Sensor data analysis method, apparatus, device, medium, and program product

Abstract

The application provides a sensor data analysis method, device, equipment, medium and program product. The method comprises the following steps: obtaining an evaluation index of a target object element, and obtaining a plurality of sensor data to be evaluated, each sensor data carrying a plurality of observation data; for each target object element in each sensor data, the target object element is matched with a preset high-precision map according to element information of the target object element, so as to obtain a matching result of the target object element in each sensor data, and the matching result comprises a single matching result corresponding to each observation data; based on the matching result, an index value of the target object element in each sensor data with respect to the evaluation index is obtained, and a quality analysis result of the target object element in each sensor is obtained based on the index value. When the sensor data analysis is performed, the individual difference is considered, the data analysis result is more objective and accurate, and support is provided for efficient use of the sensor data.

Description

Technical Field

[0001] This application relates to the field of high-precision mapping technology, and in particular to a sensor data analysis method, apparatus, equipment, medium, and program product. Background Technology

[0002] With the development of intelligent driving technology, intelligent driving vehicles rely heavily on high-precision maps for positioning and planning.

[0003] Correspondingly, to meet the high-performance requirements of intelligent driving, high-precision maps are increasingly demanding in terms of freshness. High-precision map manufacturers need to update these maps, typically by fully utilizing multi-source sensor data. Therefore, how to analyze and select appropriate sensor data as the data source for updating high-precision maps has become a crucial issue.

[0004] In related technologies, the analysis and selection of multi-source sensor data mainly involves a comprehensive evaluation of the object elements in each sensor data source, without considering the differences between sensor data for different individual object elements, resulting in inaccurate analysis results. Summary of the Invention

[0005] This application provides a sensor data analysis method, apparatus, device, medium, and program product to at least solve one of the above-mentioned technical problems.

[0006] According to one aspect of this application, a sensor data analysis method is provided, comprising:

[0007] The evaluation indicators for the target object elements are obtained, and the evaluation indicators include at least one of the following: individual accuracy indicators, individual existence indicators, and attribute correctness indicators;

[0008] Acquire data from multiple sensors to be evaluated, where each sensor data point carries multiple observation data points;

[0009] For each target object element in the sensor data, the target object element is matched with a preset high-precision map based on the element information of the target object element to obtain the matching result of the target object element in each sensor data. The matching result includes the single matching result corresponding to each observation data.

[0010] Based on the matching results, the index values ​​of the target object elements in each sensor data with respect to the evaluation index are obtained, and the quality analysis results of the target object elements in each sensor are obtained based on the index values.

[0011] In one implementation, the evaluation index is defined in the following ways:

[0012] Define individual precision indices for target object elements based on their spatial geometric information; and / or define individual existence indices for target object elements based on their existence information; and / or define attribute correctness indices for target object elements based on their element description information.

[0013] In one implementation, the element information includes location information and / or attribute information.

[0014] The step of matching the target object elements with a preset high-precision map based on the element information of the target object elements to obtain the matching result corresponding to the target object elements in each sensor data includes:

[0015] Based on the location information corresponding to the target object element, determine whether there is a map element in the preset high-precision map that matches the target object element;

[0016] In response to the determination that the map feature does not exist, the existence matching result corresponding to the target object feature is determined to be that there is no target object feature matching the map feature; and / or,

[0017] In response to the determination that the map element exists, the location matching result between the target object element and the map element is determined based on the location information corresponding to the target object element; and / or, the attribute matching result between the target object element and the map element is determined based on the attribute information corresponding to the target object element.

[0018] In one implementation, the individual accuracy index includes individual absolute accuracy and / or individual relative accuracy, and the matching result includes the positional matching result between the target object element and map elements in the preset high-precision map;

[0019] The step of obtaining the index value of the target object element in each sensor data with respect to the evaluation index based on the matching result includes:

[0020] For each sensor data point, based on the location matching result, obtain the absolute distance and / or relative distance between the target object element and the map element in each matching result;

[0021] Based on the absolute distance and / or relative distance, obtain the individual absolute accuracy and / or individual relative accuracy of the target object element in each sensor data;

[0022] Based on the individual absolute accuracy and / or individual relative accuracy and the total number of target object elements, the mean and standard deviation of the individual absolute accuracy and / or individual relative accuracy are obtained, and based on the mean and standard deviation of the individual absolute accuracy and / or individual relative accuracy, the index value of the target object element in each sensor data with respect to the individual accuracy index is obtained respectively.

[0023] In one embodiment, the sensor data includes target object features corresponding to at least one feature type, wherein the feature type includes point type and / or line type.

[0024] The step of obtaining the index value of the target object element in each sensor data with respect to the evaluation index based on the matching result includes:

[0025] For each type of element corresponding to the target object element, based on the matching result, the index value of the target object element corresponding to the element type in each sensor data with respect to the evaluation index is obtained.

[0026] In one implementation, the individual existence index includes individual recall rate and / or individual false positive rate, and the matching result includes the existence matching result between the target object element and map elements in the preset high-precision map;

[0027] The step of obtaining the index value of the target object element in each sensor data with respect to the evaluation index based on the matching result includes:

[0028] When the feature type is a point feature, for each sensor data, based on the existence matching result, a first number corresponding to the target object feature is obtained, and the individual recall rate of the target object feature is obtained according to the first number; and / or, when the feature type is a line feature, for each sensor data, based on the existence matching result, a first length of the target object feature is obtained, and the individual recall rate of the target object feature is obtained according to the first length; wherein, the first number is the number of all object features in the target object feature that match the map feature for each matching result, and the first length is the length of all object features in the target object feature that match the map feature for each matching result;

[0029] Based on the individual recall rate and the total number of target object elements, the mean and standard deviation of the individual recall rate are obtained, and based on the mean and standard deviation of the individual recall rate, the index value of the target object element in each sensor data with respect to the individual existence index is obtained respectively.

[0030] And / or,

[0031] When the feature type is a point feature, for each sensor data, based on the existence matching result, a second number corresponding to the target object feature is obtained, and the individual false positive rate of the target object feature is obtained according to the second number; and / or, when the feature type is a line feature, for each sensor data, based on the existence matching result, a second length corresponding to the target object feature is obtained, and the individual false positive rate of the target object feature is obtained according to the second length; wherein, the second number is the number of all object features in the target object feature that have a match with the map feature in each matching result, but do not have a corresponding map feature in the preset high-precision map, and the second length is the length of all object features in the target object feature that have a match with the map feature in each matching result, but do not have a corresponding map feature in the preset high-precision map;

[0032] Based on the individual false positive rate and the total number of target object elements, the mean and standard deviation of the individual false positive rate are obtained, and based on the mean and standard deviation of the individual false positive rate, the index value of the target object element in each sensor data with respect to the individual existence index is obtained.

[0033] In one implementation, the attribute correctness index includes individual attribute accuracy, and the matching result includes the attribute matching result between the target object element and the map element in the preset high-precision map.

[0034] The step of obtaining the index value of the target object element in each sensor data with respect to the evaluation index based on the matching result includes:

[0035] When the feature type is a point feature, for each sensor data, based on the attribute matching result, according to the third number of the target object features and the number of existing features in the existence matching result between the target object features and the map features, the attribute accuracy corresponding to the feature type is obtained.

[0036] When the feature type is linear, for each sensor data, based on the attribute matching result, according to the third length of the target object feature and the existence matching result between the target object feature and the map feature, the attribute accuracy corresponding to the feature type is obtained.

[0037] Wherein, the third number is the number of all object elements in the target object elements that match the map element and have the same attributes as the map element in each matching result; the third length is the length of all object elements in the target object elements that match the map element and have the same attributes as the map element in each matching result.

[0038] Based on the individual attribute accuracy rate and the total number of target object elements, the mean and standard deviation of the individual attribute accuracy rate are obtained, and based on the mean and standard deviation of the individual attribute accuracy rate, the index value of the target object element in each sensor data with respect to the attribute accuracy index is obtained.

[0039] According to another aspect of this application, a sensor data analysis apparatus is provided, comprising:

[0040] The indicator acquisition module is configured to acquire evaluation indicators for the elements of the target object. The evaluation indicators include at least one of the following: individual accuracy indicators, individual existence indicators, and attribute correctness indicators.

[0041] The data acquisition module is configured to acquire data from multiple sensors to be evaluated, wherein each sensor data carries multiple observation data.

