trajectory point, so that the server can obtain n types of sample solution data for each of the k sample trajectory points.

[0140] After the server obtains the n types of sample calculation data of each sample trajectory point among the k sample trajectory points, for the n types of sample calculation data of each sample trajectory point, the server can obtain the sample score corresponding to each type of sample calculation data among the n types of sample calculation data. Optionally, the server can maintain multiple score mapping relationships, each score mapping relationship corresponds to a type of positioning calculation data, and each score mapping relationship records the calculation results of multiple positioning calculation data of the corresponding type and the score corresponding to the calculation result of each positioning calculation data. For each type of sample calculation data among the n types of sample calculation data of each sample trajectory point, the server can determine the calculation result of the sample calculation data, and according to the calculation result of the sample calculation data, query the score mapping relationship of the corresponding type, and determine the score corresponding to the calculation result of the sample calculation data in the score mapping relationship as the sample score corresponding to the sample calculation data. Optionally, the calculation result of the positioning calculation data can be the quantization value of the positioning calculation data, and the calculation result of the sample calculation data can be the quantization value of the positioning calculation data.

[0141] In the embodiment of the present application, after the server obtains the n types of sample calculation data of each sample trajectory point among the k sample trajectory points and the sample score corresponding to each sample calculation data, that is, the sample data is obtained. To enable readers to more clearly understand the process of the server obtaining sample data, the process of the server obtaining sample data is illustrated below by way of example.

[0142] Exemplarily, taking k = 5 (that is, 5 sample trajectory points), m = 2 (that is, 2 acquisition sources), and the m acquisition sources being a GPS device and a camera as an example. The positioning data of the vehicle obtained by the server at the k sample trajectory points may include: positioning data 11 collected by the GPS device at sample trajectory point 1 and positioning data 12 collected by the camera at sample trajectory point 1, positioning data 21 collected by the GPS device at sample trajectory point 2 and positioning data 22 collected by the camera at sample trajectory point 2, positioning data 31 collected by the GPS device at sample trajectory point 3 and positioning data 32 collected by the camera at sample trajectory point 3, positioning data 41 collected by the GPS device at sample trajectory point 4 and positioning data 42 collected by the camera at sample trajectory point 4, positioning data 51 collected by the GPS device at sample trajectory point 5 and positioning data 52 collected by the camera at sample trajectory point 5.

[0143] The positioning data of sample trajectory point 1 includes positioning data 11 and positioning data 12, where:

[0144] The positioning data 11 is: $GPGGA,062802.00,3030.366090,N,11425.098182,E,1,10,0.7,20.625,M,0.000,M,,*

[0145] The positioning data 12 is: $CAM,062802.00,18,229,300,0028,0.376*

[0146] The positioning data of sample trajectory point 2 includes positioning data 21 and positioning data 22, where:

[0147] The positioning data 21 is: $GPGGA,062804.00,3030.366090,N,11425.098182,E,1,10,0.8,20.625,M,0.000,M,,*

[0148] The positioning data 22 is: $CAM,062804.00,20,229,300,0120,0.476*

[0149] The positioning data of sample trajectory point 3 includes positioning data 31 and positioning data 32, where:

[0150] The positioning data 31 is: $GPGGA,062806.00,3030.366090,N,11425.098182,E,1,10,0.8,20.625,M,0.000,M,,*

[0151] The positioning data 32 is: $CAM,062806.00,30,243,300,0150,0.876*

[0152] The positioning data of sample trajectory point 4 includes positioning data 41 and positioning data 42, where:

[0153] The positioning data 41 is: $GPGGA,062808.00,3030.366090,N,11425.098182,E,1,10,0.8,20.625,M,0.000,M,,*

[0154] The positioning data 42 is: $CAM,062808.00,31,244,300,0238,0.276*

[0155] The positioning data of sample trajectory point 5 includes positioning data 51 and positioning data 52, where:

[0156] The positioning data 51 is: $GPGGA,062811.00,3030.366090,N,11425.098182,E,1,11,0.8,20.625,M,0.000,M,,*

[0157] The positioning data 52 is: $CAM,062811.00,33,237,300,0143,0.376*

[0158] Among them, the format of the positioning data collected by the GPS device is as follows:

[0159] $GPGGA,<1>,<2>,<3>,<4>,<5>,<6>,<7>,<8>,<9>,M,<10>,M,<11>,<12>*xx <cr> <lf>

[0160] In the positioning data collected by the GPS device:

[0161] $GPGGA indicates the start guide and statement format description, indicating that the positioning data is collected by the GPS device;

[0162] <1> represents the coordinated universal time (UTC) time, in the format of hhmmss.sss;

[0163] <2> represents the latitude, in the format of ddmm.mmmm;

[0164] <3> represents the latitude hemisphere, and the corresponding data is north latitude (N) or south latitude (S);

[0165] <4> represents the longitude, in the format of dddmm.mmmm;

[0166] <5> represents the longitude hemisphere, and the corresponding data is east longitude (E) or west longitude (W);

[0167] <6> represents the GPS status, with its data range from 0 to 8, and the data is an integer. 0 indicates initialization, 1 indicates single-point positioning, 2 indicates code differential, 3 indicates invalid PPS, 4 indicates fixed solution, 5 indicates floating-point solution, 6 indicates being estimated, 7 indicates manually input fixed value, and 8 indicates simulation mode;

[0168] <7> represents the number of satellites in view, with its data range from 00 to 12. The number of satellites in view is positively correlated with the GPS positioning quality. The more satellites in view, the better the GPS positioning quality;

[0169] <8> represents the horizontal dilution of precision (HDOP), with its data range from 0.5 to 99.9;

[0170] <9> represents the altitude, with its data range from -9999.9 to 9999.9 meters;

[0171] M represents the unit of meter;

[0172] <10> represents the geoid height anomaly difference, with its data range from -9999.9 to 9999.9 meters;

[0173] M represents the unit of meter;

[0174] <11> represents the differential GPS data period (RTCM SC-104). If differential positioning is not used, <11> is empty;

[0175] <12> represents the differential reference base station label, and its data range is from 0000 to 1023;

[0176] * represents the statement end flag;

[0177] xx represents the exclusive OR checksum of all ASCII codes from $ to *;

[0178] <cr>Indicates a carriage return character and is the end marker of a statement;

[0179] <lf>Indicates a line break and is the end marker of a statement.

[0180] In the positioning data collected by the GPS device, each of $GPGGA, UTC time, latitude, longitude, GPS status, number of satellites in view, horizontal dilution of precision, altitude, geoid height anomaly difference, differential GPS data period, differential reference base station label, and xx is a positioning solution data.

[0181] Among them, the format of the positioning data collected by the camera is:

[0182] $CAM,<1>,<2>,<3>,<4>,<5>,<6>*xx <cr> <lf>

[0183] In the positioning data collected by the camera:

[0184] $CAM represents the start guide and statement format description, indicating that the positioning data is collected by the camera;

[0185] <1> represents the UTC time, in the format hhmmss.sss;

[0186] <2> represents the X coordinate, in the format xxxx;

[0187] <3> represents the Y coordinate, in the format yyyy;

[0188] <4> represents the Z coordinate, in the format zzzz;

[0189] <5> represents the number of matching feature points, in the format cccc. The number of matching feature points is positively correlated with the camera positioning quality. The more the number of matching feature points, the better the camera positioning quality;

[0190] <6> represents the camera distortion coefficient, in the format x.xxx;

[0191] * represents the statement end flag;

[0192] xx represents the exclusive OR checksum of all ASCII codes from $ to *;

[0193] <cr>Indicates the carriage return character, which is the end marker of a statement;

[0194] <lf>Represents a line break and is the end marker of a statement.

[0195] Among the positioning data collected by the camera, each of $CAM, UTC time, X coordinate, Y coordinate, Z coordinate, number of matched feature points, camera distortion coefficient, and xx is a type of positioning calculation data.

[0196] Exemplarily, taking n = 3 (that is, 3 types of sample calculation data), and the n types of sample calculation data being GPS status, number of satellites in the field of view, and number of matched feature points as an example. The n types of sample calculation data of sample trajectory point 1 extracted by the server from the positioning data of sample trajectory point 1 are respectively: 1 (GPS status), 10 (number of satellites in the field of view), and 0028 (number of matched feature points); the n types of sample calculation data of sample trajectory point 2 extracted from the positioning data of sample trajectory point 2 are respectively: 1 (GPS status), 10 (number of satellites in the field of view), and 0120 (number of matched feature points); the n types of sample calculation data of sample trajectory point 3 extracted from the positioning data of sample trajectory point 3 are respectively: 1 (GPS status), 10 (number of satellites in the field of view), and 0150 (number of matched feature points), the n types of sample calculation data of sample trajectory point 4 extracted from the positioning data of sample trajectory point 4 are respectively: 1 (GPS status), 10 (number of satellites in the field of view), and 0238 (number of matched feature points), the n types of sample calculation data of sample trajectory point 5 extracted from the positioning data of sample trajectory point 5 are respectively: 1 (GPS status), 10 (number of satellites in the field of view), and 0143 (number of matched feature points).

