Position accuracy assessment method, device, equipment, vehicle, storage medium and product

Abstract

The present application relates to a method, apparatus, device, vehicle, storage medium, and product for evaluating position accuracy. The method comprises: obtaining trajectory data and visual data collected by a vehicle during driving; generating vector data corresponding to the vehicle's driving environment based on the trajectory data and visual data; the vector data includes at least one vector element; and evaluating the position accuracy of the vector element based on the trajectory data, visual data, and vector data to obtain a position accuracy evaluation result. The present application implements an internal evaluation of position accuracy based on the inherent characteristics of the trajectory data, visual data, and vector data. Position accuracy can be evaluated without the aid of a high-precision map, thereby improving the accuracy and efficiency of position accuracy evaluation.

Description

Technical Field

[0001] The present application relates to the field of intelligent transportation information technology, and in particular to a location accuracy assessment method, device, equipment, vehicle, storage medium and product. Background Art

[0002] In recent years, the development of in-vehicle intelligent systems and cloud computing has rapidly advanced autonomous driving technology from driver assistance to autonomous driving. Currently, autonomous driving technology primarily relies on high-definition maps (HD maps) for long-range perception and lane-level planning and guidance. However, HD maps are currently primarily produced through field surveying. Factors such as high labor costs, equipment prices, and demanding operating conditions make field surveying on a large scale suboptimal. Therefore, crowdsource mapping offers an alternative. Compared to high-precision and high-cost specialized data collection methods, crowdsource mapping technology is an ideal approach for large-scale, low-cost mapping. With the advancement of artificial intelligence algorithms and sensors, crowdsource mapping can achieve a high degree of automation in the mapping process. However, due to the diverse and uneven quality of crowdsource data and the limitations of existing mapping methods, the quality of maps generated through crowdsource approaches is inherently problematic. Consequently, these maps require rigorous verification before they can be released and used. Therefore, effective and effective crowdsource mapping quality assessment methods are crucial in the autonomous driving field. Summary of the Invention

[0003] The present application provides a location accuracy assessment method, apparatus, device, vehicle, storage medium and product.

[0004] In a first aspect, the present application provides a method for evaluating position accuracy, the method comprising:

[0005] Obtain trajectory data and visual data collected by the vehicle during driving;

[0006] Generate vector data corresponding to the vehicle's driving environment based on the trajectory data and the visual data; the vector data includes at least one vector element;

[0007] Based on trajectory data, visual data, and vector data, the position accuracy of vector elements is evaluated to obtain a position accuracy evaluation result.

[0008] Based on the above technical means, this application realizes internal evaluation of position accuracy based on the inherent characteristics of trajectory data, visual data, and vector data. It can realize the evaluation of position accuracy without the help of high-precision maps, thereby improving the accuracy and efficiency of position accuracy evaluation.

[0009] Furthermore, based on the trajectory data, visual data, and vector data, the position accuracy of the vector elements is evaluated to obtain a position accuracy evaluation result, including: performing feature extraction on the vector data, trajectory data, and visual data to obtain multiple feature information; performing feature mapping on the multiple feature information to predict the position accuracy corresponding to the vector elements to obtain a position accuracy evaluation result.

[0010] Based on the above technical means, the position accuracy evaluation results of the position accuracy prediction of vector elements in this application can provide important reference information for fields such as autonomous driving and intelligent transportation systems, and help improve the safety and reliability of the system.

[0011] Furthermore, multiple feature information includes trajectory feature information, positioning feature information, visual feature information, and geometric feature information and semantic feature information of vector elements; feature extraction is performed on vector data, trajectory data, and visual data to obtain multiple feature information, including: feature extraction on vector data and trajectory data to obtain geometric feature information and semantic feature information; feature extraction on trajectory data to obtain trajectory feature information and positioning feature information; feature extraction on visual data to obtain visual feature information.

[0012] According to the above technical means, the present application performs feature extraction on visual data, trajectory data, and vector data to obtain multiple feature information, and then obtains the position accuracy of the vector elements based on the mapping of multiple feature information, with high accuracy.

[0013] Furthermore, the semantic feature information includes a field of view distance; the field of view distance is the distance between the center point of the vector element and the driving track of the vehicle.

[0014] According to the above technical means, the present application derives the field of view distance based on vector data and trajectory data. Based on the field of view distance, the accuracy of the visual data can be better corrected, thereby improving the accuracy of position accuracy prediction based on visual data.

[0015] Furthermore, the position accuracy assessment result includes the accuracy evaluation result of the vector element and the position deviation; the method also includes: in response to the accuracy evaluation result of the vector element being inaccurate, adjusting the position of the vector element based on the position deviation of the vector element to obtain adjusted vector data; based on the adjusted vector data, constructing a map corresponding to the driving environment.

[0016] According to the above technical means, the positions of the vector elements in the adjusted vector data of this application are adjusted, so the accuracy is higher, and thus the scene map constructed based on the adjusted vector data is also higher.

[0017] Furthermore, before performing feature mapping on multiple feature information, predicting the position accuracy corresponding to the vector elements, and obtaining the position accuracy evaluation result, the method also includes: based on the trajectory feature information in the multiple feature information, identifying the scene of the driving environment to obtain the scene category; based on the scene category, performing feature mapping on multiple feature information, predicting the position accuracy corresponding to the vector elements, and obtaining the position accuracy evaluation result.

[0018] According to the above technical means, the position accuracy of the vector elements based on the scene category in this application will be more intelligent and the prediction accuracy will be higher.

[0019] Furthermore, the trajectory feature information includes at least one trajectory point, each trajectory point corresponds to a star search number and a positioning state; based on the trajectory feature information in the multiple feature information, the scene of the driving environment is identified to obtain the scene category, including: determining the mean star search number of at least one star search number, and statistically analyzing the discrete distribution of at least one positioning state; in response to the mean star search number being greater than a first threshold and the discrete distribution of at least one positioning state being located in a first preset distribution set, determining that the driving environment is in an open scene; in response to the mean star search number being less than a second threshold and the ratio of the discrete distribution of at least one positioning state being located in a second preset distribution set being greater than a preset ratio, determining that the driving environment is in a satellite denial scene.