[0042] The matching module is configured to match the target object elements in each sensor data with a preset high-precision map based on the element information of the target object elements, so as to obtain the matching result corresponding to the target object elements in each sensor data. The matching result includes the single matching result corresponding to each observation data.

[0043] The quality analysis module is configured to obtain the index value of the target object element in each sensor data with respect to the evaluation index based on the matching result, and obtain the quality analysis result of the target object element in each sensor based on the index value.

[0044] In one implementation, the evaluation index is defined in the following ways: defining an individual accuracy index of the target object element based on the spatial geometric information of the target object element; and / or defining an individual existence index of the target object element based on the existence information of the target object element; and / or defining an attribute correctness index of the target object element based on the element description information of the target object element.

[0045] In one implementation, the element information includes location information and / or attribute information.

[0046] The matching module includes:

[0047] The first determining unit is configured to determine whether there is a map element in the preset high-precision map that matches the target object element based on the location information corresponding to the target object element.

[0048] The second determining unit is configured to, in response to a determination result indicating that the map feature does not exist, determine that the existence matching result corresponding to the target object feature is that no target object feature matches the map feature; and / or,

[0049] The third determining unit is configured to, in response to a determination result indicating the existence of the map element, determine a location matching result between the target object element and the map element based on the location information corresponding to the target object element; and / or determine an attribute matching result between the target object element and the map element based on the attribute information corresponding to the target object element.

[0050] In one implementation, the individual accuracy index includes individual absolute accuracy and / or individual relative accuracy, and the matching result includes the positional matching result between the target object element and map elements in the preset high-precision map;

[0051] The quality analysis module includes:

[0052] The first acquisition unit is configured to acquire, for each sensor data, the absolute distance and / or relative distance between the target object element and the map element in each matching result, based on the location matching result;

[0053] The second acquisition unit is configured to acquire the individual absolute accuracy and / or individual relative accuracy corresponding to the target object element in each sensor data based on the absolute distance and / or relative distance, and to acquire the mean and standard deviation of the individual absolute accuracy and / or individual relative accuracy based on the mean and standard deviation of the individual absolute accuracy and / or individual relative accuracy and the total number of the target object elements, and to acquire the index value of the target object element in each sensor data with respect to the individual accuracy index based on the mean and standard deviation of the individual absolute accuracy and / or individual relative accuracy.

[0054] In one embodiment, the sensor data includes target object features corresponding to at least one feature type, wherein the feature type includes point type and / or line type.

[0055] The quality analysis module is specifically configured to, for each type of element corresponding to the target object element, obtain the index value of the target object element corresponding to the element type in each sensor data with respect to the evaluation index based on the matching result.

[0056] In one implementation, the individual existence index includes individual recall rate and / or individual false positive rate, and the matching result includes the existence matching result between the target object element and map elements in the preset high-precision map;

[0057] The quality analysis module includes:

[0058] The third acquisition unit is configured to, when the feature type is a point feature, acquire a first number corresponding to the target object feature based on the existence matching result for each sensor data, and acquire the individual recall rate of the target object feature based on the first number; and / or, when the feature type is a line feature, acquire a first length of the target object feature based on the existence matching result for each sensor data, and acquire the individual recall rate of the target object feature based on the first length; wherein, the first number is the number of all object features in the target object feature that match the map feature for each matching result, and the first length is the length of all object features in the target object feature that match the map feature for each matching result;

[0059] The fourth acquisition unit is configured to acquire the mean and standard deviation of the individual recall rate based on the individual recall rate and the total number of target object elements, and acquire the index value of the target object element in each sensor data with respect to the individual existence index based on the mean and standard deviation of the individual recall rate.

[0060] And / or,

[0061] The fifth acquisition module is configured to, when the feature type is a point feature, acquire a second number corresponding to the target object feature for each sensor data based on the existence matching result, and acquire the individual false positive rate of the target object feature based on the second number; and / or, when the feature type is a line feature, acquire a second length corresponding to the target object feature for each sensor data based on the existence matching result, and acquire the individual false positive rate of the target object feature based on the second length; wherein, the second number is the number of all object features in the target object feature that match the map feature in each matching result and do not have a corresponding map feature in the preset high-precision map, and the second number is the number of all object features in the target object feature that match the map feature in each matching result and do not have a corresponding map feature in the preset high-precision map;

[0062] The sixth acquisition module is configured to acquire the mean and standard deviation of the individual false positive rate based on the individual false positive rate and the total number of target object elements, and acquire the index value of the target object element in each sensor data with respect to the individual existence index based on the mean and standard deviation of the individual recall rate.

[0063] In one implementation, the attribute correctness index includes individual attribute accuracy, and the matching result includes the attribute matching result between the target object element and the map element in the preset high-precision map.

[0064] The quality analysis module includes:

[0065] The seventh acquisition module is configured to, when the feature type is point features, acquire the attribute accuracy corresponding to the feature type for each sensor data based on the attribute matching result, according to the third number of the target object features and the number of existing features in the existence matching result between the target object features and the map features; and / or,

[0066] The eighth acquisition module, when the feature type is linear, for each sensor data, based on the attribute matching result, according to the third length of the target object feature and the existence matching result between the target object feature and the map feature, the attribute accuracy corresponding to the feature type is obtained.

[0067] Wherein, the third number is the number of all object elements in the target object elements that match the map element and have the same attributes as the map element in each matching result; the third length is the length of all object elements in the target object elements that match the map element and have the same attributes as the map element in each matching result.

[0068] Based on the individual attribute accuracy rate and the total number of target object elements, the mean and standard deviation of the individual attribute accuracy rate are obtained, and the index value of the target object element in each sensor data with respect to the attribute accuracy index is obtained respectively.

[0069] According to another aspect of this application, an electronic device is provided, comprising: at least one processor and a memory;

[0070] The memory stores computer-executed instructions;

[0071] The at least one processor executes computer execution instructions stored in the memory, causing the at least one processor to execute the sensor data analysis method.

[0072] According to another aspect of this application, a computer-readable storage medium is provided, wherein computer-executable instructions are stored in the computer-readable storage medium, and when a processor executes the computer-executable instructions, the sensor data analysis method is implemented.

[0073] According to another aspect of this application, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the sensor data analysis method.

[0074] The sensor data analysis method, apparatus, equipment, medium, and program products provided in this application acquire evaluation indicators for target object elements, obtain multiple sensor data sets to be evaluated, with each sensor data set carrying multiple observation data sets. For each target object element in the sensor data set, the target object element is matched with a preset high-precision map based on its element information to obtain a matching result for each sensor data set. The matching result includes a single matching result corresponding to each observation data set. Based on the matching result, the indicator value of the target object element in each sensor data set with respect to the evaluation indicators is obtained, and the quality analysis result of the target object element in each sensor is obtained based on the indicator value. Considering the differences between individuals, individual indicators corresponding to different individuals are used, and quality evaluation is performed separately for each individual based on multiple observation data sets for each sensor data set. This effectively improves the accuracy of sensor data quality analysis and provides support for the efficient utilization of sensor data during high-precision map updates. Attached Figure Description

[0075] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0076] Figure 1 A schematic diagram illustrating a possible index system provided for an embodiment of this application;

[0077] Figure 2 A schematic flowchart illustrating a sensor data analysis method provided in an embodiment of this application;

[0078] Figure 3 This is one of the schematic diagrams illustrating the matching results between map data and observation data in an example of this application;

[0079] Figure 4 This is the second illustration of the matching results between map data and observation data in one example of this application;

[0080] Figure 5 This is the third illustration of the matching results between map data and observation data in one example of this application;

[0081] Figure 6 This is a schematic diagram of the structure of a sensor data analysis device provided in an embodiment of this application;

[0082] Figure 7This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application;

[0083] Figure 8 This is a block diagram of a terminal device provided in an exemplary embodiment of this application. Detailed Implementation

[0084] The embodiments of this application will be explained below in conjunction with application scenarios. The sensor data analysis method provided in the embodiments of this application can be applied to the application scenario of intelligent driving, and more specifically, it can be applied to the application scenario of autonomous driving based on vehicle cloud computing. For example, the execution subject of the method provided in the embodiments of this application can be a server, and more specifically, for example, a server of a high-precision map manufacturer. The following will introduce the method provided in the embodiments of this application as the execution subject of the server.