[0197] Exemplarily, the server can maintain at least n types of score mapping relationships, each score mapping relationship corresponding to a type of positioning calculation data. The i-th score mapping relationship maintained by the server (that is, the score mapping relationship corresponding to the i-th type of positioning calculation data) can be as shown in Table 1 below, where Ri represents the i-th type of positioning calculation data (or the calculation result of the i-th type of positioning calculation data), and the score meaning is used to characterize the positioning quality:

[0198] Table 1

[0199] Ri Score Score Meaning 0 10 Precise 1 9 Small Error 2 8 Large Error ... ... ... G 0 Invalid

[0200] For the i-th type of sample calculation data among the n types of sample calculation data of each sample trajectory point, the server can determine the calculation result of this sample calculation data. Assuming the calculation result of this sample calculation data is 1, the server can query the score mapping relationship shown in Table 1 according to the calculation result 1 of this sample calculation data, and determine the score 9 corresponding to the calculation result 1 of this sample calculation data in this score mapping relationship as the sample score corresponding to this sample calculation data.

[0201] Exemplarily, taking the GPS status as the i-th positioning solution data, the score mapping relationship corresponding to the GPS status can be as shown in Table 2 below, where the meaning of GPS is the meaning of the GPS status:

[0202] Table 2

[0203] GPS Status Score GPS Meaning 0 2 Initialization 1 7 Single Point Positioning 2 8 Code Differential 3 6 Invalid PPS 4 10 Fixed Solution 5 9 Float Solution 6 4 Estimating 7 5 Manually Enter Fixed Value 8 3 Simulation Mode

[0204] For the GPS status data of each sample trajectory point, the server can query the score mapping relationship shown in Table 2 according to the GPS status data, and determine the score corresponding to the GPS status data in the score mapping relationship as the sample score corresponding to the GPS status data. For example, the GPS status data of sample trajectory point 1 is 1, and the server can determine that the sample score corresponding to the GPS status data 1 is 7 by querying the score mapping relationship shown in Table 2 according to the GPS status data 1.

[0205] Step 202: Obtain an initial scoring model, where the initial scoring model includes n model variables and n model parameters corresponding to the n model variables.

[0206] Optionally, the initial scoring model can include n model variables and n model parameters corresponding to the n model variables. The n model parameters can be the weight parameters of the n model variables. The n model variables can correspond to n types of target solution data. In the initial scoring model, the parameter values of the n model parameters can be unknowns or initial values.

[0207] Exemplarily, the expression of the initial scoring model can be: S represents the scoring score, n represents the number of model variables, R i represents the i-th model variable among the n model variables of the initial scoring model, w i represents the i-th model parameter among the n model parameters of the initial scoring model. The i-th model parameter is the weight parameter of the i-th model variable. The i-th model variable can correspond to the i-th type of target solution data among the n types of target solution data.

[0208] Step 203: Determine the target parameter values of the n model parameters according to the sample data.

[0209] As described above, the sample data includes n types of sample solution data. Optionally, the n types of sample solution data can correspond one-to-one to n types of target solution data, and the type of each sample solution data is the same as the type of the corresponding target solution data. Optionally, determining the target parameter values of the n model parameters according to the sample data may include: determining the parameter value of each model parameter among the n model parameters according to the sample data. In the embodiment of the present application, taking the determination of the target parameter value of the i-th model parameter as an example, the i-th model parameter is the model parameter corresponding to the i-th type of target solution data, 1≤i≤n, and i is an integer.

[0210] Exemplarily, please refer to Figure 3 , which shows a flowchart of a method for determining the target parameter value of the i-th model parameter according to the sample data provided by the embodiment of the present application. Refer to Figure 3 , and this method may include the following steps:

[0211] Sub-step 2031: Determine the conflict quantification value between the i-th type of sample solution data and the j-th type of sample solution data.

[0212] Wherein, the conflict quantification value is used to characterize the conflict between the i-th type of sample solution data and the j-th type of sample solution data, 1≤i≤n, 1≤j≤n, i≠j, and both i and j are integers.

[0213] As described above, the sample data includes n types of sample solution data of each of the k sample trajectory points, and each type of sample solution data of each sample trajectory point corresponds to a sample score. Optionally, the server can use the conflict quantification formula to determine the conflict quantification value between the i-th type of sample solution data and the j-th type of sample solution data.

[0214] Wherein, the conflict quantification formula can be: r ij represents the conflict quantification value between the i-th type of sample solution data and the j-th type of sample solution data, q = 1, 2, 3... k, and d q represents the difference between the normalized score corresponding to the i-th type of sample solution data of the q-th sample trajectory point among the k sample trajectory points and the normalized score corresponding to the j-th type of sample solution data of the q-th sample trajectory point. According to this conflict quantification formula, it can be understood that r ij represents the conflict quantification value between the i-th type of sample solution data of the k sample trajectory points and the j-th type of sample solution data of the k sample trajectory points, that is, r ij is the comprehensive quantification value of the conflict between the i-th type of sample solution data of the k sample trajectory points and the j-th type of sample solution data of the k sample trajectory points.

[0215] Among them, the normalized score corresponding to the i-th sample calculation data of the q-th sample trajectory point is obtained by normalizing the sample score of the i-th sample calculation data of the q-th sample trajectory point, and the normalized score corresponding to the j-th sample calculation data of the q-th sample trajectory point is obtained by normalizing the sample score of the j-th sample calculation data of the q-th sample trajectory point. The server can first normalize the sample score of the i-th sample calculation data of the q-th sample trajectory point and the sample score of the j-th sample calculation data of the q-th sample trajectory point respectively to obtain the normalized score corresponding to the i-th sample calculation data of the q-th sample trajectory point and the normalized score corresponding to the j-th sample calculation data of the q-th sample trajectory point, and then substitute the k value, the normalized score corresponding to the i-th sample calculation data of the q-th sample trajectory point, and the normalized score corresponding to the j-th sample calculation data of the q-th sample trajectory point into the above conflict quantification formula for calculation to obtain the conflict quantification value between the i-th sample calculation data and the j-th sample calculation data.

[0216] In the embodiment of the present application, for each type of sample calculation data, the sample data may include k pieces of sample calculation data of this type, and each piece of sample calculation data of this type is the sample calculation data of a sample trajectory point. Exemplarily, continuing with the example in step 201, k = 5, n = 3, and the 3 types of sample calculation data are GPS status, number of satellites in the field of view, and number of matching feature points as an example. Then, the sample data includes 5 GPS statuses, 5 numbers of satellites in the field of view, and 5 numbers of matching feature points. The 5 GPS statuses are: 1, 1, 1, 1, 1. The 5 GPS statuses correspond one-to-one with 5 sample trajectory points. The 5 numbers of satellites in the field of view are 10, 10, 10, 10, 10. The 5 numbers of satellites in the field of view correspond one-to-one with the 5 sample trajectory points. The 5 numbers of matching feature points are 0028, 0120, 0150, 0238, 0143. The 5 numbers of matching feature points correspond one-to-one with the 5 sample trajectory points. Optionally, the server can use the normalization formula to normalize the sample score corresponding to the i-th sample calculation data (there are k pieces of sample calculation data of this type in the sample data) and the sample score corresponding to the j-th sample calculation data (there are k pieces of sample calculation data of this type in the sample data) respectively.

[0217] Among them, the normalization formula can be x ap represents the normalized score of the a-th sample score among the k sample scores corresponding to the p-th sample calculation data (there are a total of k pieces of the p-th sample calculation data in the sample data) among the n types of sample calculation data (for example, the sample score corresponding to the p-th sample calculation data of the a-th sample trajectory point), f p (a) represents the a-th sample score among the k sample scores corresponding to the p-th type of sample solution data, f p * represents the optimal score among the k sample scores corresponding to the p-th type of sample solution data, f p* represents the worst score among the k sample scores corresponding to the p-th type of sample solution data, where p = i or j, 1 ≤ a ≤ k, and a is an integer. The optimal score among the k sample scores can be the maximum value among the k sample scores, and the worst score among the k sample scores can be the minimum value among the k sample scores.

[0218] Optionally, the server can substitute the sample score corresponding to the i-th type of sample solution data of the q-th sample trajectory point, the optimal score among the k sample scores corresponding to the i-th type of sample solution data, and the worst score among the k sample scores corresponding to the i-th type of sample solution data into the above normalization formula for calculation respectively, to obtain the normalized score corresponding to the i-th type of sample solution data of the q-th sample trajectory point. Similarly, the server can substitute the sample score corresponding to the j-th type of sample solution data of the q-th sample trajectory point, the optimal score among the k sample scores corresponding to the j-th type of sample solution data, and the worst score among the k sample scores corresponding to the j-th type of sample solution data into the above normalization formula for calculation respectively, to obtain the normalized score corresponding to the j-th type of sample solution data of the q-th sample trajectory point.

[0219] Exemplarily, taking the normalization of the scores shown in Table 1 using the above normalization formula as an example, after the normalization of the scores shown in Table 1, the normalized scores corresponding to the i-th type of positioning solution data can be as shown in Table 3 below:

[0220] Table 3

[0221] Ri Score Meaning 0 1 Precise 1 0.9 Small Error 2 0.8 Large Error ... ... ... G 0 Invalid

[0222] Sub-step 2032: Determine the discrimination power quantization value of the i-th type of sample solution data according to the conflict quantization value between the i-th type of sample solution data and the n types of sample solution data, and the score variance corresponding to the i-th type of sample solution data.