[0020] According to the above technical means, this application distinguishes the current vehicle driving scene based on the number of satellite searches and positioning status in the trajectory feature information, and then predicts the position accuracy of vector elements in different scenes with high accuracy.

[0021] Furthermore, based on the trajectory data, visual data, and vector data, the position accuracy of the vector elements is evaluated to obtain a position accuracy evaluation result, including: using a preset position accuracy evaluation model, based on the trajectory data, visual data, and vector data, the position accuracy of the vector elements is evaluated to obtain a position accuracy evaluation result.

[0022] According to the above technical means, this application is based on a trained preset position accuracy evaluation model, which can directly evaluate the position accuracy of vector elements based on trajectory data, visual data and vector data. It can realize the evaluation of position accuracy without the help of true value elements in high-precision maps, thereby improving the efficiency of position accuracy evaluation.

[0023] Furthermore, the preset position accuracy evaluation model is set with corresponding model parameters for different scenarios: based on trajectory data, visual data, and vector data, the position accuracy of the vector elements is evaluated to obtain a position accuracy evaluation result, including: determining the scene category of the vehicle's driving environment; using the preset position accuracy evaluation model, calling the model parameters that match the scene category, and based on trajectory data, visual data, and vector data, evaluating the position accuracy of the vector elements to obtain a position accuracy evaluation result.

[0024] According to the above technical means, the model parameters of the model for position accuracy evaluation of vector elements in this application vary according to different scene categories, can better adapt to changes in scenes, and the accuracy of position accuracy evaluation is higher.

[0025] Furthermore, before obtaining the position accuracy evaluation result, the method further includes: obtaining sample data; the sample data includes trajectory sample data, visual sample data, and vector sample data matching the trajectory sample data and the visual sample data; using the position accuracy evaluation model to be trained, based on the sample data, predicting the sample position accuracy of the sample vector element to obtain the sample position prediction result; adjusting the model parameters of the position accuracy evaluation model to be trained based on the sample prediction result and the target prediction result corresponding to the sample data to obtain the preset position accuracy evaluation model.

[0026] According to the above technical means, the preset neural network model of this application is obtained by batch training the neural network using sample data. The model output result is the position accuracy evaluation result, which is used directly subsequently, thereby improving the efficiency of position accuracy evaluation.

[0027] In a second aspect, the present application provides a position accuracy assessment device, the position accuracy assessment device comprising:

[0028] An acquisition module is used to acquire trajectory data and visual data collected by the vehicle during driving;

[0029] A generation module, configured to generate vector data corresponding to the vehicle's driving environment based on the trajectory data and the visual data; the vector data includes at least one vector element;

[0030] The evaluation module is used to evaluate the position accuracy of vector elements based on trajectory data, visual data, and vector data to obtain a position accuracy evaluation result.

[0031] In a third aspect, the present application provides an electronic device, which includes: a processor, a memory, and a communication bus; the processor implements the above-mentioned position accuracy assessment method when executing the running program stored in the memory.

[0032] In a fourth aspect, the present application provides a vehicle comprising the above-mentioned electronic device.

[0033] In a fifth aspect, the present application provides a computer-readable storage medium, characterized in that a computer program is stored thereon, and when the computer program is executed by a processor, the above-mentioned position accuracy assessment method is implemented.

[0034] In a sixth aspect, the present application provides a computer program product, including a computer program or instructions, characterized in that when the computer program or instructions are executed by a processor, the above-mentioned position accuracy assessment method is implemented.

[0035] Beneficial effects of this application:

[0036] (1) This application implements internal evaluation of position accuracy based on the inherent characteristics of trajectory data, visual data, and vector data. It can evaluate position accuracy without the help of high-precision maps, thereby improving the accuracy and efficiency of position accuracy evaluation.

[0037] (2) The position accuracy evaluation results of the position accuracy prediction of vector elements in this application can provide important reference information for fields such as autonomous driving and intelligent transportation systems, helping to improve the safety and reliability of the system.

[0038] (3) This application performs feature extraction on visual data, trajectory data, and vector data to obtain multiple feature information, and then maps the multiple feature information to obtain the position accuracy of the vector elements with high accuracy.

[0039] (4) In this application, the positions of the vector elements in the adjusted vector data are adjusted, so the accuracy is high. Therefore, the scene map constructed based on the adjusted vector data is also high in accuracy.

[0040] (5) The location accuracy of vector elements based on scene categories in this application will be more intelligent and the prediction accuracy will be higher.

[0041] (6) This application is based on a pre-trained location accuracy assessment model and can directly assess the location accuracy of vector elements based on trajectory data, visual data, and vector data. It can achieve location accuracy assessment without the help of true value elements in high-precision maps, thereby improving the efficiency of location accuracy assessment.