[0085] Optionally, a network connection is established between the server, terminal devices, and intelligent vehicles. The terminal devices collect multi-source sensor data, which can be understood as data from multiple types of sensors, including radar sensors, ultrasonic sensors, lidar sensors, and optical sensors (e.g., video cameras). The data can be sensor semantic data. The server receives the multi-source sensor data transmitted from the terminal devices, performs quality analysis on the data, selects the best sensor data as the data source to update the high-precision map data, and then synchronizes the updated high-precision map data to the intelligent vehicles. The intelligent vehicles can then use the high-precision map data to assist in autonomous driving. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, and cloud computing.

[0086] In related technologies, the semantic data quality evaluation schemes for multi-source sensors described above are typically based on reference data. The main differences lie in the selection of reference data, the selection of quality factors, and the quantification of quality factors. Different evaluation schemes are generally based on the assumption that there is global consistency in capability among similar individuals (i.e., object elements in the map). They treat all similar individuals as a whole for evaluation, without considering the differences between individual sensor semantic data. They fail to use a single overall indicator to uniformly describe the data quality, resulting in generally unobjective evaluations and inaccurate analysis results.

[0087] In view of this, the inventive concept implemented in this application provides a quality evaluation index system based on the stability of individual capabilities, such as Figure 1As shown, this allows for the use of (individual) evaluation indicators in the quality evaluation index system to obtain individual quality, and then further analyze the quality of sensor data based on individual quality. This enables efficient evaluation of sensor data quality, facilitating better utilization of sensor data based on its quality. For example, when using high-quality sensor data to update high-precision maps, the accuracy of high-precision maps can be effectively guaranteed, meeting user needs and improving user experience.

[0088] In this embodiment, individual precision refers to the matching between the spatial geometry (e.g., location) of a certain object element in the target object elements in the sensor data and the true value (map elements in the preset high-precision map), which is used to represent the accuracy of spatial geometric information.

[0089] Individual precision metrics: For precision, the expected value (mean) of individual precision represents the systematic error level of the data source, while the standard deviation represents the noise level of the data source. Specific refined metrics can be used as follows: the smaller the value of the individual precision metric, the higher the precision of the data.

[0090] 1) The mean of individual precision: represents the systematic error of the data source; the smaller the value, the smaller the systematic error of the data source;

[0091] The standard deviation of the mean of individual precision represents the stability of the systematic error of the data source; the smaller the value, the more stable the systematic capability of the data source and the more consistent the level of data quality capability.

[0092] The mean of the standard deviation of individual precision: represents the randomness of the data source and data quality; the smaller the value, the less random the data quality, and the fewer times it takes to reach the convergence threshold.

[0093] The standard deviation of individual precision represents the fluctuation of randomness in the data source; the smaller the value, the smaller the difference in the number of times different individuals need to reach the convergence threshold.

[0094] 2) Individual Existence Indicators: Existence refers to the consistency between the existence of elements in the data and the true value. Specific detailed indicators for existence include:

[0095] The expected value (mean) of individual recall: represents the systemic capability level of the data source; the larger the value, the higher the systematic recall and the higher the quality level.

[0096] The standard deviation of individual recall represents the stability (range of fluctuation) of the system capability of the data source; the smaller the value, the smaller the difference in recall between individuals, and the more stable the system capability.

[0097] The expected value (mean) of the individual false positive rate represents the system capability level of the data source; the smaller the value, the lower the systemic false positive rate and the higher the quality level.

[0098] The standard deviation of individual false positive rates represents the stability (fluctuation range) of the data source's system capability; the smaller the value, the smaller the difference in false positive rates between individuals, and the more stable the system capability.

[0099] 3) Attribute accuracy metrics: Attribute accuracy refers to the consistency between the descriptive information of elements in the data (i.e., the semantic information of the records), such as the consistency between the category and the true value. Similar to existence, specific detailed metrics for attribute accuracy can be as follows:

[0100] The mathematical expectation (mean) of attribute accuracy represents the system capability level of the data source; the larger the value, the higher the systematic attribute accuracy and the higher the quality level.

[0101] The standard deviation of attribute accuracy represents the stability (range of fluctuation) of the data source's system capability. The smaller the value, the smaller the difference in attribute accuracy between individuals, and the more stable the system capability.

[0102] As can be understood, in this embodiment, the mean of individual precision, that is, the average of the individual precisions among multiple observations of a certain individual in the sensor data, represents the error magnitude of that individual corresponding to the sensor data. The mean of the average of individual precisions, that is, the average of the individual precisions among all individuals in the sensor data, represents the overall error magnitude of the sensor data. Other detailed indicators are similar and will not be elaborated here.

[0103] The technical solution of this application and how it solves the above-mentioned technical problems will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that these specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0104] Please refer to Figure 2 , Figure 2 This is a flowchart illustrating the sensor data analysis method provided in the embodiments of this application. Taking a server as the executing entity, the method includes steps S201-S204.

[0105] Step S201: Obtain the evaluation indicators of the target object elements. The evaluation indicators shall include at least one of the following: individual accuracy indicators, individual existence indicators, and attribute correctness indicators.

[0106] Understandably, the target object element refers to the geographic element in the sensor data that needs to be evaluated for quality. It can be one or more, such as traffic signs, i.e., traffic signs, lane markings, traffic lights, guardrails, etc. In some embodiments, it can also be other geographic elements.

[0107] Compared to related technologies, current methods for evaluating sensor data quality typically assess all object elements within the data holistically. However, due to variations in scenarios and other factors, significant differences exist between individuals within the same category (e.g., signs). Based on common holistic evaluation methods, these individual differences are often overlooked, resulting in an inability to accurately describe data quality. This embodiment, however, fully considers the differences in capabilities between individuals. By defining individual evaluation indicators, it proposes an evaluation method based on the stability of individual capabilities. This further enhances the objectivity, accuracy, and comprehensiveness of data quality (capability) evaluation results, providing users with a stronger basis for assessing and understanding data quality and data source capabilities, as well as for data source selection and utilization.

[0108] In this embodiment, the evaluation index of the target object element (hereinafter referred to as the individual evaluation index) includes any one or any combination of individual accuracy index, individual existence index and attribute correctness index, which are used to evaluate the quality of a single object element in the sensor data.

[0109] In one implementation, individual evaluation indicators can be defined by the server. This embodiment uses multiple dimensions such as spatial geometric information, the existence of elements, and element description information to define individual evaluation indicators. The method of defining evaluation indicators in step S201 includes:

[0110] Define individual precision indices for target object elements based on their spatial geometric information; and / or define individual existence indices for target object elements based on their existence information; and / or define attribute correctness indices for target object elements based on their feature description information.

[0111] Understandably, spatial geometric information includes spatial location, such as the positional matching relationship between the target object element and the corresponding map element in the preset high-precision map. The existence of the target object element refers to whether there is a map element in the preset high-precision map that corresponds to the target object element. The element description information refers to whether the description of the element is consistent with the description of the corresponding map element in the preset high-precision map. Taking lane markings as an example, their attribute information can include type attributes and color attributes, such as dashed line type and white. The attribute information will be described in the element description information to identify the correctness of the attributes between the target object element and the corresponding map element.

[0112] Step S202: Obtain multiple sensor data to be evaluated, wherein each sensor data carries multiple observation data.

[0113] In this embodiment, the multiple sensor data to be evaluated are the sensor semantic data used for high-precision map updates. Optionally, the sensor data can be sensor semantic data or other types of data, and there are no restrictions on this.

[0114] Optionally, the sensor data may include one or more target object elements, wherein the target object elements may be traffic signs, i.e., traffic signs, lane markings, traffic lights, guardrails.

[0115] In this embodiment, in order to further improve the accuracy of sensor data quality evaluation, the individual quality of each sensor data is evaluated for the multiple observation data carried in each sensor data, and then the sensor data is evaluated as a whole by combining the quality of all individuals. Compared with the related technology that directly evaluates the sensor data, the sensor data evaluation results that take into account individual differences are more objective.

[0116] Step S203: For each target object element in the sensor data, match the target object element with a preset high-precision map according to the element information of the target object element to obtain the matching result of the target object element in each sensor data. The matching result includes the single matching result corresponding to each observation data.