[0223] Among them, the discrimination force quantization value is used to characterize the discrimination force of the i-th type of sample calculation data. The variance of the scores corresponding to the i-th type of sample calculation data can be the variance of the sample scores of the i-th type of sample calculation data of k sample trajectory points (one i-th type of sample calculation data for each sample trajectory point, a total of k i-th type of sample calculation data). Optionally, the variance of the scores corresponding to the i-th type of sample calculation data is the variance of the normalized scores corresponding to the i-th type of sample calculation data of the k sample trajectory points. The server can determine the discrimination force quantization value of the i-th type of sample calculation data by using the discrimination force quantization formula according to the conflict quantization value between the i-th type of sample calculation data and the n types of sample calculation data, and the variance of the normalized scores corresponding to the i-th type of sample calculation data.

[0224] Among them, the discrimination force quantization formula can be: C i represents the discrimination force quantization value of the i-th type of sample calculation data, r ij represents the conflict quantization value between the i-th type of sample calculation data and the j-th type of sample calculation data calculated in sub-step 2031, σ i represents the variance of the normalized scores corresponding to the i-th type of sample calculation data.

[0225] Optionally, the server can first calculate the variance of the normalized scores corresponding to the i-th type of sample calculation data of k sample trajectory points, and then substitute the variance of the normalized scores corresponding to the i-th type of sample calculation data of the k sample trajectory points and the conflict quantization value between the i-th type of sample calculation data and the j-th type of sample calculation data calculated in sub-step 2031 into the above discrimination force quantization formula for calculation to obtain the discrimination force quantization value of the i-th type of sample calculation data.

[0226] Sub-step 2033: Determine the target parameter value of the i-th model parameter among the n model parameters according to the discrimination force quantization value of the i-th type of sample calculation data and the discrimination force quantization values of the n types of sample calculation data.

[0227] Optionally, the server can determine the target parameter value of the i-th model parameter among the n model parameters by using the model parameter formula according to the discrimination force quantization value of the i-th type of sample calculation data and the discrimination force quantization values of the n types of sample calculation data. The model parameter formula can be: w i represents the i-th model parameter, C i represents the discrimination force quantization value of the i-th type of sample calculation data, C j represents the discrimination force quantization value of the j-th type of sample calculation data, j = 1, 2, 3... n.

[0228] Optionally, the server may determine the discrimination quantization value C of the j-th sample solution data with reference to the method in sub-step 2032 j , and use the discrimination quantization value C of the j-th sample solution data j and the discrimination quantization value C of the i-th sample solution data calculated in sub-step 2032 i to substitute into the model parameter formula for calculation respectively, so as to obtain the target parameter value of the i-th model parameter.

[0229] Step 204: Update the parameter values of the n model parameters in the initial scoring model according to the target parameter values of the n model parameters, so as to obtain a data scoring model.

[0230] The server may update the parameter values of the n model parameters in the initial scoring model obtained in step 202 according to the target parameter values of the n model parameters determined in step 203, and determine the scoring model with updated model parameters as the data scoring model. Exemplarily, the server uses the w calculated in step 203 1 , w 2 , w 3 ...w n to replace w in the initial scoring model of step 202 1 , w 2 , w 3 ...w n respectively, so as to update the parameter values of the n model parameters in the initial scoring model.

[0231] In summary, for the training method of the data scoring model provided in the embodiments of the present application, after the server obtains the sample data and the initial scoring model, it determines the target parameter values of the n model parameters according to the sample data, and updates the parameter values of the n model parameters in the initial scoring model according to the target parameter values of the n model parameters to obtain the data scoring model, so as to train the data scoring model, which can facilitate the subsequent use of the data scoring model to score the target trajectory points.

[0232] Exemplarily, please refer to Figure 4 , which shows a flowchart of a training method of a data filtering model provided in the embodiments of the present application. This method can be applied to Figure 1 the implementation environment shown. Refer to Figure 4 , this method may include the following steps:

[0233] Step 401: Obtain sample data.

[0234] The implementation process of this step 401 can refer to Figure 2 For step 201 in the illustrated embodiment, it will not be elaborated in the embodiments of the present application. It should be noted that the sample data in step 401 and the sample data in step 201 above may be the same batch of sample data or may not be the same batch of sample data. The embodiments of the present application do not limit this.

[0235] Step 402: Obtain an initial filtering model, which includes model variables and threshold parameters corresponding to the model variables.

[0236] Optionally, the initial filtering model may include model variables and threshold parameters corresponding to the model variables. The model variables may correspond to scoring values, and the threshold parameters may be score threshold parameters of the model variables. In the initial filtering model, the parameter values of the threshold parameters may be unknowns or initial values. By way of example, the expression of the initial filtering model may be: where Res represents the filtering result, S represents the model variable, and S threshold represents the threshold parameter, 1 represents retaining the trajectory point (i.e., the trajectory point meets the conditions), and 0 represents filtering out the trajectory point (i.e., the trajectory point does not meet the conditions).

[0237] Step 403: Determine the target parameter value of the threshold parameter according to the sample data.

[0238] Optionally, the sample data includes n types of sample calculation data for each of the k sample trajectory points. The n types of sample calculation data for each sample trajectory point include at least one positioning calculation data in the positioning data collected by each of the m acquisition sources at the sample trajectory point, where k≥2 and k is an integer.

[0239] By way of example, please refer to Figure 5 , which shows a flowchart of a method for determining the target parameter value of the threshold parameter according to the sample data provided by the embodiments of the present application. Refer to Figure 5 , and this method may include the following steps:

[0240] Sub-step 4031: For each of the k sample trajectory points, score the sample trajectory point according to the data scoring model and the n types of sample calculation data of the sample trajectory point to obtain the scoring value of the sample trajectory point.

[0241] For each of the k sample trajectory points, the server may input the n types of sample calculation data of the sample trajectory point into the data scoring model, and use the data scoring model to score the sample trajectory point to obtain the scoring value of the sample trajectory point. By way of example, the server inputs the n types of sample calculation data of each sample trajectory point into the data scoring model Enable the data scoring model to score the sample trajectory points based on the n types of sample calculation data of the sample trajectory points, and output the scoring value S of the sample trajectory points.

[0242] Optionally, each type of sample calculation data of each sample trajectory point among the k sample trajectory points corresponds to a sample score. The n model variables in the data scoring model can be the scores corresponding to the n types of target calculation data (for example, the n model variables in the data scoring model can be the normalized scores corresponding to the n types of target calculation data) variables. The n types of sample calculation data of each sample trajectory point correspond one-to-one with the n types of target calculation data. For each sample trajectory point among the k sample trajectory points, the server can score the sample trajectory point according to the data scoring model and the sample scores corresponding to the n types of sample calculation data of the sample trajectory point, and obtain the scoring value of the sample trajectory point.

[0243] Exemplarily, taking k = 5 (that is, 5 sample trajectory points), n = 3 (that is, 3 types of sample calculation data), and the 3 types of sample calculation data being GPS status, number of satellites in the field of view, and number of matching feature points as an example, the data scoring model can be S = w 1 ×R 1 +w 2 ×R 2 +w 3 ×R 3 , R 1 , R 2 , R 3 are the model variables corresponding to the GPS status, number of satellites in the field of view, and number of matching feature points in sequence, and w 1 , w 2 , w 3 are the model parameters corresponding to R 1 , R 2 , R 3 in sequence. For each sample trajectory point from sample trajectory point 1 to sample trajectory point 5, the server can input the sample scores corresponding to the GPS status, number of satellites in the field of view, and number of matching feature points of the sample trajectory point into the data scoring model S = w 1 ×R 1 +w 2 ×R 2 +w 3 ×R 3 to enable the data scoring model to score the sample trajectory point and obtain the scoring value of the sample trajectory point. Exemplarily, the scoring values of sample trajectory point 1 to sample trajectory point 5 are S1, S2, S3, S4, and S5 in sequence.

[0244] Sub-step 4032: Determine the target parameter value of the threshold parameter according to the scoring values of the k sample trajectory points.

[0245] Optionally, the server may select one of the scoring scores of the k sample trajectory points as the target parameter value of the threshold parameter. For example, the server may sort the scoring scores of the k sample trajectory points and select one of the scoring scores of the k sample trajectory points as the target parameter value of the threshold parameter according to the sorting order.

[0246] Optionally, the server may sort the scoring scores of the k sample trajectory points in descending order of the scoring scores, and determine the minimum score among the top 10% of the scoring scores in the sorting result as the target parameter value of the threshold parameter. For example, the scoring scores of the k sample trajectory points may be S1, S2, S3, S4, S5, S6, S7... St (since there may be at least two sample trajectory points with equal scoring scores among the k sample trajectory points, the number of scoring scores of the k sample trajectory points may not be equal to k). The sorting result obtained by the server sorting the scoring scores of the k sample trajectory points in descending order of the scoring scores is: St, S2, S1, S4, S5, S6, S7... S3. The top 10% of the scoring scores in the sorting result may be St, S2, S1, S4, S5, S6, and the minimum score among the top 10% of the scoring scores is S6. Therefore, the server determines that the target parameter value of the threshold parameter is S6.