[0042] (7) The model parameters of the model used in this application to evaluate the position accuracy of vector elements vary according to different scene categories, and can better adapt to changes in the scene, and the accuracy of the position accuracy evaluation is higher. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1A flow chart of a method for evaluating position accuracy provided in an embodiment of the present application;

[0044] Figure 2 A schematic diagram of an exemplary process for obtaining a position accuracy evaluation result provided in an embodiment of the present application Figure 1 ;

[0045] Figure 3 A schematic diagram of an exemplary process for extracting feature information provided in an embodiment of the present application;

[0046] Figure 4 A schematic diagram of an exemplary position accuracy evaluation result provided in an embodiment of the present application;

[0047] Figure 5 A schematic diagram of an exemplary process for constructing an environment map provided in an embodiment of the present application Figure 1 ;

[0048] Figure 6 A schematic diagram of an exemplary process for obtaining a position accuracy evaluation result provided in an embodiment of the present application Figure 2 ;

[0049] Figure 7 A schematic diagram of an exemplary scene recognition process provided in an embodiment of the present application;

[0050] Figure 8 A schematic diagram of an exemplary process for obtaining a position accuracy evaluation result provided in an embodiment of the present application Figure 3 ;

[0051] Figure 9 A schematic diagram of an exemplary model training method provided in the embodiment of the present application Figure 1 ;

[0052] Figure 10 A schematic diagram of an exemplary model training method provided in the embodiment of the present application Figure 2 ;

[0053] Figure 11 A schematic diagram of an exemplary process for constructing an environment map provided in an embodiment of the present application Figure 2 ;

[0054] Figure 12 A flowchart of an exemplary method for evaluating position accuracy provided in an embodiment of the present application;

[0055] Figure 13 A schematic diagram of the structure of a position accuracy assessment device provided in an embodiment of the present application;

[0056] Figure 14 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0057] The following will describe the embodiments of the present application with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand the other advantages and effects of the present application from the contents disclosed in this specification. The present application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be understood that the preferred embodiments are only for the purpose of illustrating the present application and are not intended to limit the scope of protection of the present application.

[0058] In recent years, the development of in-vehicle intelligent systems and cloud computing has rapidly advanced autonomous driving technology from driver assistance to autonomous driving. Currently, autonomous driving technology primarily relies on high-definition maps (HD maps) for long-range perception and lane-level planning and guidance. HD maps are currently primarily produced through field surveying. High-precision maps obtained through high-precision sensor measurements offer high accuracy and are often used as reference ground truth data for algorithm development or map production (see patent CN116124178A, which discloses a multi-source map quality assessment method, system, and storage medium). Real-time kinematic (RTK), static measurement using the Global Navigation Satellite System (GNSS), and drone mapping can generate multi-source map data within a specific area, achieving centimeter-level accuracy. However, due to labor costs, high equipment prices, and demanding operating conditions, field surveying over larger areas is not an optimal solution. Therefore, multi-source mapping offers an alternative to field surveying. Compared to high-precision and high-cost specialized acquisition methods, multi-source mapping technology is an ideal approach for large-scale, low-cost mapping. With the development of artificial intelligence algorithms and sensors, crowdsource mapping can achieve a high degree of automation in the mapping process. However, due to the diverse sources of crowdsource data, uneven quality, and the limitations of existing mapping methods, the quality of maps generated through crowdsourcing is inherently problematic. Crowdsourced maps must undergo rigorous verification before they can be released and used. Therefore, effective and rational methods for evaluating the quality of crowdsourced maps are crucial in the autonomous driving field.

[0059] Currently, in response to the above needs, patent CN116124178A discloses a crowdsource map quality assessment method, system, and storage medium. These methods can perform quality assessments based on element comparisons between crowdsource map information sent from the cloud and road information collected by on-board equipment. This requires communication between cloud-side and vehicle-side data, but the scope of the test and assessment is limited to areas with a true value reference. The assessment process involves matching vehicle-side data with cloud-side data, which incurs a certain amount of computational overhead. Patent CN117930297A discloses a vehicle positioning method and system based on an on-board camera and map. These methods can calculate a first lateral distance and a second lateral distance based on different methods, and use the difference between the two lateral distances as an indicator of whether the actual residual matches the prediction error of the Long Short-Term Memory (LSTM) model as an indicator of whether the GNSS signal has deviated. This primarily improves the accuracy of vehicle positioning. However, methods for evaluating the positional accuracy of crowdsourced data without a high-precision basemap are still lacking. Methods for evaluating positional accuracy and correcting for positional deviations in scenarios without a basemap have not been fully explored. The paper "Positional Accuracy Evaluation Integrated with EgoSensors for Crowdsourced HD Map" assesses the quality of map features directly based on empirical rules, applying specific rules to map data for each scenario. However, the design of these rules relies heavily on subjective judgment, and their generalization capabilities need to be improved.

[0060] In view of the above technical problems, the present application provides a method for evaluating position accuracy, which is implemented by an electronic device. Figure 1 A flow chart of a method for evaluating position accuracy provided in an embodiment of the present application is shown as follows: Figure 1 As shown, the process includes the following steps S101 to S103:

[0061] Step S101: Acquire trajectory data and visual data collected during the vehicle's driving process.

[0062] In an embodiment of the present application, an electronic device can obtain trajectory data and visual data corresponding to a vehicle during driving. For example, the electronic device can obtain the trajectory data and visual data through onboard sensors, where the trajectory data can be obtained by a GNSS sensor and an inertial measurement unit (IMU) sensor, and the visual data can be obtained based on an onboard visual sensor.

[0063] For example, trajectory data comes from inertial navigation sensors (inertial measurement unit sensors), which include important features such as the vehicle's absolute position, speed, direction of travel, and the number of satellites captured by the GNSS sensor. These features reflect the vehicle's physical environment and the quality of satellite signals along the road. If the physical environment is obstructed or the satellite signal quality is poor, the vehicle may cause offsets in mapping elements and missed detections along that road.

[0064] For example, the electronic device may perform pre-processing on the trajectory data, such as denoising and smoothing, to improve the accuracy and reliability of the data.

[0065] For example, the visual data is collected by an on-board camera and includes environmental information such as road signs, lane lines, and traffic lights, and is intended to supplement the missing information in the trajectory data.

[0066] Exemplarily, the electronic device may perform pre-processing on the image data (visual data), such as grayscale conversion, binarization, denoising, etc., to facilitate subsequent feature extraction and recognition.

[0067] In the embodiments of the present application, the electronic device is a device with a location accuracy assessment function, which may be a tablet computer, a laptop computer, a PDA, a personal digital assistant (PDA), a desktop computer, etc. The exemplary electronic devices are not limited here.