[0117] Optionally, the feature information corresponding to the target object element includes location information and / or attribute information. Location information includes the coordinate values ​​of the target object element, which can be three-dimensional coordinate values ​​(x, y, z). Specifically, it can be the coordinate values ​​of a point on the target object element (e.g., the center point of a traffic sign, the shape point of a lane marking). Attribute information represents the characteristic description of the target object element, including information such as color, orientation, width, and type. For example, when the target object element is a traffic sign, its corresponding attribute information includes the sign's shape. As another example, when the target object element is a lane marking, its corresponding attribute information includes the marking color and marking type.

[0118] The coordinate values ​​of the target object element include coordinate values ​​in multiple directions, including the horizontal direction (equivalent to the x-axis direction), the vertical direction (equivalent to the y-axis direction), and the elevation direction (equivalent to the z-axis direction).

[0119] It should be noted that those skilled in the art can select appropriate high-precision map data as the preset high-precision map based on practical applications. For example, a high-precision map with the highest or consistent freshness in the same area as the sensor data to be evaluated can be selected as the approximate true value of the observed real world, or the high-precision map data with the closest data freshness, or the highest, consistent, or closest freshness among them, that is, at the latest time point, at the same time point, or at the closest time point. Therefore, using high-precision maps to analyze the quality of sensor data is equivalent to using the actual situation, i.e., the quality of the sensor data in the real world, which can ensure the determination of quality, i.e., the accuracy of the quality evaluation.

[0120] In one implementation, step S203, which matches the target object elements with a preset high-precision map based on the element information of the target object elements to obtain the matching result corresponding to the target object elements in each sensor data, may include the following steps:

[0121] Based on the location information corresponding to the target object element, determine whether there is a map element in the preset high-precision map that matches the target object element;

[0122] In response to the determination that the map feature does not exist, the existence matching result corresponding to the target object feature is determined to be that there is no target object feature matching the map feature; and / or,

[0123] In response to the determination that the map element exists, the location matching result between the target object element and the map element is determined based on the location information corresponding to the target object element; and / or, the attribute matching result between the target object element and the map element is determined based on the attribute information corresponding to the target object element.

[0124] In this embodiment, the matching of each observation data in the sensor data with the high-precision map data is used as the basis for judging the quality of sensor data (individual and overall) in order to analyze the quality of multi-source sensor data.

[0125] Specifically, based on the location information of the target object element, the system searches for a corresponding map element on a high-precision map and determines whether the corresponding map element exists in the high-precision map to confirm the existence of the target object element. It is understandable that multiple observations from the same sensor may yield different matching results; that is, some observations may show a presence match while others show no match. When a corresponding map element exists, the system further determines the location matching result based on the location information between the elements (the target object element and its corresponding map element), such as the distance difference between the x, y, and z axes. In one implementation, the location matching result can be divided into multiple levels, such as high, medium, and low matching degree, which can be adaptively set according to the magnitude of the distance difference. Furthermore, the system can also determine the attribute matching result based on the attribute information between the elements. For example, if the target object element is a road line with red attribute (description) information, while the map element's attribute information is white, this indicates an inconsistent attribute matching result.

[0126] Understandably, related technologies do not consider the stability of individual data from the sensor itself, leading to a lack of objectivity in the overall evaluation of the data. In other words, varying sensor stability will result in different degrees of deviation for data from multiple observations. Therefore, the matching results for each observation will also differ, and the greater the difference, the more unstable the data and the lower its quality. This embodiment aims to further improve the evaluation quality of sensor data by analyzing individual differences.

[0127] Step S204: Based on the matching results, obtain the index value of the target object element in each sensor data with respect to the evaluation index, and obtain the quality analysis result of the target object element in each sensor based on the index value.

[0128] In this embodiment, the evaluation index values ​​obtained based on the matching results incorporate the matching results of multiple observations of each sensor data, taking into account the differences between individuals and covering the quality evaluation of individuals, thus effectively improving the accuracy of sensor data quality analysis.

[0129] In one implementation, the individual accuracy index includes individual absolute accuracy and / or individual relative accuracy, and the matching result includes the positional matching result between the target object element and map elements in the preset high-precision map;

[0130] Step S204, based on the matching result, obtains the index value of the target object element in each sensor data with respect to the evaluation index, which may include the following steps:

[0131] For each sensor data point, based on the location matching result, obtain the absolute distance and / or relative distance between the target object element and the map element in each matching result;

[0132] Based on the absolute distance and / or relative distance, obtain the individual absolute accuracy and / or individual relative accuracy corresponding to the target object element in each sensor data. Based on the mean and standard deviation of the individual absolute accuracy and / or individual relative accuracy and the total number of the target object elements, obtain the mean and standard deviation of the individual absolute accuracy and / or individual relative accuracy. Based on the mean and standard deviation of the individual absolute accuracy and / or individual relative accuracy, obtain the index value of the target object element in each sensor data with respect to the individual accuracy index.

[0133] Understandably, the difference between absolute distance and relative distance can be attributed to the reference point. Absolute distance requires no specific reference point and is a fixed distance, such as the Himalayas at an altitude of over 8800 meters. This is absolute distance, and in this embodiment, it refers to the absolute distance calculated for a single target object element. Relative distance, on the other hand, is distance compared to a reference point, such as the distance between a tree and a person. In this embodiment, it refers to the relative distance calculated for two target object elements.

[0134] Specifically, the process of obtaining the absolute precision of an individual and its mean and standard deviation can be achieved by combining the following content and formulas:

[0135] Taking sensor data as an example, the first step is to statistically analyze the quality distribution (i.e., standard deviation and mean) of the absolute accuracy of each individual (target object element) after multiple observations and reports, using the following formula:

[0136]

[0137] In the formula, σ 1x This represents the standard deviation of the absolute precision of multiple observations reported by a single individual along the x-axis, where the x-axis can include horizontal, vertical, and height dimensions; μ 1x This represents the mean (mathematical expectation) of the absolute precision of multiple observations reported by a single individual; x i This represents the position of the element in the x-th match (i.e., an existence match that results in existence) along the x-th dimension; x i ' represents the position of the map feature corresponding to the i-th matched target object feature in the x-dimensional direction; x i -x i ' represents the absolute distance in the x dimension between the i-th matched target object feature and its corresponding map feature; n represents the total number of matched target object features reported by an individual across multiple observations.

[0138] For example, x1-x1' is the absolute vertical distance of the first reported sign (existence matching result is existence), and the distribution of capabilities is expressed by the mean (mathematical expectation) and standard deviation.

[0139] In some embodiments, matching results may be considered only in one dimension. In this embodiment, it is considered that some target object elements have feature information in other dimensions besides the horizontal dimension. In order to further improve the accuracy of data analysis, the evaluation is carried out by combining various dimensions (for example, taking the average of various dimensions).

[0140] The second step involves statistically analyzing and calculating the mathematical expectation of the absolute accuracy of each individual sensor to evaluate the stability of the data source capabilities of each sensor.

[0141] In the formula, μ 1μx σ is the mean of the absolute precision of all individual observations reported multiple times, along the x-axis, used to describe the systematic error of the data source; 1μx μ represents the standard deviation of the mean of absolute precision from multiple observations reported by all individuals, used to describe the stability of systematic errors in the data source; xi represents the mean of the absolute precision of the i-th individual in the x-dimensional region; m represents the total number of individuals.

[0142] The third step is to statistically analyze and calculate the standard deviation of the absolute accuracy of each individual sensor to evaluate the stability of random noise in each sensor data source.

[0143] In the formula, μ 1σx σ represents the mean of the standard deviations of the absolute precision of all individual observations reported multiple times, used to describe the level of random noise in the data source; 1σx This represents the standard deviation of the absolute precision of all individual observations reported multiple times, used to describe the random noise fluctuations in the data source; σ xi It represents the standard deviation of the absolute precision of the i-th individual in the x-dimensional dimension.

[0144] Specifically, the process of obtaining individual relative accuracy and its mean and standard deviation can be achieved by combining the following content and formulas:

[0145] Step 1: Analyze the quality distribution of relative precision for each individual's multiple observations and reports:

[0146]

[0147] In the formula, μ 2x σ represents the mean of the relative precision of multiple observation data reports for a single individual along the x-axis; 2xX represents the standard deviation of relative accuracy across multiple observations and reports from a single individual; n: the feature pairs of object features matched across multiple observations and reports from a single individual; i1 This represents the position of an object feature in the i-th pair of features along the x-axis; X i2 X′ represents the position of the other object feature in the i-th pair of features along the x-axis. i1 In a high-precision map, X′ represents the position of one feature in the i-th pair of features along the x-axis. i2 In a high-precision map, X represents the position of the other object feature in the i-th pair of features along the x-axis. i1 -X i2 X′ represents the relative distance of the i-th pair of features along the x-axis. i1 -X′ i2 This represents the relative distance of the i-th pair of features in the X dimension in a high-precision map.