[0247] Those skilled in the art can easily understand that the solution for determining the target parameter value of the threshold parameter according to the scoring scores of the k sample trajectory points provided in the embodiments of the present application is only exemplary. In actual applications, the server may determine the target parameter value of the threshold parameter according to other implementation manners. For example, the server may determine the minimum score among the top 15% of the scoring scores in the sorting result of the scoring scores of the k sample trajectory points as the target parameter value of the threshold parameter. For another example, the server may determine the minimum score among the top 20% of the scoring scores in the sorting result of the scoring scores of the k sample trajectory points as the target parameter value of the threshold parameter. For still another example, the server may determine the maximum score among the scoring scores of the k sample trajectory points as the target parameter value of the threshold parameter. The embodiments of the present application do not limit this. It should be noted that the selection of the target parameter value of the threshold parameter is related to the filtering effect of the data filtering model. If the target parameter value of the threshold parameter is too large, it will cause excessive filtering and fewer trajectory points will be filtered out. If the target parameter value of the threshold parameter is too small, it is difficult to effectively filter out noise trajectory points.

[0248] Step 404: Update the parameter value of the threshold parameter in the initial filtering model according to the target parameter value of the threshold parameter to obtain a data filtering model.

[0249] The server may update the parameter value of the threshold parameter in the initial filtering model obtained in step 402 according to the target parameter value of the threshold parameter determined in step 403, and determine the filtering model with updated model parameters as the data filtering model. Exemplarily, the server uses the S calculated in step 403 threshold to replace the S in the initial filtering model of step 402 threshold , so as to update the parameter value of the threshold parameter in the initial filtering model.

[0250] In summary, for the training method of the data filtering model provided in the embodiments of the present application, after the server obtains the sample data and the initial filtering model, it determines the target parameter value of the threshold parameter according to the sample data, and updates the parameter value of the threshold parameter in the initial filtering model according to the target parameter value of the threshold parameter to obtain the data filtering model, so as to train the data filtering model, which can facilitate the subsequent use of the data filtering model to filter the target trajectory points.

[0251] For the training method of the data filtering model provided in the embodiments of the present application, by setting the parameter value of the threshold parameter in the data filtering model, different degrees of trajectory point filtering can be realized, and the noise trajectory points can be eliminated.

[0252] In the embodiments of the present application, after the above Figures 2 to 5 embodiments, the server can train the data scoring model and the data filtering model. During the driving of the vehicle, the server can use the above data scoring model to score the trajectory points, and filter the trajectory points of the vehicle based on the scoring value of the data scoring model and the data filtering model. The following combines Figure 6 and Figure 7 to introduce the filtering process of the trajectory points, that is, the above Figures 2 to 5 embodiments are the model training process, and the following Figure 6 and Figure 7 are the model application processes.

[0253] Exemplarily, please refer to Figure 6 , which shows a flowchart of a method for filtering trajectory points provided in the embodiments of the present application. This method can be applied to the Figure 1 shown implementation environment. Refer to Figure 6 , this method may include the following steps:

[0254] Step 601, obtain n types of target solution data in the positioning data collected by m acquisition sources at the same target trajectory point, and the n types of target solution data include at least one positioning solution data in the positioning data collected by each acquisition source.

[0255] Among them, the target trajectory point is any trajectory point during the vehicle's driving, n≥m≥2, and both m and n are integers. The m acquisition sources are located on the same vehicle (such as vehicle 022). In the embodiments of the present application, there are two possible implementation manners for the server to obtain n types of target calculation data from the positioning data collected by the m acquisition sources at the same target trajectory point:

[0256] The first implementation manner: The server receives the positioning data collected by the m acquisition sources on the vehicle at the same target trajectory point reported by the vehicle, and extracts n types of target calculation data from the positioning data collected by the m acquisition sources on the vehicle at the target trajectory point reported by the vehicle.

[0257] Optionally, during the vehicle's driving, the m acquisition sources on the vehicle can collect the positioning data of the vehicle at the target trajectory point, and the vehicle can report the positioning data collected by the m acquisition sources at the target trajectory point to the server. Among them, the positioning data collected by the m acquisition sources at the target trajectory point can include m positioning data (that is, the positioning data of the vehicle at the target trajectory point is a set of data composed of m positioning data, and the m positioning data correspond to the m acquisition sources). Each of the m positioning data can include at least one positioning calculation data, and the positioning calculation data included in the m positioning data are not all the same. After the server receives the m positioning data collected by the m acquisition sources on the vehicle at the target trajectory point reported by the vehicle, it can extract at least one type of target calculation data from each of the m positioning data, so that the server can extract n types of target calculation data from the m positioning data.

[0258] Exemplarily, taking m = 2 (that is, 2 acquisition sources), the m acquisition sources are a GPS device and a camera, n = 3 (that is, 3 types of sample calculation data), and the n types of sample calculation data are GPS status, the number of satellites in the field of view, and the number of matching feature points as an example. The positioning data of the vehicle at the target trajectory point collected by the m acquisition sources can include: the positioning data 61 of the GPS device at the target trajectory point and the positioning data 62 collected by the camera at the target trajectory point. Among them:

[0259] The positioning data 61 is: $GPGGA,062802.00,3030.366090,N,11425.098182,E,1,12,0.7,20.625,M,0.000,M,,*$, $

[0260] The positioning data 62 is: 062802.00,0033,0237,3000,0143,0.376*

[0261] The formats of the positioning data 61 and the positioning data 62 and the meanings of the respective data therein can Figure 2 Step 201 in the illustrated embodiment will not be elaborated in the embodiments of the present application.

[0262] The n types of sample calculation data extracted by the server from the positioning data of the target trajectory point are respectively: 1 (GPS status), 12 (number of satellites in the field of view), and 0143 (number of matching feature points). That is, the n types of target calculation data in the positioning data collected by the m acquisition sources of the vehicle at the same target trajectory point obtained by the server are 1, 12, and 0143.

[0263] The second implementation manner: The server receives the n types of target calculation data in the positioning data collected by the m acquisition sources on the vehicle at the same target trajectory point reported by the vehicle.

[0264] Optionally, during the driving of the vehicle, the m acquisition sources on the vehicle can collect the positioning data of the vehicle at the target trajectory point. The positioning data collected by the m acquisition sources at the target trajectory point can include m positioning data, and each of the m positioning data can contain at least one type of positioning calculation data. The vehicle can extract at least one type of target calculation data from each of the m positioning data, so that the vehicle can extract n types of target calculation data from the m positioning data. After that, the vehicle can report the n types of target calculation data to the server, and the server can receive the n types of target calculation data. For example, the server receives the n types of target calculation data 1, 12, and 0143 in the positioning data collected by the m acquisition sources of the vehicle at the same target trajectory point sent by the vehicle.

[0265] Step 602: Score the target trajectory point according to the data scoring model and the n types of target calculation data to obtain the scoring value of the target trajectory point.

[0266] The server can input the n types of target calculation data in the positioning data collected by the m acquisition sources at the target trajectory point into the data scoring model, and use the data scoring model to score the target trajectory point to obtain the scoring value of the target trajectory point. As described above, it is easy to understand that the data scoring model is obtained by updating the parameter values of the n model parameters in the initial scoring model according to the target parameter values of the n model parameters. The initial scoring model includes n model variables and n model parameters corresponding to the n model variables, and the target parameter values of the n model parameters are determined according to the sample data. The process of determining the target parameter values of the n model parameters according to the sample data can be as Figure 3 shown, which will not be elaborated in the embodiments of the present application.

[0267] Optionally, the n model variables of the data scoring model may be score variables corresponding to n types of target solution data, and the server may score the target trajectory point according to the data scoring model and the scores corresponding to the n types of target solution data in the positioning data collected by the m acquisition sources at the target trajectory point.

[0268] Exemplarily, please refer to Figure 7 , which shows a flowchart of a method for scoring a target trajectory point according to a data scoring model and n types of target solution data of the target trajectory point provided by an embodiment of the present application. Refer to Figure 7 , this method may include the following steps:

[0269] Sub-step 6021: Determine the scores corresponding to the n types of target solution data.

[0270] Optionally, the server may maintain multiple score mapping relationships, each score mapping relationship corresponding to a type of positioning solution data, and each score mapping relationship records the calculation results of multiple positioning solution data of the corresponding type and the scores corresponding to the calculation results of each positioning solution data. For each type of target solution data among the n types of target solution data of the target trajectory point, the server may determine the calculation result of the target solution data, query the corresponding score mapping relationship according to the calculation result of the target solution data, and determine the score corresponding to the calculation result of the target solution data in the score mapping relationship as the target score corresponding to the target solution data. Among them, the calculation result of the target solution data may be the quantization value of the positioning solution data.

[0271] Sub-step 6022: Score the target trajectory point according to the data scoring model and the scores corresponding to the n types of target solution data to obtain the scoring score of the target trajectory point.

[0272] Optionally, the server may input the scores corresponding to the n types of target solution data in the positioning data collected by the m acquisition sources at the target trajectory point into the data scoring model, so that the data scoring model scores the target trajectory point according to the scores corresponding to the n types of target solution data. Exemplarily, the server may substitute the scores corresponding to the n types of target solution data into the data scoring model for calculation, and use the calculation result as the scoring score of the target trajectory point.