[0068] Step S102: Generate vector data corresponding to the vehicle's driving environment based on the trajectory data and the visual data; the vector data includes at least one vector element.

[0069] In an embodiment of the present application, the electronic device may utilize a Simultaneous Localization and Mapping (SLAM) algorithm to generate vector data corresponding to the vehicle's driving environment based on trajectory data and visual data.

[0070] In an embodiment of the present application, the vector elements include at least arrows, lane boundary lines, road signs, and traffic facilities.

[0071] For example, vector data generation can be achieved by: extracting key feature points, such as turning points and stopping points, from trajectory data; using computer vision technology to extract key visual features, such as road signs, traffic lights, and obstacles, from image data; generating road vector data: combining trajectory key feature points and key visual features, using geometric modeling and interpolation algorithms to generate vector data, such as road centerlines and boundary lines. Smoothing and optimizing the generated vector data to improve data accuracy and usability; generating traffic element vector data: generating corresponding vector data based on the location and attribute information of identified traffic elements, such as traffic lights and intersections; integrating and correlating the traffic element vector data with the road vector data to construct a complete vector map of the driving environment.

[0072] For example, after the vector data is generated, the generated vector data may be post-processed, such as removing redundant data, repairing broken line segments, etc., and formatting and standardizing the data for subsequent applications.

[0073] Step S103: Evaluate the position accuracy of the vector element based on the trajectory data, the visual data, and the vector data to obtain a position accuracy evaluation result.

[0074] In an embodiment of the present application, after acquiring trajectory data, visual data, and vector data, the electronic device will evaluate the position accuracy of the vector elements based on their respective data features to obtain a position accuracy evaluation result.

[0075] For example, the electronic device may use a neural network model to construct a position accuracy assessment model, and then use the position accuracy assessment model to perform assessment to obtain a position accuracy assessment result.

[0076] For example, electronic devices can also match trajectory data with vector data to determine which vector elements the vehicle interacted with during travel (e.g., driving on a road, approaching a traffic sign, etc.); record location information at the time of matching, including Global Positioning System (GPS) coordinates, timestamp, and corresponding vector element type; and verify the matching results using visual data. The positional relationship between the vector elements and the images captured by the camera can be compared to determine the positional accuracy of the vector elements. Of course, image recognition, target detection, and other technologies can also be used to assist in the verification process.

[0077] Compared with the technical solutions in related technologies that require the use of high-precision maps, this application realizes internal evaluation of position accuracy based on the inherent characteristics of trajectory data, visual data, and vector data. It can realize the evaluation of position accuracy without the help of high-precision maps, thereby improving the accuracy and efficiency of position accuracy evaluation.

[0078] In some embodiments, as Figure 2 As shown, when the electronic device executes the above step S103, it may further include the following steps S201 and S202:

[0079] Step S201: extract features from vector data, trajectory data, and visual data to obtain multiple feature information.

[0080] In an embodiment of the present application, the electronic device performs feature extraction on the vector data, trajectory data, and visual data to obtain a plurality of feature information.

[0081] For example, electronic devices can perform feature extraction based on deep learning neural networks, or based on machine learning algorithms, or of course based on statistics. The feature extraction method can be set based on actual needs and application scenarios, and this application does not limit this.

[0082] In an embodiment of the present application, the electronic device can perform feature extraction on the vector data, trajectory data, and visual data after fusing them to obtain multiple feature information, or it can extract corresponding features based on a single data, for example, perform feature extraction on the visual data to extract visual feature information.

[0083] Step S202: Perform feature mapping on multiple feature information, predict the position accuracy corresponding to the vector elements, and obtain a position accuracy evaluation result.

[0084] In an embodiment of the present application, the electronic device can map multiple feature information into a feature space to form a feature vector, and can use appropriate feature scaling and normalization methods to ensure the consistency and comparability of the feature vector, select a suitable position accuracy assessment model for prediction, and obtain a position accuracy assessment result.

[0085] In this way, the predicted position accuracy evaluation results for the position accuracy of vector elements can provide important reference information for fields such as autonomous driving and intelligent transportation systems, helping to improve the safety and reliability of the system.

[0086] In some embodiments, the plurality of feature information includes trajectory feature information, positioning feature information, visual feature information, and geometric feature information and semantic feature information of vector elements; Figure 3 As shown, when the electronic device executes the above step S201, it can execute the following steps S301 to S303:

[0087] Step S301: extract features from vector data and trajectory data to obtain geometric feature information and semantic feature information.

[0088] In an embodiment of the present application, feature extraction is performed on vector data corresponding to each trajectory point in the trajectory data to obtain geometric feature information and semantic feature information.

[0089] In an embodiment of the present application, geometric feature information may include shape features, position features, and topological features; shape features may be the boundary shape of vector elements (such as roads, buildings, traffic signs, etc.), and their perimeter, area, aspect ratio, etc. may be calculated; for linear elements (such as roads), their length, curvature, direction, etc. may be calculated; the position of the vector element in the geographic coordinate system may be determined, including longitude and latitude, altitude, etc.; the relative position relationship between the vector element and other elements (such as adjacent roads and intersections) may be calculated; the connection relationship between vector elements may be analyzed, such as the intersection of roads, the adjacent relationship of buildings, etc.; and network structure features may be extracted, such as the connectivity of the road network, the degree of the node, etc.

[0090] In an embodiment of the present application, semantic feature information may include category information, functional information, and attribute information; wherein, the category of the vector element (such as road type, building purpose, etc.) is determined based on the attributes and metadata of the vector element; classification labels or codes are used to represent these category information; functional information: analyze the function or role of the vector element, such as the indicative meaning of a traffic sign.

[0091] Step S302: extract features from the trajectory data to obtain trajectory feature information and positioning feature information.

[0092] In an embodiment of the present application, the electronic device can extract features from the trajectory data to obtain trajectory feature information and positioning feature information. The trajectory feature information may include trajectory points, timestamps, and category information; the positioning feature information may include the number of satellite searches corresponding to each trajectory point.