[0148] Taking signage as an example, the relative precision of a pair of elements is calculated as follows:

[0149] X 11 -X 12 It is the vertical relative distance between the first and second reported signs; X′ 11 -X′ 12 It is the vertical relative distance between the first and second reported signboards and the corresponding map signs; (X) 11 -X 12 )-(X′ 11 -X′ 12 The relative accuracy of the first and second signs reported in the longitudinal direction. It is understood that the difference between the relative distance and the absolute distance mentioned above can be taken as the absolute value.

[0150] The second step involves statistically analyzing and calculating the expected value (mean) of individual relative accuracy to evaluate the stability of the source ability.

[0151] In the formula, μ 2μx σ represents the mean of the absolute precision values ​​of multiple observations reported by all individuals, used to describe the systematic error of the data source; 2μx μ represents the standard deviation of the mean of absolute precision from multiple observations reported by all individuals, used to describe the stability of systematic errors in the data source; xi Let m represent the mean of the relative precision of the i-th pair of individuals in the x-dimensional dimension, and m represent the total number of pairs of individuals.

[0152] Step 3: Statistically analyze the standard deviation of individual relative precision to evaluate the stability of the source random noise.

[0153] In the formula, μ 2σxThis represents the mean of the standard deviations of the relative precision across multiple observations reported by all individuals, along the x-axis, and is used to describe the level of random noise capability of the data source; σ 2σx This represents the standard deviation of the relative precision across multiple observations reported by all individuals, along the x-axis, used to describe the random noise fluctuations in the data source; σ xi represents the standard deviation of the absolute precision of the i-th individual pair in the x-dimensional dimension; m represents the total number of individual pairs.

[0154] In one embodiment, the sensor data includes target object elements corresponding to at least one element type, the element type including point type and / or line type.

[0155] Specifically, feature types include linear and point types. When calculating individual recall and / or attribute accuracy (i.e., the corresponding individual existence index and attribute accuracy index values), to improve the accuracy of quality analysis, the calculation process differs for object features corresponding to different feature types. For example, when the object feature is a traffic sign (i.e., the feature type is traffic sign), its corresponding object type is point type. As another example, when the object feature is a lane marking (i.e., the feature type is lane marking), its corresponding object type is linear type.

[0156] Further, in step S204, based on the matching result, obtaining the index value of the target object element in each sensor data with respect to the evaluation index may include the following steps:

[0157] For each type of element corresponding to the target object element, based on the matching result, the index value of the target object element corresponding to the element type in each sensor data with respect to the evaluation index is obtained.

[0158] In one implementation, the individual existence index includes individual recall rate and / or individual false positive rate, and the matching result includes the existence matching result between the target object element and map elements in the preset high-precision map;

[0159] The step of obtaining the index value of the target object element in each sensor data with respect to the evaluation index based on the matching result includes:

[0160] When the feature type is a point feature, for each sensor data, based on the existence matching result, a second number corresponding to the target object feature is obtained, and the individual recall rate of the target object feature is obtained according to the first number; and / or, when the feature type is a line feature, for each sensor data, based on the existence matching result, a first length of the target object feature is obtained, and the individual recall rate of the target object feature is obtained according to the first length; wherein, the first number is the number of all object features in the target object feature that match the map feature for each matching result, and the first length is the length of all object features in the target object feature that match the map feature for each matching result;

[0161] Based on the individual recall rate and the total number of target object elements, the mean and standard deviation of the individual recall rate are obtained. Then, based on the mean and standard deviation of the individual recall rate, the index value of the target object element in each sensor data point with respect to the individual existence index is obtained; and / or,

[0162] When the feature type is a point feature, for each sensor data, based on the existence matching result, a second number corresponding to the target object feature is obtained, and the individual false positive rate of the target object feature is obtained according to the second number; and / or, when the feature type is a line feature, for each sensor data, based on the existence matching result, a second length corresponding to the target object feature is obtained, and the individual false positive rate of the target object feature is obtained according to the second length; wherein, the second number is the number of all object features in the target object feature that have a match with the map feature in each matching result, and the corresponding map feature does not exist in the preset high-precision map; the second number is the number of all object features in the target object feature that have a match with the map feature in each matching result, and the corresponding map feature does not exist in the preset high-precision map;

[0163] Based on the individual false positive rate and the total number of target object elements, the mean and standard deviation of the individual false positive rate are obtained, and based on the mean and standard deviation of the individual recall rate, the index value of the target object element in each sensor data with respect to the individual existence index is obtained.

[0164] In this embodiment, the individual recall rate represents the probability that an individual is matched with a map feature present in a preset high-precision map. The higher the individual recall rate, the higher the consistency between the individual and the high-precision map in the sensor data, and the more stable the data. The higher the mean and standard deviation of the recall rates of all individuals in a certain sensor data, the higher the quality of the corresponding sensor data. The individual false positive rate represents the probability that an individual is mismatched (or misreported) with a map feature that does not exist in the preset high-precision map. The higher the individual false positive rate, the lower the consistency between the individual and the high-precision map in the sensor data, and the more unstable the data. The lower the mean and standard deviation of the false positive rates of all individuals in a certain sensor data, the higher the quality of the corresponding sensor data.

[0165] Specifically, the value of the individual existence index is calculated using the individual recall rate. (Taking sensor data as an example), the calculation process can be obtained from the following content and formula:

[0166] Step 1: Calculate the individual recall for each individual across multiple observations:

[0167] In the formula, D represents the total number of times an individual is observed; R point R represents the individual recall rate when the individual is a point feature; line The individual recall rate is represented when the individual is a linear feature; Cm1 represents the number of all matched object features when an individual is observed and reported multiple times; Lm1 represents the length of all matched object features when an individual is observed and reported multiple times; L1 represents the length of a linear individual.

[0168] Step 2: Statistically evaluate the stability of the source capability by calculating the expected recall rate for each individual. In the formula, μ R R represents the mean of individual recall rates across multiple observations and reports, used to describe the systematic capability of the data source; m represents the total number of individuals, and R0 represents the mean of individual recall rates across multiple observations and reports. i This represents the individual recall rate of the i-th object element.

[0169] Step 3: Statistically analyze the standard deviation of the recall rate for each individual to evaluate the stability of the source random noise. In the formula, σ R It represents the standard deviation of the individual recall rate across multiple observations and reports, used to describe the systematic stability of the data source.

[0170] In a further implementation, the value of the individual existence index can also be calculated using (or solely using) the individual false positive rate. (Taking a sensor data set as an example), the calculation process can be obtained based on the following content and formula:

[0171] Step 1: Calculate the individual false positive rate for each individual across multiple observations:

[0172] In the formula, D represents the total number of times an individual is observed; P point When representing an individual as a point feature, the individual false positive rate; P line When an individual is a linear feature, the false reporting rate is represented by Cm2; Cm2 represents the number of object features that are falsely reported (the map does not have a corresponding map feature, but there is a reported existence matching result) when an individual is observed and reported multiple times; Lm2 represents the length of the reported features in all false reports when an individual is observed and reported multiple times; L2 represents the length of a linear individual.

[0173] Step 2: Statistically analyze the expected value of individual false positive rates to evaluate the stability of source capabilities.

[0174] In the formula, μ P P represents the mean of the individual false reporting rates across multiple observations by all individuals, used to describe the systematic capability of the data source; i represents the individual false positive rate of the i-th object element; m represents the total number of individuals.

[0175] Step 3: Statistically analyze the standard deviation of individual false positive rates to evaluate the stability of the source random noise. In the formula, σ P The standard deviation of the individual false reporting rate represents the number of observations reported by all individuals, used to describe the systematic stability of the data source; μ P This represents the average false reporting rate for all individuals across multiple observations and reports.

[0176] In one implementation, the attribute correctness index includes individual attribute accuracy, and the matching result includes the attribute matching result between the target object element and the map element in the preset high-precision map.