[0273] Step 603: Filter the target trajectory point according to the data filtering model and the scoring score of the target trajectory point. As easily understood as described above, the data filtering model is obtained by updating the parameter value of the threshold parameter in the initial filtering model according to the target parameter value of the threshold parameter. The initial filtering model includes model variables and the threshold parameter corresponding to the model variables, and the target parameter value of the threshold parameter is determined according to the sample data. The process of determining the target parameter value of the threshold parameter according to the sample data may be as Figure 5 As shown, the embodiments of the present application will not be elaborated here.

[0274] Optionally, the server may input the scoring value of the target trajectory point into a data filtering model, so that the data filtering model filters the target trajectory point according to the scoring value of the target trajectory point. Exemplarily, the server may substitute the scoring value of the target trajectory point into the data filtering model for calculation, and filter the target trajectory point according to the calculation result output by the data filtering model.

[0275] Exemplarily, the server may input the scoring value of the target trajectory point into the data filtering model as shown in step 404 after updating the threshold parameter, and judge the relationship between the scoring value of the target trajectory point and the target parameter value of the threshold parameter through the data filtering model. If the scoring value of the target trajectory point is greater than or equal to the target parameter value of the threshold parameter, the calculation result is 1, and the server retains the target trajectory point. If the scoring value of the target trajectory point is less than the target parameter value of the threshold parameter, the calculation result is 0, and the server filters out the target trajectory point (that is, discards the target trajectory point). Thus, the filtering of the target trajectory point can be realized.

[0276] In summary, for the trajectory point filtering method provided by the embodiments of the present application, after the server obtains n types of target solution data in the positioning data collected by m acquisition sources at the same target trajectory point, it scores the target trajectory point according to the data scoring model and the n types of target solution data, obtains the scoring value of the target trajectory point, and filters the target trajectory point according to the data filtering model and the scoring value of the target trajectory point. The n types of target solution data include at least one positioning solution data in the positioning data collected by each acquisition source. The trajectory point filtering method can comprehensively consider filtering trajectory points from the perspectives of different acquisition sources, which helps to improve the filtering effect. The trajectory point filtering method provided by the embodiments of the present application can realize online real-time cleaning of trajectory points, without strict prior knowledge, and can comprehensively consider filtering trajectory points from the perspectives of different acquisition sources. The trajectory points filtered based on the trajectory point filtering method can be well applied to location-based services such as vehicle trajectory restoration, providing reliable basic data for location-based services.

[0277] The following is the device embodiment of the present application, which can be used to execute the method embodiment of the present application. For details not disclosed in the device embodiment of the present application, please refer to the method embodiment of the present application.

[0278] Please refer to Figure 8 , which shows a schematic logical structure diagram of a trajectory point filtering device 800 provided by the embodiments of the present application. The trajectory point filtering device 800 may be a server or a functional component in the server. The server may be Figure 1 the data center server 001 in the shown implementation environment. Refer to Figure 8 , the trajectory point filtering device 800 may include, but is not limited to:

[0279] An acquisition module 810, configured to acquire n types of target solution data from the positioning data collected by m acquisition sources at the same target trajectory point, where the n types of target solution data include at least one positioning solution data in the positioning data collected by each acquisition source. The target trajectory point is any trajectory point during the vehicle driving process, n≥m≥2, and both m and n are integers;

[0280] A scoring module 820, configured to score the target trajectory point according to a data scoring model and the n types of target solution data to obtain a scoring value of the target trajectory point;

[0281] A filtering module 830, configured to filter the target trajectory point according to a data filtering model and the scoring value of the target trajectory point.

[0282] Optionally, the data scoring model is obtained by updating the parameter values of n model parameters in an initial scoring model according to the target parameter values of the n model parameters. The initial scoring model includes n model variables and n model parameters corresponding to the n model variables, and the target parameter values of the n model parameters are determined according to sample data.

[0283] Optionally, the sample data includes n types of sample solution data corresponding to the n types of target solution data;

[0284] The target parameter values of the n model parameters are determined through the following steps:

[0285] Determine the conflict quantification value between the i-th type of sample solution data and the j-th type of sample solution data, where the conflict quantification value is used to characterize the conflict between the i-th type of sample solution data and the j-th type of sample solution data, 1≤i≤n, 1≤j≤n, i≠j, and both i and j are integers;

[0286] According to the conflict quantification value between the i-th type of sample solution data and the n types of sample solution data, and the variance of the score corresponding to the i-th type of sample solution data, determine the discrimination quantification value of the i-th type of sample solution data, where the discrimination quantification value is used to characterize the discrimination of the i-th type of sample solution data;

[0287] According to the discrimination quantification value of the i-th type of sample solution data and the discrimination quantification values of the n types of sample solution data, determine the target parameter value of the i-th model parameter among the n model parameters.

[0288] Optionally, the sample data includes n types of sample calculation data for each of the k sample trajectory points. The n types of sample calculation data for each sample trajectory point include at least one type of positioning calculation data collected by each of the m acquisition sources at the sample trajectory point. Each type of sample calculation data for each sample trajectory point corresponds to a sample score. The variance of the scores corresponding to the i-th type of sample calculation data is the variance of the normalized scores corresponding to the i-th type of sample calculation data of the k sample trajectory points. k≥2, and k is an integer;

[0289] Determining the conflict quantification value between the i-th type of sample calculation data and the j-th type of sample calculation data includes:

[0290] Using the conflict quantification formula to determine the conflict quantification value between the i-th type of sample calculation data and the j-th type of sample calculation data;

[0291] According to the conflict quantification value between the i-th type of sample calculation data and the n types of sample calculation data, and the variance of the scores corresponding to the i-th type of sample calculation data, determining the discrimination quantification value of the i-th type of sample calculation data includes:

[0292] According to the conflict quantification value between the i-th type of sample calculation data and the n types of sample calculation data, and the variance of the normalized scores corresponding to the i-th type of sample calculation data, using the discrimination quantification formula to determine the discrimination quantification value of the i-th type of sample calculation data;

[0293] According to the discrimination quantification value of the i-th type of sample calculation data and the discrimination quantification values of the n types of sample calculation data, determining the target parameter value of the i-th model parameter among the n model parameters includes:

[0294] According to the discrimination quantification value of the i-th type of sample calculation data and the discrimination quantification values of the n types of sample calculation data, using the model parameter formula to determine the target parameter value of the i-th model parameter among the n model parameters;

[0295] Among them, the conflict quantification formula is

[0296] The discrimination quantification formula is

[0297] The model parameter formula is:

[0298] r ij represents the conflict quantification value between the i-th type of sample calculation data and the j-th type of sample calculation data. q = 1, 2, 3... k, d q represents the difference between the normalized score corresponding to the i-th type of sample calculation data of the q-th sample trajectory point among the k sample trajectory points and the normalized score corresponding to the j-th type of sample calculation data of the q-th sample trajectory point, C i Represents the discrimination quantization value of the i-th sample solution data, σ i Represents the variance of the normalized scores corresponding to the i-th sample solution data, w i Represents the i-th model parameter, C j Represents the discrimination quantization value of the j-th sample solution data, where j = 1, 2, 3... n.

[0299] Optionally, the normalized score corresponding to the i-th sample solution data is obtained by normalizing the sample score corresponding to the i-th sample solution data, and the normalized score corresponding to the j-th sample solution data is obtained by normalizing the sample score corresponding to the j-th sample solution data.

[0300] Optionally, the normalized score corresponding to the i-th sample solution data is obtained by normalizing the sample score corresponding to the i-th sample solution data using a normalization formula, and the normalized score corresponding to the j-th sample solution data is obtained by normalizing the sample score corresponding to the j-th sample solution data using the normalization formula;

[0301] where the normalization formula is x ap Represents the normalized score of the a-th sample score among the k sample scores corresponding to the p-th sample solution data among the n sample solution data, f p (a) represents the a-th sample score among the k sample scores corresponding to the p-th sample solution data, f p * represents the optimal score among the k sample scores corresponding to the p-th sample solution data, f p* Represents the worst score among the k sample scores corresponding to the p-th sample solution data, where p = i or j, 1 ≤ a ≤ k, and a is an integer.

[0302] Optionally, the n model variables of the data scoring model are the score variables corresponding to the n target solution data;

[0303] The scoring module 820 is used to:

[0304] Determine the scores corresponding to the n target solution data;

[0305] Score the target trajectory point according to the data scoring model and the scores corresponding to the n target solution data to obtain the scoring value of the target trajectory point.

[0306] Optionally, the data filtering model is obtained by updating the parameter value of the threshold parameter in the initial filtering model according to the target parameter value of the threshold parameter. The initial filtering model includes the model variable and the threshold parameter corresponding to the model variable, and the target parameter value of the threshold parameter is determined according to the sample data.

[0307] Optionally, the sample data includes n types of sample calculation data for each of the k sample trajectory points, and the n types of sample calculation data for each sample trajectory point include at least one positioning calculation data in the positioning data collected by each of the m acquisition sources at the sample trajectory point, where k≥2 and k is an integer;

[0308] The target parameter value of the threshold parameter is determined through the following steps:

[0309] For each of the k sample trajectory points, score the sample trajectory point according to the data scoring model and the n types of sample calculation data of the sample trajectory point to obtain the scoring value of the sample trajectory point;

[0310] Determine the target parameter value of the threshold parameter according to the scoring values of the k sample trajectory points.