[0093] Step S303: extract features from the visual data to obtain visual feature information.

[0094] In an embodiment of the present application, an electronic device may perform feature extraction on visual data to obtain visual feature information. For example, image features may be extracted using traditional methods such as Scale-Invariant Feature Transform (SIFT) and Speeded-Up Robust Features (SURF), or using deep learning methods such as pre-trained Convolutional Neural Network (CNN) models (such as VGG16, ResNet, and Inception).

[0095] In this way, feature extraction is performed on visual data, trajectory data, and vector data to obtain multiple feature information, and then the position accuracy of the vector elements is obtained based on the mapping of the multiple feature information, with high accuracy.

[0096] In some embodiments, the semantic feature information includes a field of view distance; the field of view distance is the distance between the center point of the vector element and the driving track of the vehicle.

[0097] In an embodiment of the present application, the semantic feature information also includes field of view distance, which is derived feature information based on vector data and trajectory data. The field of view distance is the shortest Euclidean distance between the center point of the vector element and the vehicle's driving trajectory.

[0098] like Figure 4 As shown, the arrow represents vehicle-side mapping element 41, the dashed rectangle represents vehicle-side mapping lane line element 42, and the dashed line represents trajectory 43. The shortest Euclidean distance from the center point of vehicle-side mapping element 41 to trajectory 43 is the field of view distance 44; the shortest Euclidean distance from the center point of vehicle-side mapping lane line element 42 to trajectory 43 is the field of view distance 44. Vehicle-side mapping elements and vehicle-side mapping lane line elements are vector elements.

[0099] Exemplarily, the field of view distance is the distance from the vehicle-mounted sensor to the landmark, and the positioning point of the moving vehicle is regarded as the position of the vehicle-mounted sensor. Exemplarily, the field of view distance can be the shortest Euclidean distance between the center point of the vector element and the vehicle's driving trajectory line.

[0100] In an embodiment of the present application, the field of view distance is derived based on the vector data and the trajectory data. Based on the field of view distance, the accuracy of the visual data can be better corrected, thereby improving the accuracy of the position accuracy prediction based on the visual data.

[0101] In some embodiments, the position accuracy assessment result includes the accuracy evaluation result of the vector element and the position deviation; Figure 5 As shown, the electronic device may further include the following steps S501 and S502:

[0102] Step S501 : In response to an accuracy evaluation result of a vector element being inaccurate, the position of the vector element is adjusted based on a position deviation of the vector element to obtain adjusted vector data.

[0103] In the embodiment of the present application, if the electronic device responds to the inaccurate accuracy evaluation result of the vector element, it can adjust the position of the vector element based on the position deviation of the vector element. Because the electronic device predicts the position accuracy of the vector element, when obtaining the position accuracy evaluation result, it also includes the position deviation of the vector element. Figure 4As shown, the true value element 45 and the true value lane line element 46 are the element true values, and the absolute position error 47 is the position deviation between the predicted position of the vector element and the element true value. The electronic device can adjust the position of the vector element based on the position deviation to obtain adjusted vector data.

[0104] Step S502: construct a map corresponding to the driving environment based on the adjusted vector data.

[0105] In an embodiment of the present application, after obtaining the adjusted vector data, the electronic device can construct a map corresponding to the driving environment based on the adjusted vector data, and then perform assisted driving based on the constructed map.

[0106] In this way, since the positions of the vector elements in the adjusted vector data are adjusted, the accuracy is higher, and thus the scene map constructed based on the adjusted vector data is also more accurate.

[0107] In some embodiments, before the electronic device performs step S302, Figure 6 As shown, the following steps S601 and S602 may also be performed:

[0108] Step S601: Based on the trajectory feature information in the plurality of feature information, the scene of the driving environment is identified to obtain a scene category.

[0109] In an embodiment of the present application, an electronic device identifies a driving scene based on trajectory feature information from multiple feature information to determine a scene category. Trajectory feature information may include the number of satellites searched, positioning status, direction, speed, and other information. The electronic device can perform scene recognition based on some or all of this information to determine a scene category. For example, the scene categories may be open (non-tunnel) or satellite-denied (tunnel).

[0110] Step S602: Based on the scene category, feature mapping is performed on multiple feature information to predict the position accuracy corresponding to the vector elements to obtain a position accuracy evaluation result.

[0111] In the embodiments of the present application, electronic devices can use different prediction methods for different scene categories. Considering that the data obtained in tunnel scenarios may be biased or less accurate, and of lower quality than data obtained in non-tunnel scenarios, some methods can be used to enhance the data quality. Alternatively, refined processing can be performed based on the obtained low-quality data before performing the prediction work for the position accuracy of the vector elements.

[0112] In this way, the location accuracy of vector features based on scene categories will be more intelligent and the prediction accuracy will be higher.

[0113] In some embodiments, the trajectory feature information includes at least one trajectory point, each trajectory point corresponds to a search number and a positioning state; when the electronic device executes step S601, Figure 7 As shown, the following steps S701 to S703 may be included:

[0114] Step S701: Determine a mean value of at least one star search number, and calculate a discrete distribution of at least one positioning state.

[0115] In an embodiment of the present application, the electronic device can obtain the number of searched stars and the positioning status of each of the at least one trajectory point.

[0116] Exemplarily, the positioning status may include different types, such as two-dimensional positioning, three-dimensional positioning, single-point positioning, differential positioning, etc. The type of positioning status may be limited based on actual needs and application scenarios, which is not limited in this application.

[0117] In an embodiment of the present application, after obtaining at least one star search number, the electronic device determines the mean of the at least one star search number to obtain the mean star search number, and calculates a discrete distribution of at least one positioning state. The discrete distribution of positioning states can provide information about the probability distribution of the current positioning state, enabling more accurate location determination.