[0177] Step S204, based on the matching result, obtains the index value of the target object element in each sensor data with respect to the evaluation index, which may include the following steps:

[0178] When the feature type is a point feature, for each sensor data, based on the attribute matching result, according to the third number of the target object features and the number of existing features in the existence matching result between the target object features and the map features, the attribute accuracy corresponding to the feature type is obtained.

[0179] When the feature type is linear, for each sensor data, based on the attribute matching result, according to the third length of the target object feature and the existence matching result between the target object feature and the map feature, the attribute accuracy corresponding to the feature type is obtained.

[0180] Wherein, the third number is the number of all object elements in the target object elements that match the map element and have the same attributes as the map element in each matching result; the third length is the length of all object elements in the target object elements that match the map element and have the same attributes as the map element in each matching result.

[0181] Based on the individual attribute accuracy rate and the total number of target object elements, the mean and standard deviation of the individual attribute accuracy rate are obtained, and the index value of the target object element in each sensor data with respect to the attribute accuracy index is obtained respectively.

[0182] In this embodiment, when matching target object elements with map elements, the attribute matching results between target object elements and map elements are further obtained to analyze the attribute accuracy of individuals and improve the accuracy of individual quality analysis results.

[0183] Specifically, the attribute accuracy index is calculated using the individual attribute accuracy. (Taking sensor data as an example), the calculation process can be obtained from the following content and formula:

[0184] Step 1: Calculate the attribute accuracy of each individual across multiple observations:

[0185] In the formula, H point This represents the accuracy of individual attributes when an individual is a point-type feature; H line This represents the accuracy of individual attributes when the individual is a linear feature; Cm represents the number of reports of a point feature that, after multiple observations and reports, matches the map feature (i.e., the existence match result is "existence"); C represents the number of map feature reports of a point feature that, after multiple observations and reports, match the map feature with the same attributes; Lm3 represents the length of map feature reports of a linear feature that, after multiple observations and reports, match the map feature with the correct attributes; L3 represents the length of linear feature reports of a linear feature that, after multiple observations and reports, match the map feature with the correct attributes. It's understandable that attributes must match; for example, a road line includes a color attribute, and the map feature's description of the color attribute is red, so the target object feature's description of the color attribute is also red.

[0186] To facilitate understanding of the embodiments of this application, a specific example of this embodiment will be described below. In this specific example, the individual absolute precision and individual recall of the target object element - signage - are used to evaluate different sensor data:

[0187] Step 1: Define evaluation metrics: 1) Metrics can be further defined by understanding sensor capabilities, such as the ability to provide GPS tracks, center point coordinates (x, y, z) of traffic signs, and the type of traffic signs; 2) Determine evaluation metrics: absolute accuracy of signs (horizontal, vertical, and elevation), relative accuracy of signs (horizontal), individual recall rate of signs, individual false positive rate of signs, and individual type (i.e., attribute) accuracy of signs.

[0188] Step 2: Acquire data from multiple sensors, such as Figure 3 As shown in the figure, ○ represents map data, that is, the sign in the reference map (i.e., map elements). The numbers ①-⑤ in ○ represent the data to be evaluated in a certain sensor data - the sign corresponding to 5 observations (i.e. target object elements). For ease of understanding, only 5 observations are used as an example here. In the actual application of the evaluation method, the more observations, the better to improve the accuracy of data evaluation. This is just an example of the evaluation process.

[0189] Step 3: For each sensor data point, match the signs in the data to be evaluated with the signs in the reference map. Taking one sensor data point as an example, combine... Figure 4 As shown in Table 1 below:

[0190]

[0191] As is understandable, in Table 1 above, "Y" indicates a successful match, meaning the existence match result is "existent," "√" corresponds to a match, and "×" corresponds to a no-match. From Figure 4 As can be seen from the signs on the reference map: Sign ①: 4 out of 5 observations matched; Sign ②: 3 out of 5 observations matched; Sign ③: 3 out of 5 observations matched; Sign ④: 5 out of 5 observations matched; Sign ⑤: 1 out of 5 observations matched. Considering that false alarms may occur in practical applications, for example, sign ①' (i.e., Figure 4 Medium gray fill ①) Observation five times, false alarms four times, sign ②' (i.e., Figure 4 (②) Five observations, one false alarm. At this point, the individual false alarm rate can be used for indicator evaluation.

[0192] After matching the above observation data, the absolute accuracy of each individual is calculated, and the quality distribution of absolute accuracy for each individual after multiple observations is statistically analyzed: the absolute positional difference between the data reported by each individual in each observation and the matched reference map features is calculated, such as... Figure 5 As shown, D21 is an individual instance, representing the absolute positional difference between the second reported data and the matched reference map feature. The statistical results are shown in Table 2 below.

[0193]

[0194] Based on Table 2, the quality distribution of individual absolute precision is calculated using the data reported by each individual after five observations:

[0195]

[0196]

[0197] By analogy, we can obtain: μ2 = 0.95, σ2 = 0.1871; μ3 = 0.9, σ3 ​​= 0.0432; μ4 = 1.05, σ4 = 0.2162; μ5 = 0.865, σ5 = 0.065.

[0198] Based on the statistical analysis of the expected value of the absolute precision of each individual, the stability of the source ability is evaluated:

[0199] The stability of the source random noise is evaluated by statistically analyzing the standard deviation of the absolute precision of each individual.

[0200] Furthermore, individual recall rates can be calculated using Table 1 or Table 2: R1 = 4 / 5 = 80%; R2 = 3 / 5 = 60%; R3 = 3 / 5 = 60%; R4 = 5 / 5 = 100%; R5 = 2 / 5 = 40%;

[0201] The expected value of individual recall is calculated and statistically evaluated to assess the stability of source capability. The stability of the source random noise is evaluated by statistically analyzing the standard deviation of individual recall rates.

[0202] Alternatively, when analyzing data quality using both individual precision and individual recall, the mean can be used for joint evaluation.

[0203] In the example above, the mathematical expectation of the individual absolute precision is... Represents the systematic error level and standard deviation of the data source. This represents the noise level of the data source, utilizing the N values ​​between sensor data. μ and N σThe corresponding index values ​​can be used to analyze the error level and noise level of sensor data. The same applies to individual recall. In this process, the differences between individuals are taken into account. For individual indicators corresponding to different individuals, and combined with multiple observation data of each sensor data, quality evaluation is carried out for each individual, which effectively improves the accuracy of sensor data quality analysis.

[0204] This application provides a corresponding sensor data analysis device, such as... Figure 6 As shown, the system includes: an indicator acquisition module 61, configured to acquire evaluation indicators for target object elements, wherein the evaluation indicators include at least one of the following: individual accuracy indicator, individual existence indicator, and attribute correctness indicator; a data acquisition module 62, configured to acquire multiple sensor data to be evaluated, wherein each sensor data carries multiple observation data; a matching module 63, configured to match the target object elements in each sensor data with a preset high-precision map according to the element information of the target object elements, so as to obtain the matching result corresponding to the target object elements in each sensor data, wherein the matching result includes a single matching result corresponding to each observation data; and a quality analysis module 64, configured to acquire the indicator value of the target object elements in each sensor data with respect to the evaluation indicators based on the matching results, and acquire the quality analysis result of the target object elements in each sensor based on the indicator value.

[0205] In one implementation, the evaluation index is defined in the following ways: defining an individual accuracy index of the target object element based on the spatial geometric information of the target object element; and / or defining an individual existence index of the target object element based on the existence information of the target object element; and / or defining an attribute correctness index of the target object element based on the element description information of the target object element.

[0206] In one embodiment, the element information includes location information and / or attribute information. The matching module 63 includes: a first determining unit, configured to determine whether there is a map element matching the target object element in the preset high-precision map based on the location information corresponding to the target object element; a second determining unit, configured to determine the existence matching result corresponding to the target object element as not having a map element in response to the determination result that there is no target object element matching the map element; and / or, a third determining unit, configured to determine the location matching result between the target object element and the map element based on the location information corresponding to the target object element in response to the determination result that there is a map element; and / or, determine the attribute matching result between the target object element and the map element based on the attribute information corresponding to the target object element.