[0311] Optionally, each type of sample calculation data of each sample trajectory point corresponds to a sample score,

[0312] For each of the k sample trajectory points, score the sample trajectory point according to the data scoring model and the n types of sample calculation data of the sample trajectory point to obtain the scoring value of the sample trajectory point, including:

[0313] For each of the k sample trajectory points, score the sample trajectory point according to the data scoring model and the sample scores corresponding to the n types of sample calculation data of the sample trajectory point to obtain the scoring value of the sample trajectory point.

[0314] In summary, for the trajectory point filtering device provided in the embodiment of the present application, after the acquisition module acquires n types of target calculation data in the positioning data collected by m acquisition sources at the same target trajectory point, the scoring module scores the target trajectory point according to the data scoring model and the n types of target calculation data to obtain the scoring value of the target trajectory point, and the filtering module filters the target trajectory point according to the data filtering model and the scoring value of the target trajectory point. The n types of target calculation data include at least one positioning calculation data in the positioning data collected by each acquisition source. The trajectory point filtering device can comprehensively consider filtering trajectory points from the perspectives of different acquisition sources, which helps to improve the filtering effect.

[0315] Please refer to Figure 9 , which shows a schematic logical structure diagram of a training device 900 for a data scoring model provided in the embodiment of the present application. The training device 900 for the data scoring model can be a server or a functional component in the server. Exemplarily, the server can be Figure 1 the data center server 001 in the shown implementation environment. Refer to Figure 9 , the training device 900 of the data scoring model may include, but is not limited to:

[0316] An acquisition module 910, configured to acquire sample data and an initial scoring model, where the initial scoring model includes n model variables and n model parameters corresponding to the n model variables, n≥2, and n is an integer;

[0317] A determination module 920, configured to determine target parameter values of the n model parameters according to the sample data;

[0318] An update module 930, configured to update the parameter values of the n model parameters in the initial scoring model according to the target parameter values of the n model parameters to obtain a data scoring model.

[0319] Optionally, the sample data includes n types of sample solution data corresponding to n types of target solution data;

[0320] The determination module 920 is configured to:

[0321] Determine a conflict quantification value between the i-th type of sample solution data and the j-th type of sample solution data, where the conflict quantification value is used to characterize the conflict between the i-th type of sample solution data and the j-th type of sample solution data, 1≤i≤n, 1≤j≤n, i≠j, and both i and j are integers;

[0322] According to the conflict quantification value between the i-th type of sample solution data and the n types of sample solution data, and the score variance corresponding to the i-th type of sample solution data, determine a discrimination quantification value of the i-th type of sample solution data, where the discrimination quantification value is used to characterize the discrimination of the i-th type of sample solution data;

[0323] According to the discrimination quantification value of the i-th type of sample solution data and the discrimination quantification values of the n types of sample solution data, determine the target parameter value of the i-th model parameter among the n model parameters.

[0324] Optionally, the sample data includes n types of sample solution data of each of the k sample trajectory points, the n types of sample solution data of each sample trajectory point include at least one positioning solution data in the positioning data collected by each of the m acquisition sources at the sample trajectory point, each type of the sample solution data of each sample trajectory point corresponds to a sample score, and the score variance corresponding to the i-th type of sample solution data is the variance of the normalized scores corresponding to the i-th type of sample solution data of the k sample trajectory points, k≥2, n≥m≥2, and both k and m are integers;

[0325] The determination module 920 is configured to:

[0326] Use a conflict quantification formula to determine the conflict quantification value between the i-th type of sample solution data and the j-th type of sample solution data;

[0327] According to the conflict quantification value between the i-th sample solution data and the n sample solution data, and the variance of the normalized scores corresponding to the i-th sample solution data, the discrimination quantification value of the i-th sample solution data is determined by using the discrimination quantification formula;

[0328] According to the discrimination quantification value of the i-th sample solution data and the discrimination quantification values of the n sample solution data, the target parameter value of the i-th model parameter among the n model parameters is determined by using the model parameter formula;

[0329] Among them, the conflict quantification formula is

[0330] The discrimination quantification formula is

[0331] The model parameter formula is:

[0332] r ij represents the conflict quantification value between the i-th sample solution data and the j-th sample solution data, q = 1, 2, 3... k, d q represents the difference between the normalized score corresponding to the i-th sample solution data and the normalized score corresponding to the j-th sample solution data of the q-th sample trajectory point among the k sample trajectory points, C i represents the discrimination quantification value of the i-th sample solution data, σ i represents the variance of the normalized scores corresponding to the i-th sample solution data, w i represents the i-th model parameter, C j represents the discrimination quantification value of the j-th sample solution data, j = 1, 2, 3... n.

[0333] Optionally, please continue to refer to Figure 9 , the training device 900 of the data scoring model further includes:

[0334] A normalization module 940, configured to perform normalization processing on the sample scores corresponding to the i-th sample solution data and the sample scores corresponding to the j-th sample solution data respectively, to obtain the normalized scores corresponding to the i-th sample solution data and the normalized scores corresponding to the j-th sample solution data.

[0335] Optionally, the normalization module 940 is configured to perform normalization processing on the sample scores corresponding to the i-th sample solution data and the corresponding sample scores of the j-th sample solution data respectively by using the normalization formula;

[0336] Among them, the normalization formula is x ap It represents the normalized score of the a-th sample score among the k sample scores corresponding to the p-th sample solution data in the n types of sample solution data, f p (a) represents the a-th sample score among the k sample scores corresponding to the p-th sample solution data, f p * represents the optimal score among the k sample scores corresponding to the p-th sample solution data, f p* It represents the worst score among the k sample scores corresponding to the p-th sample solution data, p = i or j, 1 ≤ a ≤ k, and a is an integer.

[0337] In summary, for the training device of the data scoring model provided in the embodiment of the present application, after the acquisition module obtains the sample data and the initial scoring model, the determination module determines the target parameter values of the n model parameters according to the sample data, and the update module updates the parameter values of the n model parameters in the initial scoring model according to the target parameter values of the n model parameters to obtain the data scoring model, which can facilitate obtaining the data scoring model.

[0338] Please refer to Figure 10 , which shows the logical structure diagram of a training device 1000 of a data filtering model provided in the embodiment of the present application. The training device 1000 of the data filtering model can be a server or a functional component in the server. Exemplarily, the server can be Figure 1 the data center server 001 in the shown implementation environment. Refer to Figure 10 , the training device 1000 of the data filtering model can include but is not limited to:

[0339] An acquisition module 1010, configured to acquire sample data and an initial filtering model, where the initial filtering model includes model variables and threshold parameters corresponding to the model variables;

[0340] A determination module 1020, configured to determine the target parameter values of the threshold parameters according to the sample data;

[0341] An update module 1030, configured to update the parameter values of the threshold parameters in the initial filtering model according to the target parameter values of the threshold parameters to obtain a data filtering model.

[0342] Optionally, the sample data includes n types of sample solution data of each sample trajectory point among k sample trajectory points, and the n types of sample solution data of each sample trajectory point include at least one positioning solution data in the positioning data collected by each of the m acquisition sources at the sample trajectory point, k ≥ 2, n ≥ m ≥ 2, and k, n, and m are all integers;

[0343] The determination module 1020 is configured to:

[0344] For each of the k sample trajectory points, score the sample trajectory point according to the data scoring model and the n types of sample calculation data of the sample trajectory point to obtain the scoring value of the sample trajectory point;

[0345] Determine the target parameter value of the threshold parameter according to the scoring values of the k sample trajectory points.

[0346] Optionally, each type of sample calculation data of each sample trajectory point corresponds to a sample score,

[0347] The determination module 1020 is configured to, for each of the k sample trajectory points, score the sample trajectory point according to the data scoring model and the sample scores corresponding to the n types of sample calculation data of the sample trajectory point to obtain the scoring value of the sample trajectory point.

[0348] In summary, for the training device of the data filtering model provided in the embodiment of the present application, after the acquisition module obtains the sample data and the initial filtering model, the determination module determines the target parameter value of the threshold parameter according to the sample data, and the update module updates the parameter value of the threshold parameter in the initial filtering model according to the target parameter value of the threshold parameter to obtain the data filtering model, which can facilitate obtaining the data filtering model.

[0349] Please refer to Figure 11 , which shows a schematic hardware structure diagram of a processing device 1100 provided in the embodiment of the present application. The processing device 1100 can be at least one of a trajectory point filtering device, a training device of a data scoring model, or a training device of a data filtering model, and the processing device 1100 can be a server. For example, the processing device 1100 can be Figure 1 the data center server in the shown implementation environment.

[0350] See Figure 11 , the processing device 1100 includes a processor 1102, a memory 1104, a communication interface 1106, and a bus 1108. The processor 1102, the memory 1104, and the communication interface 1106 are communicatively connected to each other through the bus 1108. Those skilled in the art should understand that Figure 11 the connection manner between the processor 1102, the memory 1104, and the communication interface 1106 shown is only exemplary. During implementation, the processor 1102, the memory 1104, and the communication interface 1106 can also be communicatively connected to each other using other connection manners besides the bus 1108.

[0351] Among them, the memory 1104 can be used to store instructions 11042 and data 11044. In the embodiments of the present application, the memory 1104 can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), flash memory, optical memory, and registers. And, the memory 1104 can include a hard disk and / or memory.