[0118] Step S702: In response to the average number of search satellites being greater than a first threshold, and the discrete distribution of at least one positioning state being within a first preset distribution set, it is determined that the driving environment is in an open scene.

[0119] In the embodiment of the present application, if the average value of the number of satellite searches is greater than the first threshold value, and the discrete distribution of at least one positioning state is in the first preset distribution set, it is determined that the vehicle's driving environment is in an open scene. The first threshold value can be 10, of course, it can also be other values, which can be set based on actual needs and application scenarios, and this application does not limit this; the first preset distribution set can be Of course, it can also be other sets, which can be set based on actual needs and application scenarios, and this application does not limit this.

[0120] Step S703: In response to the average value of the number of satellite searches being less than a second threshold value, and the ratio of the discrete distribution of at least one positioning state being in a second preset distribution set being greater than a preset ratio, it is determined that the driving environment is in a satellite denial scenario.

[0121] In an embodiment of the present application, if the mean value of the number of satellite searches is less than a second threshold, and the ratio of the discrete distribution of at least one positioning state in the second preset distribution set is greater than a preset ratio, it is determined that the driving environment is in a satellite denial scenario. The second threshold can be 5, or other values. The second threshold can be dependent on the first threshold, or can exist independently. It can be set based on actual needs and application scenarios, and this application does not limit this. The second preset distribution set can be {1, 6}, or of course other sets. There is no intersection between the first preset distribution set and the second preset distribution set. The positioning states concentrated in the first preset distribution set are characterized as good signals, and the positioning states concentrated in the second preset distribution set are characterized as bad signals. Of course, the relationship between the first preset distribution set and the second preset distribution set can be set based on actual needs and application scenarios, and this application does not limit this. The preset ratio can be one-half, one-third, or other ratios. For example, the ratio can be set based on actual needs and application scenarios, and this application does not limit this.

[0122] In this way, based on the number of satellite searches and positioning status in the trajectory feature information, the current vehicle driving scene can be distinguished, and then the position accuracy of the vector elements can be predicted according to different scenes with high accuracy.

[0123] In some embodiments, when executing the above step S103, the electronic device may include the following steps: using a preset position accuracy evaluation model, based on trajectory data, visual data, and vector data, to evaluate the position accuracy of the vector element to obtain a position accuracy evaluation result.

[0124] In an embodiment of the present application, the electronic device can use a preset position accuracy assessment model to assess the position accuracy of vector elements based on trajectory data, visual data, and vector data. The preset position accuracy assessment model is a pre-trained model based on a neural network that has the function of assessing the position accuracy of vector elements.

[0125] In this way, based on the trained preset position accuracy assessment model, the position accuracy of vector elements can be evaluated directly based on trajectory data, visual data and vector data. The position accuracy can be evaluated without the help of true value elements in high-precision maps, thereby improving the efficiency of position accuracy assessment.

[0126] In some embodiments, the preset location accuracy assessment model is configured with corresponding model parameters for different scenarios; when the electronic device executes the above step S103, Figure 8 As shown, the following steps S801 and S802 may be included:

[0127] Step S801: Determine the scene category of the vehicle's driving environment.

[0128] In an embodiment of the present application, since the quality of trajectory data, visual data, and vector data obtained in different scenarios is different, the electronic device can determine the scene category of the vehicle's driving environment, and then perform subsequent processing based on the scene category, making the processing more accurate.

[0129] Step S802: Using a preset position accuracy assessment model, calling model parameters that match the scene category, and evaluating the position accuracy of vector elements based on trajectory data, visual data, and vector data to obtain a position accuracy assessment result.

[0130] In an embodiment of the present application, after obtaining the corresponding scene category, the electronic device can set different model parameters for the preset position accuracy evaluation model for the scene category of the vehicle's driving environment, taking into account data of different qualities. In this way, the preset position accuracy evaluation model is used to call the model parameters that match the scene category, and the position accuracy of the vector elements is evaluated based on the trajectory data, visual data, and vector data to obtain the position accuracy evaluation result.

[0131] In this way, the model parameters of the model for evaluating the position accuracy of vector elements vary according to different scene categories, which can better adapt to changes in the scene and improve the accuracy of the position accuracy evaluation.

[0132] In some embodiments, before the electronic device performs "using a preset position accuracy assessment model to assess the position accuracy of the vector element based on the trajectory data, the visual data, and the vector data to obtain a position accuracy assessment result", Figure 9 As shown, the following steps S901 to S903 may also be included:

[0133] Step S901 : Acquire sample data; the sample data includes trajectory sample data, visual sample data, and vector sample data matching the trajectory sample data and the visual sample data.

[0134] Exemplarily, the preparation of the training set (sample data) includes collecting samples containing trajectory data, visual data, vector data, and field of view distance-derived features from crowd-source data, ensuring that these samples cover different road scenarios and conditions, and labeling each sample data, including the true value of its position accuracy; performing feature selection steps, analyzing the initially extracted features, using statistical methods to screen out features that have a significant impact on the evaluation of position accuracy, further reducing the feature dimensions through dimensionality reduction technology, and retaining the features that are most important to the model prediction effect.

[0135] Step S902: using the position accuracy evaluation model to be trained, based on the sample data, predict the sample position accuracy of the sample vector element to obtain a sample position prediction result.

[0136] In an embodiment of the present application, the electronic device can construct the architecture of a neural network model (a position accuracy evaluation model to be trained), including an input layer, multiple hidden layers and an output layer, and select an appropriate loss function and optimization algorithm; then, the sample data is input into the position accuracy evaluation model to be trained, and the sample position accuracy of the sample vector elements is predicted to obtain the sample position prediction result.

[0137] Step S903: Based on the sample prediction results and the target prediction results corresponding to the sample data, the model parameters of the position accuracy evaluation model to be trained are adjusted to obtain a preset position accuracy evaluation model.