[0207] In one embodiment, the individual accuracy index includes individual absolute accuracy and / or individual relative accuracy, and the matching result includes the positional matching result between the target object element and map elements in a preset high-precision map; the quality analysis module 64 includes:

[0208] The first acquisition unit is configured to, for each sensor data, acquire the absolute distance and / or relative distance between the target object element and the map element in each matching result, based on the location matching result; the second acquisition unit is configured to, based on the absolute distance and / or relative distance, acquire the individual absolute accuracy and / or individual relative accuracy corresponding to the target object element in each sensor data, and based on the mean and standard deviation of the individual absolute accuracy and / or individual relative accuracy and the total number of target object elements, acquire the mean and standard deviation of the individual absolute accuracy and / or individual relative accuracy, and based on the mean and standard deviation of the individual absolute accuracy and / or individual relative accuracy, acquire the index value of the target object element in each sensor data with respect to the individual accuracy index.

[0209] In one embodiment, the sensor data includes target object elements corresponding to at least one element type, the element type including point type and / or line type, and the quality analysis module 64 is specifically configured to, for each element type corresponding to the target object element, obtain the index value of the target object element corresponding to the element type in each sensor data with respect to the evaluation index based on the matching result.

[0210] In one implementation, the individual existence index includes individual recall rate and / or individual false positive rate, and the matching result includes the existence matching result between the target object element and map elements in a preset high-precision map; the quality analysis module 64 includes:

[0211] The third acquisition unit is configured to, when the feature type is a point feature, acquire a second number corresponding to the target object feature for each sensor data based on the existence matching result, and acquire the individual recall rate of the target object feature based on the first number; and / or, when the feature type is a line feature, acquire a first length of the target object feature for each sensor data based on the existence matching result, and acquire the individual recall rate of the target object feature based on the first length; wherein, the first number is the number of all object features in the target object feature that match the map feature for each matching result, and the first length is the length of all object features in the target object feature that match the map feature for each matching result; the fourth acquisition unit is configured to acquire the mean and standard deviation of the individual recall rate based on the individual recall rate and the total number of target object features, and acquire the index value of the target object feature in each sensor data with respect to the individual existence index based on the mean and standard deviation of the individual recall rate;

[0212] And / or, the fifth acquisition module is configured to, when the feature type is a point feature, acquire a second number corresponding to the target object feature for each sensor data based on the existence matching result, and acquire the individual false positive rate of the target object feature based on the second number; and / or, when the feature type is a line feature, acquire a second length corresponding to the target object feature for each sensor data based on the existence matching result, and acquire the individual false positive rate of the target object feature based on the second length; wherein, the second number is the number of the target object features for each matching result. The second number is the number of all object elements that match the map element but do not exist in the preset high-precision map. The second number is the number of all object elements in the target object element that match the map element but do not exist in the preset high-precision map for each matching result. The sixth acquisition module is configured to obtain the mean and standard deviation of the individual false positive rate based on the individual false positive rate and the total number of target object elements, and obtain the index value of the target object element in each sensor data with respect to the individual existence index based on the mean and standard deviation of the individual recall rate.

[0213] In one implementation, the attribute correctness index includes individual attribute accuracy, and the matching result includes attribute matching results between the target object element and map elements in a preset high-precision map; the quality analysis module 64 includes:

[0214] The seventh acquisition module is configured to, when the feature type is a point feature, acquire the attribute accuracy corresponding to the feature type for each sensor data based on the attribute matching result, according to the third number of the target object feature and the number of features that exist in the existence matching result between the target object feature and the map feature; and / or, the eighth acquisition module, when the feature type is a line feature, acquires the attribute accuracy corresponding to the feature type for each sensor data based on the attribute matching result, according to the third length of the target object feature and the number of features that exist in the existence matching result between the target object feature and the map feature;

[0215] Wherein, the third number is the number of all object elements in the target object elements that match the map element and have the same attributes as the map element in each matching result; the third length is the length of all object elements in the target object elements that match the map element and have the same attributes as the map element in each matching result.

[0216] Based on the individual attribute accuracy rate and the total number of target object elements, the mean and standard deviation of the individual attribute accuracy rate are obtained, and the index value of the target object element in each sensor data with respect to the attribute accuracy index is obtained respectively.

[0217] For relevant instructions, please refer to the corresponding text. Figures 2-5 The relevant descriptions and effects of the steps in the corresponding embodiments are understood, and will not be elaborated on here.

[0218] This application also provides an electronic device, such as... Figure 7 As shown, it includes: at least one processor 71 and a memory 72; the memory 72 stores computer-executable instructions; the at least one processor 71 executes the computer-executable instructions stored in the memory 72, causing the at least one processor 72 to execute the sensor data analysis method, wherein the memory 72 and the processor 71 are connected via a bus 73.

[0219] For relevant instructions, please refer to the corresponding text. Figures 2-5 The relevant descriptions and effects of the steps in the corresponding embodiments are understood, and will not be elaborated on here.

[0220] This application also provides a computer-readable storage medium storing computer-executable instructions. When a processor executes the computer-executable instructions, the sensor data analysis method described above is implemented.

[0221] The computer-readable storage medium can be ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc. See the corresponding explanations. Figures 2-5 The relevant descriptions and effects of the steps in the corresponding embodiments are understood, and will not be elaborated on here.

[0222] This application also provides a chip, including a memory and a processor. The memory stores a computer program, and the processor retrieves and runs the computer program from the memory to execute this application. Figures 2-5 The sensor data analysis method provided in any of the corresponding embodiments.

[0223] One embodiment of this application provides a computer program product, including a computer program that, when executed by a processor, implements this application. Figures 2-5 The sensor data analysis method provided in any of the corresponding embodiments.

[0224] Figure 8 This is a block diagram illustrating an exemplary embodiment of the present application of a terminal device 800, which may be a mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, medical device, fitness device, personal digital assistant, etc.

[0225] The terminal device 800 may include one or more of the following components: processing component 802, memory 804, power supply component 806, multimedia component 808, audio component 810, input / output (I / O) interface 812, sensor component 814, and communication component 816.

[0226] Processing component 802 typically controls the overall operation of terminal device 800, such as operations associated with display, telephone calls, data communication, camera operation, and recording. Processing component 802 may include one or more processors 820 to execute instructions to complete all or part of the steps of the methods described above. Furthermore, processing component 802 may include one or more modules to facilitate interaction between processing component 802 and other components. For example, processing component 802 may include a multimedia module to facilitate interaction between multimedia component 808 and processing component 802.

[0227] Memory 804 is configured to store various types of data to support operation on terminal device 800. Examples of this data include instructions for any application or method operating on terminal device 800, contact data, phonebook data, messages, pictures, videos, etc. Memory 804 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0228] Power supply component 806 provides power to various components of terminal device 800. Power supply component 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to terminal device 800.

[0229] Multimedia component 808 includes a screen that provides an output interface between terminal device 800 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touchscreen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors may sense not only the boundaries of touch or swipe actions but also the duration and pressure associated with the touch or swipe operation. In some embodiments, multimedia component 808 includes a front-facing camera and / or a rear-facing camera. When terminal device 800 is in an operating mode, such as a shooting mode or video mode, the front-facing camera and / or rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0230] Audio component 810 is configured to output and / or input audio signals. For example, audio component 810 includes a microphone (MIC) configured to receive external audio signals when terminal device 800 is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 804 or transmitted via communication component 816. In some embodiments, audio component 810 also includes a speaker for outputting audio signals.

[0231] I / O interface 812 provides an interface between processing component 802 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, power buttons, and lock buttons.

[0232] Sensor assembly 814 includes one or more sensors for providing status assessments of various aspects of terminal device 800. For example, sensor assembly 814 can detect the on / off state of terminal device 800, the relative positioning of components such as the display and keypad of terminal device 800, changes in the position of terminal device 800 or a component of terminal device 800, the presence or absence of user contact with terminal device 800, the orientation or acceleration / deceleration of terminal device 800, and temperature changes of terminal device 800. Sensor assembly 814 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 814 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, sensor assembly 814 may also include an accelerometer, a gyroscope, a magnetometer, a pressure sensor, or a temperature sensor.

[0233] Communication component 816 is configured to facilitate wired or wireless communication between terminal device 800 and other devices. Terminal device 800 can access wireless networks based on communication standards, such as WiFi, 3G, 4G, 5G, or other standard communication networks, or combinations thereof. In one exemplary embodiment, communication component 816 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 816 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0234] In an exemplary embodiment, the terminal device 800 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the functions described in this application. Figures 2-5 The method provided in any of the corresponding embodiments.