[0352] Among them, the processor 1102 can be a general-purpose processor. The general-purpose processor can be a processor that executes specific steps and / or operations by reading and executing instructions (such as instructions 11042) stored in a memory (such as memory 1104). The general-purpose processor may use data (such as data 11044) stored in the memory (such as memory 1104) during the execution of the above steps and / or operations. The general-purpose processor can be, for example but not limited to, a central processing unit (CPU). In addition, the processor 1102 can also be a dedicated processor. The dedicated processor can be a processor specifically designed to execute specific steps and / or operations. The dedicated processor can be, for example but not limited to, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), and a field-programmable gate array (FPGA). In addition, the processor 1102 can also be a combination of multiple processors, such as a multi-core processor. The processor 1102 can include one or more circuits to execute all or part of the steps of the method provided in the above embodiments.

[0353] Among them, the communication interface 1106 may include input / output (I / O) interfaces, physical interfaces, and logical interfaces, etc., which are used to implement the interconnection of components inside the processing device 1100, as well as interfaces for implementing the interconnection between the processing device 1100 and other devices (such as network devices or user devices). The physical interface may be a gigabit Ethernet (GE) interface, which can be used to implement the interconnection between the processing device 1100 and other devices (such as vehicles or edge servers), and the logical interface is an interface inside the processing device 1100, which can be used to implement the interconnection of components inside the processing device 1100. It is easy to understand that the communication interface 1106 can be used for the processing device 1100 to communicate with other devices. For example, the communication interface 1106 is used for sending and receiving information between the processing device 1100 and the vehicle.

[0354] Among them, the bus 1108 can be of any type and is a communication bus used to implement the interconnection of the processor 1102, the memory 1104, and the communication interface 1106, such as a system bus.

[0355] The above components can be respectively arranged on independent chips, or at least partially or entirely arranged on the same chip. Whether to arrange each component on a different chip independently or integrate them on one or more chips often depends on the needs of product design. The embodiments of the present application do not limit the specific implementation forms of the above components.

[0356] Figure 11 The shown processing device 1100 is only exemplary. During implementation, the processing device 1100 may further include other components, which will not be listed one by one herein. The Figure 11 shown processing device 1100 can implement the method provided by the embodiments of the present application by executing all or part of the steps of the method provided by the above embodiments.

[0357] Please refer to Figure 12 , which shows a schematic diagram of the hardware structure of another processing device 1200 provided by the embodiments of the present application. The processing device 1100 can be at least one of a trajectory point filtering device, a training device for a data scoring model, or a training device for a data filtering model, and the processing device 1200 can be a vehicle. See Figure 12 , the processing device 1200 includes a processor 1202, a memory 1204, a data acquisition source 1206, a communication interface 1208, and a bus 1210. The processor 1202, the memory 1204, the data acquisition source 1206, and the communication interface 1208 are communicatively connected to each other through the bus 1210. The data acquisition source 1206 is used to acquire the positioning data of the trajectory points of the trajectory point filtering device 1200, and the data acquisition source 1206 may be, for example but not limited to, a GPS device, a camera, etc. Descriptions of the processor 1202, the memory 1204, the communication interface 1208, and the bus 1210 may refer to Figure 9 the embodiments shown, and details thereof will not be repeated in this embodiment of the present application. Figure 12 The trajectory point filtering device 1200 shown is merely exemplary. In the implementation process, the processing device 1200 may further include other components, which will not be listed one by one herein.

[0358] An embodiment of the present application provides a trajectory point filtering system, which may include a server and a vehicle. The server may include, for example, Figure 8 or Figure 11 the trajectory point filtering device shown, and the vehicle may include, for example, Figure 12 the trajectory point filtering device shown. Optionally, the server may further include, for example, Figure 9 or Figure 11 the training device of the data scoring model shown, and / or, for example, Figure 10 or Figure 11 the trajectory point filtering device shown. Exemplarily, the trajectory point filtering system may be, for example, Figure 1 the trajectory point filtering system shown.

[0359] An embodiment of the present application provides a computer-readable storage medium, in which a computer program is stored. When the computer program runs on a computer, the computer is caused to execute all or part of the steps of the method provided in the embodiment shown in Figures 2 to 7 .

[0360] An embodiment of the present application provides a computer program product containing instructions. When the computer program product runs on a computer, the computer is caused to execute all or part of the steps of the method provided in the embodiment shown in Figures 2 to 7 .

[0361] An embodiment of the present application provides a chip, which includes a programmable logic circuit and / or program instructions. When the chip runs, it is used to implement all or part of the steps of the method provided in the embodiment shown in Figures 2 to 7 .

[0362] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product, which includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (such as coaxial cable, optical fiber, digital subscriber line) or wireless (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more integrated available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium, or a semiconductor medium (such as a solid-state drive), etc.

[0363] In the present application, terms such as "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance. The term "at least one" means one or more, and "a plurality" means two or more, unless otherwise clearly defined. The term "and / or" is merely a description of the association relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this article generally represents an "or" relationship between the associated objects before and after.

[0364] The method embodiments and device embodiments and other different types of embodiments provided in the embodiments of the present application can be referred to each other, and the embodiments of the present application do not limit this. The order of operations in the method embodiments provided in the embodiments of the present application can be appropriately adjusted, and the operations can also be increased or decreased according to the situation. Any person skilled in the art within the technical scope disclosed in the present application can easily think of a changed method, which should be covered by the protection scope of the present application, so it will not be elaborated here.

[0365] In the corresponding embodiments provided by the present application, it should be understood that the disclosed devices and the like can be implemented in other constitutive manners. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces. The indirect coupling or communication connection of devices or units can be in an electrical or other form.

[0366] The units described as separate components may or may not be physically separated. The components described as units may or may not be physical units. They can be located in one place or distributed to multiple network devices (such as terminal devices). Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0367] As described above, it is only an exemplary embodiment of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.< / lf> < / cr> < / lf> < / cr> < / lf> < / cr> < / lf> < / cr>

Claims

1. A method for filtering trajectory points, characterized in that, the method includes: obtaining n types of target calculation data in the positioning data collected by m acquisition sources at the same target trajectory point, where the n types of target calculation data include at least one positioning calculation data in the positioning data collected by each of the acquisition sources, the target trajectory point is any trajectory point during vehicle driving, n≥m≥2, and both m and n are integers; scoring the target trajectory point according to a data scoring model and the n types of target calculation data to obtain a scoring value of the target trajectory point; filtering the target trajectory point according to a data filtering model and the scoring value of the target trajectory point; wherein, the data scoring model is obtained by updating the parameter values of n model parameters in an initial scoring model according to the target parameter values of the n model parameters, and the target parameter values of the n model parameters are determined according to sample data; the sample data includes n types of sample calculation data of each of k sample trajectory points, the n types of sample calculation data correspond to the n types of target calculation data, each type of sample calculation data of each sample trajectory point corresponds to a sample score, k≥2, k is an integer, and the target parameter values of the n model parameters are determined through the following steps: determining a conflict quantification value between the i-th type of sample calculation data and the j-th type of sample calculation data, where the conflict quantification value is used to characterize the conflict between the i-th type of sample calculation data and the j-th type of sample calculation data, 1≤i≤n, 1≤j≤n, i≠j, and both i and j are integers; determining a discrimination quantification value of the i-th type of sample calculation data according to the conflict quantification value between the i-th type of sample calculation data and the n types of sample calculation data, and the variance of the normalized scores corresponding to the i-th type of sample calculation data, where the discrimination quantification value is used to characterize the discrimination of the i-th type of sample calculation data; determining the target parameter value of the i-th model parameter among the n model parameters according to the discrimination quantification value of the i-th type of sample calculation data and the discrimination quantification values of the n types of sample calculation data; Among them, the normalized score corresponding to the i-th type of sample calculation data is obtained by normalizing the sample score corresponding to the i-th type of sample calculation data using a normalization formula, and the normalization formula is x ap represents the normalized score of the a-th sample score among the k sample scores corresponding to the p-th sample calculation data among the n types of sample calculation data, f p (a) represents the a-th sample score among the k sample scores corresponding to the p-th sample calculation data, represents the optimal score among the k sample scores corresponding to the p-th sample calculation data, represents the worst score among the k sample scores corresponding to the p-th sample calculation data, p = i, 1 ≤ a ≤ k, and a is an integer.

2. The method according to claim 1, characterized in that, the initial scoring model includes n model variables and the n model parameters corresponding to the n model variables.

3. The method according to claim 1, characterized in that, the n types of sample calculation data of each sample trajectory point include at least one positioning calculation data in the positioning data collected by each of the m acquisition sources at the sample trajectory point; the determining the conflict quantification value between the i-th type of sample calculation data and the j-th type of sample calculation data includes: using a conflict quantification formula to determine the conflict quantification value between the i-th type of sample calculation data and the j-th type of sample calculation data; Determining the discrimination quantization value of the i-th sample solution data based on the conflict quantization value between the i-th sample solution data and the n sample solution data, and the variance of the normalized scores corresponding to the i-th sample solution data, includes: determining the discrimination quantization value of the i-th sample solution data by using a discrimination quantization formula based on the conflict quantization value between the i-th sample solution data and the n sample solution data, and the variance of the normalized scores corresponding to the i-th sample solution data; Determining the target parameter value of the i-th model parameter among the n model parameters based on the discrimination quantization value of the i-th sample solution data and the discrimination quantization values of the n sample solution data, includes: determining the target parameter value of the i-th model parameter among the n model parameters by using a model parameter formula based on the discrimination quantization value of the i-th sample solution data and the discrimination quantization values of the n sample solution data; wherein, the conflict quantification formula is The discrimination force quantization formula is The model parameter formula is as follows: r ij represents the conflict quantification value between the i-th sample solution data and the j-th sample solution data, q = 1, 2, 3... k, d q represents the difference between the normalized score corresponding to the i-th sample solution data and the normalized score corresponding to the j-th sample solution data of the q-th sample trajectory point among the k sample trajectory points, C i represents the discrimination quantification value of the i-th sample solution data, σ i represents the variance of the normalized score corresponding to the i-th sample solution data, w i represents the i-th model parameter, C j represents the discrimination quantification value of the j-th sample solution data, j = 1, 2, 3... n.