[0138] Exemplarily, the electronic device inputs the training set data into the neural network model, calculates the output result (sample position prediction result) through forward propagation, and calculates the error (loss information) between the model output result and the true value.

[0139] For example, based on the error between the model output result and the true value, the model parameters are updated through the back propagation algorithm, and the model is verified and evaluated through the validation set to identify and solve the overfitting or underfitting problem, and finally a preset position accuracy evaluation model is obtained.

[0140] Among them, the preset neural network model (preset position accuracy evaluation model) is obtained by batch training the neural network using a training set (sample data), and the model output result is the position accuracy evaluation result.

[0141] like Figure 10 As shown, an exemplary training method for a position accuracy evaluation model is provided. Exemplarily, the training method can be applied to a training device, which may include a data acquisition unit, a data association unit, a data matching subunit, a data visualization subunit, a scene recognition unit, and a position accuracy prediction unit including a model feature extraction subunit, a model training subunit, a position accuracy prediction unit, an output subunit, an adjustment subunit, and a loss function subunit.

[0142] Exemplarily, the training method may include the following steps S1001 to S1004, which are discussed herein in conjunction with the training device:

[0143] Step S1001: Obtain a training set.

[0144] Here, the training device will use the data acquisition unit to specify the data time range and route; the data acquisition unit can obtain the full set of matching data obtained by the same vehicle collection platform (on-board sensor) within the specified time range: trajectory data, vector data and image (visual) data.

[0145] The data association unit specifies the time length of a single sample. It can segment the full dataset according to the specified time length, creating data samples (sample data) at a specified scale. For example, by inputting the data time range and route parameters (start time, end time, route name, and mapping date), the corresponding paths for trajectory data, visual data, and vector data are obtained. Based on the corresponding paths for trajectory data, visual data, and vector data, the trajectory data, visual data, and vector data are retrieved from the corresponding storage areas and extracted and packaged according to a standardized data package organization format.

[0146] The data matching subunit develops corresponding string processing and matching logic to match the different timestamp assignment methods for trajectory data, visual data, and vector data. This allows multiple data samples of the same time length to be obtained. Since the timestamp assignment methods for trajectory data, visual data, and vector data are different, different data can be distinguished based on the timestamp string.

[0147] The data visualization sub-unit can obtain the full set of trajectory data, vector data and visual data obtained by the data acquisition unit from the same vehicle collection platform (on-board sensor) within a specified time range. It can realize the visualization function through interface interaction and realize the visualization of vector data, trajectory data and positioning data (visual data) at the same time.

[0148] Step S1002: scene recognition.

[0149] Here, the scene recognition unit can extract the number of search stars in the trajectory data ( )、Positioning status( ) Construct feature vector , for the feature vector, the variation of the feature vector in the window is determined in the form of a sliding window. Based on the variation of the feature vector in the sliding window (the time length of a single sample), it is determined whether the current trajectory is in a tunnel, that is, a satellite denial environment. Exemplarily, the determination method may include: S1, constructing a time series, constructing a target GNSS data feature sequence through a sliding window, and counting the statistical features of the sliding window (see the matching subunit in the above step); S2, performing feature statistics, counting the mean of the number of satellite searches, and counting the discrete distribution of the positioning state, and judging the open scene and the satellite denial scene according to the established rules. Exemplary rules may be: The mean is greater than the threshold 10, and Distributed in It is determined to be an open scene; The mean is less than 5, and Distributed in If the ratio is greater than 50%, it is determined to be a satellite denial link; S3, perform sliding window update and dynamic judgment, and repeat the above judgment by sliding the window with a fixed step size, and record the recognition results of each window. In this way, the corresponding scene recognition results can be obtained for multiple data samples of the same time length.

[0150] Step S1003: Position accuracy prediction.

[0151] Here, the model feature extraction subunit is used to extract a multidimensional feature space (trajectory features, vector features, visual features, and derived features (geometric and semantic features)) from trajectory data, visual data, and vector data (multi-source map data). This feature data is then serialized in a tensor format and input into a preset neural network model (the position accuracy prediction model to be trained). Trajectory features are obtained by processing trajectory data using a specific neural network feature extraction subunit (e.g., a one-dimensional convolutional network). Visual features are obtained by processing visual data collected by the vehicle's camera using a corresponding neural network (e.g., a two-dimensional convolutional network). Vector features are generated using simultaneous localization and mapping (SLAM) technology. Derived features (semantic features) primarily refer to field of view distance, which is the distance from the vehicle's sensor to a landmark, with the positioning point of the moving vehicle being considered the location of the vehicle's sensor.

[0152] In this way, a position accuracy evaluation model was constructed based on a neural network, which fully considered the temporal characteristics of the data and the relationship between multiple types of data, thereby improving the effect of position accuracy evaluation.

[0153] It should be noted that when performing position accuracy prediction, the scene recognition results are input as parameters into the position accuracy prediction unit. A specific position accuracy classifier (position accuracy prediction unit) is configured with corresponding parameters for different scene labels. This specific position accuracy classifier is pre-trained based on samples from the corresponding category. The position accuracy of each element is evaluated based on the characteristics of each scene. For example, each individual vehicle-side vector element is assigned a label of "accurate" or "inaccurate."

[0154] The output subunit decodes the trajectory, visual, and vector features obtained by the model feature extraction subunit, performs position accuracy classification and prediction, and writes the resulting position accuracy classification results into output data according to a fixed data type and stores them on a medium. The model feature extraction subunit extracts appropriate features for evaluation model construction, using different types of features such as statistical, frequency, and time domain features. Feature selection methods are also used to identify features that significantly impact quality evaluation. An appropriate evaluation model (the position accuracy assessment model to be trained) is then selected for construction and training. The model is then tuned based on a validation set for optimal performance, and then integrated with vehicle motion models and sensor fusion techniques for evaluation. Visual features use image processing techniques to calculate the geometric transformation between the image and the road coordinate system, thereby deriving the vehicle's lane position. This is then compared with trajectory data to detect sections of track drift and offset, helping to correct discrepancies between trajectory and vector mapping (vector elements).