[0235] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 804 including instructions, which can be executed by a processor 820 of a terminal device 800 to perform the above-described method. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0236] This application also provides a non-transitory computer-readable storage medium, which, when the instructions in the storage medium are executed by the processor of a terminal device, enables the terminal device 800 to perform the above-described embodiments of this application. Figures 2-5 The method provided in any of the corresponding embodiments.

[0237] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0238] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the application disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

[0239] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A sensor data analysis method, characterized in that, include: The evaluation indicators for the target object elements are obtained, and the evaluation indicators include at least one of the following: individual accuracy indicators, individual existence indicators, and attribute correctness indicators; Acquire data from multiple sensors to be evaluated, where each sensor data point carries multiple observation data points; For each target object element in the sensor data, the target object element is matched with a preset high-precision map based on the element information of the target object element to obtain the matching result of the target object element in each sensor data. The matching result includes the single matching result corresponding to each observation data. Based on the matching results, the index value of the target object element in each sensor data with respect to the evaluation index is obtained, and the quality analysis result of the target object element in each sensor is obtained based on the index value. The index value is calculated based on the single matching result corresponding to each target object element in multiple observations. The individual accuracy index includes individual absolute accuracy and / or individual relative accuracy, and the matching result includes the position matching result between the target object element and the map element in the preset high-precision map. The step of obtaining the index value of the target object element in each sensor data with respect to the evaluation index based on the matching result includes: For each sensor data point, based on the location matching result, obtain the absolute distance and / or relative distance between the target object element and the map element in each matching result; Based on the absolute distance and / or relative distance, obtain the individual absolute accuracy and / or individual relative accuracy of the target object element in each sensor data; Based on the individual absolute accuracy and / or individual relative accuracy and the total number of target object elements, the mean and standard deviation of the individual absolute accuracy and / or individual relative accuracy are obtained, and based on the mean and standard deviation of the individual absolute accuracy and / or individual relative accuracy, the index value of the target object element in each sensor data with respect to the individual accuracy index is obtained respectively.

2. The method according to claim 1, characterized in that, The evaluation indicators are defined in the following ways: Define individual precision indices for target object elements based on their spatial geometric information; and / or define individual existence indices for target object elements based on their existence information; and / or define attribute correctness indices for target object elements based on their element description information.

3. The method according to claim 1 or 2, characterized in that, The element information includes location information and / or attribute information. The step of matching the target object elements with a preset high-precision map based on the element information of the target object elements to obtain the matching result corresponding to the target object elements in each sensor data includes: Based on the location information corresponding to the target object element, determine whether there is a map element in the preset high-precision map that matches the target object element; In response to the determination that the map feature does not exist, the existence matching result corresponding to the target object feature is determined to be that there is no target object feature matching the map feature; and / or, In response to the determination that the map element exists, the location matching result between the target object element and the map element is determined based on the location information corresponding to the target object element; and / or, the attribute matching result between the target object element and the map element is determined based on the attribute information corresponding to the target object element.

4. The method according to claim 1 or 2, characterized in that, The sensor data includes target object features corresponding to at least one feature type, wherein the feature type includes point type and / or line type. The step of obtaining the index value of the target object element in each sensor data with respect to the evaluation index based on the matching result includes: For each type of element corresponding to the target object element, based on the matching result, the index value of the target object element corresponding to the element type in each sensor data with respect to the evaluation index is obtained.

5. The method according to claim 4, characterized in that, The individual existence index includes individual recall rate and / or individual false positive rate, and the matching result includes the existence matching result between the target object element and the map element in the preset high-precision map; The step of obtaining the index value of the target object element in each sensor data with respect to the evaluation index based on the matching result includes: When the feature type is a point feature, for each sensor data, based on the existence matching result, a first number corresponding to the target object feature is obtained, and the individual recall rate of the target object feature is obtained according to the first number; and / or, when the feature type is a line feature, for each sensor data, based on the existence matching result, a first length of the target object feature is obtained, and the individual recall rate of the target object feature is obtained according to the first length; wherein, the first number is the number of all object features in the target object feature that match the map feature for each matching result, and the first length is the length of all object features in the target object feature that match the map feature for each matching result; Based on the individual recall rate and the total number of target object elements, the mean and standard deviation of the individual recall rate are obtained, and based on the mean and standard deviation of the individual recall rate, the index value of the target object element in each sensor data with respect to the individual existence index is obtained respectively. And / or, When the feature type is a point feature, for each sensor data, based on the existence matching result, a second number corresponding to the target object feature is obtained, and the individual false positive rate of the target object feature is obtained according to the second number; and / or, when the feature type is a line feature, for each sensor data, based on the existence matching result, a second length corresponding to the target object feature is obtained, and the individual false positive rate of the target object feature is obtained according to the second length; wherein, the second number is the number of all object features in the target object feature that have a match with the map feature in each matching result, but do not have a corresponding map feature in the preset high-precision map, and the second length is the length of all object features in the target object feature that have a match with the map feature in each matching result, but do not have a corresponding map feature in the preset high-precision map; Based on the individual false positive rate and the total number of target object elements, the mean and standard deviation of the individual false positive rate are obtained, and based on the mean and standard deviation of the individual false positive rate, the index value of the target object element in each sensor data with respect to the individual existence index is obtained.

6. The method according to claim 4, characterized in that, The attribute correctness index includes the individual attribute accuracy rate, and the matching result includes the attribute matching result between the target object element and the map element in the preset high-precision map. The step of obtaining the index value of the target object element in each sensor data with respect to the evaluation index based on the matching result includes: When the feature type is a point feature, for each sensor data, based on the attribute matching result, according to the third number of the target object features and the number of existing features in the existence matching result between the target object features and the map features, the attribute accuracy corresponding to the feature type is obtained. When the feature type is linear, for each sensor data, based on the attribute matching result, according to the third length of the target object feature and the existence matching result between the target object feature and the map feature, the attribute accuracy corresponding to the feature type is obtained. Wherein, the third number is the number of all object elements in the target object elements that match the map element and have the same attributes as the map element in each matching result; the third length is the length of all object elements in the target object elements that match the map element and have the same attributes as the map element in each matching result. The mean and standard deviation of the individual attribute accuracy rate are obtained based on the individual attribute accuracy rate and the total number of target object elements. The index value of the target object element in each sensor data with respect to the attribute accuracy index is obtained based on the mean and standard deviation of the individual attribute accuracy rate.

7. A sensor data analysis device, characterized in that, include: The indicator acquisition module is configured to acquire evaluation indicators for the elements of the target object. The evaluation indicators include at least one of the following: individual accuracy indicators, individual existence indicators, and attribute correctness indicators. The data acquisition module is configured to acquire data from multiple sensors to be evaluated, wherein each sensor data carries multiple observation data. The matching module is configured to match the target object elements in each sensor data with a preset high-precision map based on the element information of the target object elements, so as to obtain the matching result corresponding to the target object elements in each sensor data. The matching result includes the single matching result corresponding to each observation data. The quality analysis module is configured to obtain the index value of the target object element in each sensor data with respect to the evaluation index based on the matching result, and obtain the quality analysis result of the target object element in each sensor based on the index value. The index value is calculated based on the single matching result corresponding to each target object element in multiple observations. The individual accuracy index includes individual absolute accuracy and / or individual relative accuracy, and the matching result includes the position matching result between the target object element and the map element in the preset high-precision map. The quality analysis module includes: The first acquisition unit is configured to acquire, for each sensor data, the absolute distance and / or relative distance between the target object element and the map element in each matching result, based on the location matching result; The second acquisition unit is configured to acquire the individual absolute accuracy and / or individual relative accuracy corresponding to the target object element in each sensor data based on the absolute distance and / or relative distance; and to acquire the mean and standard deviation of the individual absolute accuracy and / or individual relative accuracy based on the individual absolute accuracy and / or individual relative accuracy and the total number of the target object elements; and to acquire the index value of the target object element in each sensor data with respect to the individual accuracy index based on the mean and standard deviation of the individual absolute accuracy and / or individual relative accuracy.

8. An electronic device, characterized in that, include: At least one processor and memory; The memory stores computer-executed instructions; The at least one processor executes computer execution instructions stored in the memory, causing the at least one processor to perform the sensor data analysis method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by the processor, implement the sensor data analysis method as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the sensor data analysis method as described in any one of claims 1 to 6.

Images