4. The method according to claim 1, wherein, the normalized score corresponding to the j-th sample solution data is obtained by normalizing the sample score corresponding to the j-th sample solution data by using the normalization formula. When normalizing the sample score corresponding to the j-th sample solution data by using the normalization formula, in the normalization formula, p = j.

5. The method according to any one of claims 1 to 4, wherein, the n model variables of the data scoring model are the score variables corresponding to the n target solution data; Scoring the target trajectory point according to the data scoring model and the n target solution data to obtain the scoring score of the target trajectory point, includes: determining the scores corresponding to the n target solution data; scoring the target trajectory point according to the data scoring model and the scores corresponding to the n target solution data to obtain the scoring score of the target trajectory point.

6. The method according to any one of claims 1 to 4, wherein, the data filtering model is obtained by updating the parameter value of the threshold parameter in the initial filtering model according to the target parameter value of the threshold parameter. The initial filtering model includes model variables and the threshold parameter corresponding to the model variables, and the target parameter value of the threshold parameter is determined according to the sample data.

7. The method according to claim 6, wherein, the target parameter value of the threshold parameter is determined by the following steps: For each sample trajectory point among the k sample trajectory points, scoring the sample trajectory point according to the data scoring model and the n sample solution data of the sample trajectory point to obtain the scoring score of the sample trajectory point; determining the target parameter value of the threshold parameter according to the scoring scores of the k sample trajectory points.

8. The method according to claim 7, wherein, For each of the k sample trajectory points, scoring the sample trajectory point according to the data scoring model and the n types of sample calculation data of the sample trajectory point to obtain a scoring value of the sample trajectory point, including: For each of the k sample trajectory points, scoring the sample trajectory point according to the data scoring model and the sample scores corresponding to the n types of sample calculation data of the sample trajectory point to obtain a scoring value of the sample trajectory point.

9. A trajectory point filtering device Characterized in that The device includes: An acquisition module, configured to acquire n types of target calculation data in the positioning data acquired by m acquisition sources at the same target trajectory point, where the n types of target calculation data include at least one positioning calculation data in the positioning data acquired by each acquisition source, the target trajectory point is any trajectory point during vehicle driving, n≥m≥2, and both m and n are integers; A scoring module, configured to score the target trajectory point according to a data scoring model and the n types of target calculation data to obtain a scoring value of the target trajectory point; A filtering module, configured to filter the target trajectory point according to a data filtering model and the scoring value of the target trajectory point; Wherein, the data scoring model is obtained by updating the parameter values of n model parameters in the initial scoring model according to the target parameter values of the n model parameters, and the target parameter values of the n model parameters are determined according to sample data; The sample data includes n types of sample calculation data of each of the k sample trajectory points, the n types of sample calculation data correspond to the n types of target calculation data, each type of sample calculation data of each sample trajectory point corresponds to a sample score, k≥2, k is an integer, and the target parameter values of the n model parameters are determined through the following steps: Determine the conflict quantification value between the i-th type of sample calculation data and the j-th type of sample calculation data, where the conflict quantification value is used to characterize the conflict between the i-th type of sample calculation data and the j-th type of sample calculation data, 1≤i≤n, 1≤j≤n, i≠j, and both i and j are integers; According to the conflict quantification value between the i-th type of sample calculation data and the n types of sample calculation data, and the variance of the normalized scores corresponding to the i-th type of sample calculation data, determine the discrimination quantification value of the i-th type of sample calculation data, where the discrimination quantification value is used to characterize the discrimination of the i-th type of sample calculation data; According to the discrimination quantification value of the i-th type of sample calculation data and the discrimination quantification values of the n types of sample calculation data, determine the target parameter value of the i-th model parameter among the n model parameters; Among them, the normalized score corresponding to the i-th sample solution data is obtained by normalizing the sample score corresponding to the i-th sample solution data using the normalization formula, and the normalization formula is x ap represents the normalized score of the a-th sample score among the k sample scores corresponding to the p-th sample solution data among the n sample solution data, f p (a) represents the a-th sample score among the k sample scores corresponding to the p-th sample solution data, represents the optimal score among the k sample scores corresponding to the p-th sample solution data, represents the worst score among the k sample scores corresponding to the p-th sample solution data, p = i, 1 ≤ a ≤ k, and a is an integer.

10. The device according to claim 9 Characterized in that The initial scoring model includes n model variables and the n model parameters corresponding to the n model variables.

11. The device according to claim 9 Characterized in that The n types of sample calculation data for each of the sample trajectory points include at least one type of positioning calculation data in the positioning data collected by each of the m acquisition sources at the sample trajectory point; Determining the conflict quantification value between the i-th type of sample calculation data and the j-th type of sample calculation data includes: using a conflict quantification formula to determine the conflict quantification value between the i-th type of sample calculation data and the j-th type of sample calculation data; Determining the discrimination quantification value of the i-th type of sample calculation data according to the conflict quantification value between the i-th type of sample calculation data and the n types of sample calculation data, and the variance of the normalized score corresponding to the i-th type of sample calculation data includes: according to the conflict quantification value between the i-th type of sample calculation data and the n types of sample calculation data, and the variance of the normalized score corresponding to the i-th type of sample calculation data, using a discrimination quantification formula to determine the discrimination quantification value of the i-th type of sample calculation data; Determining the target parameter value of the i-th model parameter among the n model parameters according to the discrimination quantification value of the i-th type of sample calculation data and the discrimination quantification values of the n types of sample calculation data includes: according to the discrimination quantification value of the i-th type of sample calculation data and the discrimination quantification values of the n types of sample calculation data, using a model parameter formula to determine the target parameter value of the i-th model parameter among the n model parameters; Among them, the conflict quantification formula is The discrimination force quantization formula is The formula for the model parameters is as follows: r ij represents the conflict quantification value between the i-th sample solution data and the j-th sample solution data, q = 1, 2, 3... k, d q represents the difference between the normalized score corresponding to the i-th sample solution data and the normalized score corresponding to the j-th sample solution data of the q-th sample trajectory point among the k sample trajectory points, C i represents the discrimination quantification value of the i-th sample solution data, σ i represents the variance of the normalized score corresponding to the i-th sample solution data, w i represents the i-th model parameter, C j represents the discrimination quantification value of the j-th sample solution data, j = 1, 2, 3... n.

12. The apparatus according to claim 9, wherein, The normalized score corresponding to the j-th type of sample calculation data is obtained by normalizing the sample score corresponding to the j-th type of sample calculation data using the normalization formula. When normalizing the sample score corresponding to the j-th type of sample calculation data using the normalization formula, in the normalization formula, p = j.

13. The apparatus according to any one of claims 9 to 12, wherein, The n model variables of the data scoring model are score variables corresponding to the n types of target calculation data; The scoring module is configured to: Determine the scores corresponding to the n types of target calculation data; Score the target trajectory point according to the data scoring model and the scores corresponding to the n types of target calculation data to obtain the scoring score of the target trajectory point.

14. The apparatus according to any one of claims 9 to 12, The data filtering model is obtained by updating the parameter value of the threshold parameter in the initial filtering model according to the target parameter value of the threshold parameter. The initial filtering model includes model variables and the threshold parameter corresponding to the model variables, and the target parameter value of the threshold parameter is determined according to the sample data.

15. The apparatus according to claim 14, wherein, The target parameter value of the threshold parameter is determined through the following steps: For each of the k sample trajectory points, score the sample trajectory point according to the data scoring model and the n types of sample calculation data of the sample trajectory point to obtain the scoring score of the sample trajectory point; Determine the target parameter value of the threshold parameter according to the scoring scores of the k sample trajectory points.

16. The device according to claim 15, wherein, for each of the k sample trajectory points, scoring the sample trajectory point according to the data scoring model and the n types of sample calculation data of the sample trajectory point to obtain a scoring value of the sample trajectory point, including: for each of the k sample trajectory points, scoring the sample trajectory point according to the data scoring model and the sample scores corresponding to the n types of sample calculation data of the sample trajectory point to obtain a scoring value of the sample trajectory point.

17. A trajectory point filtering device, wherein, comprising: a processor and a memory, wherein a program is stored in the memory, and the processor is configured to call the program stored in the memory, so that the trajectory point filtering device executes the trajectory point filtering method according to any one of claims 1 to 8.

18. A computer-readable storage medium, wherein, a computer program is stored in the computer-readable storage medium, and when the computer program runs on a computer, the computer is caused to execute the trajectory point filtering method according to any one of claims 1 to 8.

Images