[0155] Step S1004: model training.

[0156] Here, the model training subunit is used to train the preset neural network model (the position accuracy prediction model to be trained) based on the training sample data and the true value label, compare the predicted result with the true result to obtain the error, and back-propagate the error to each network layer of the model to adjust the weights and parameters of the network layer to obtain the preset position accuracy prediction model. Exemplarily, the training sample data can refer to the multi-source map data collected by the vehicle and the location information and semantic label information of the corresponding high-precision map true value elements. The semantic label information includes the classification information in the map ground mark, where if the type of ground mark is a lane line, it should include the classification of the real line and the dashed line. The preset neural network model is trained based on the sample multi-source map data and the corresponding high-precision map true value elements.

[0157] Among them, the adjustment subunit is used to adjust the feature weights mentioned in the model feature extraction subunit of the preset neural network model based on the differences between the sample crowd-source map data locations and the corresponding true value data locations, as well as the differences between the sample crowd-source map data attributes and the true value semantic labels (the model parameters of the to-be-trained location accuracy assessment model are adjusted based on the loss information to obtain the preset location accuracy assessment model).

[0158] The loss function subunit is used to calculate the cross entropy between the locations of the multi-source map data collected by the vehicle and the high-precision map location labels in the sample data to obtain the loss value (loss information).

[0159] The adjustment subunit is used to adjust the parameters of the preset neural network model through back propagation based on the loss value.

[0160] In this way, a trained preset neural network model (preset position accuracy evaluation model) can be obtained.

[0161] like Figure 11 As shown, an exemplary map generation method is provided, which may include the following steps S1101 and S1102:

[0162] Step S1101: Input trajectory data, visual data, and vector data into a preset neural network model, and output location accuracy based on crowdsourced map data.

[0163] Here, trajectory data and visual data are obtained first, and then vector data is obtained based on the trajectory data and visual data through synchronous positioning and mapping algorithms.

[0164] In this way, without the need for remote sensing imagery and field measurements, the position accuracy of road markings such as arrows and lane boundaries in the evaluation map can be evaluated on road sections in various scenarios, and each vector element (road sign) can be assigned a corresponding label based on the position accuracy prediction result.

[0165] Step S1102: assisting in the production of crowd-source maps according to the location accuracy based on crowd-source map data output by the model.

[0166] Here, the location accuracy output by the model is used to assist in the production of multi-source maps. In addition to outputting the location accuracy, the model also outputs the location deviation.

[0167] In this way, internal evaluation is achieved based on the characteristics of the crowd-source map data itself, reducing the dependence on high-precision maps in the crowd-source map production process and improving the efficiency of crowd-source map production. It also breaks away from the dependence of crowd-source map data quality assessment on the true value in high-precision maps, evaluates data quality based on the internal characteristics of the data, and saves map production costs.

[0168] like Figure 12 As shown, an exemplary implementation of a method for evaluating the location accuracy of a multi-source map is provided, including steps S1201 to S1205:

[0169] Step S1201: data acquisition.

[0170] Here, the vehicle's onboard sensors can acquire raw sensor data 121. This raw data includes raw trajectory data and raw visual data. Raw trajectory data is acquired based on GNSS and IMU sensors and includes important features such as the vehicle's absolute position, speed, direction of travel, and the number of satellites captured by the GNSS sensors. Raw visual data includes environmental information such as road signs, lane markings, and traffic lights. This data is intended to supplement missing information in the trajectory data and provide the model with richer environmental perception capabilities. The collected visual data is typically stored in image format and can assist in scene recognition.

[0171] Step S1202: Data extraction.

[0172] Here, the required trajectory data and visual data (multi-source sensor data 122) can be extracted from the original trajectory data and the original visual data. Because the original data may contain all data of all time periods, the required data needs to be extracted here. For example, the position, driving direction and number of satellites captured by the GNSS sensor included in the trajectory data within a specified time range can be extracted from the original trajectory data. The speed and the like can be omitted. For example, trajectory data (absolute position and number of satellites) and visual data (collected by the camera at a fixed frequency are required, and extraction means visual data that matches the trajectory data) are extracted from the original data. Which feature information to extract can be set based on actual needs and application scenarios, and this application does not limit this.

[0173] Here, synchronous positioning and mapping algorithms can also be used to determine corresponding vector data (vehicle-side high-freshness data 126: vectorization results 127 and trajectory data 128) based on trajectory data (positioning signal 123 and inertial navigation data 124) and visual data (visual information 125). The vector data includes road signs, lane lines, traffic facilities, etc. mapped from multiple sources, including geometric attributes such as the center of mass, perimeter, and area of the vector elements (road signs).

[0174] Step S1203: feature extraction.

[0175] Here, feature extraction is performed on the trajectory data, vector data, and visual data to obtain a feature space: geometric features 129 , semantic features 1210 , trajectory features 1211 , positioning features 1212 , and visual features 1213 .

[0176] Step S1204: scene classification.

[0177] Here, based on the number of satellites searched and the positioning status in the trajectory feature data, the corresponding scene category is determined. The scene categories include: non-tunnel scene 1214 (open scene) and tunnel scene 1215 (satellite denial scene).

[0178] Step S1205: Map features.

[0179] Here, based on the scene category, the position accuracy prediction model 1216 (preset position accuracy prediction model) of the corresponding parameters is called to map features, and then the geometric features 129, semantic features 1210, trajectory features 1211, positioning features 1212 and visual features 1213 are input to perform position accuracy prediction to obtain the position accuracy evaluation results of each vector element.

[0180] Step S1206: Evaluation.

[0181] It should be noted that, in addition to outputting the accuracy of each vector element, the preset position accuracy prediction model can also output information such as confidence and position deviation, thereby obtaining the position accuracy evaluation result 1217.

[0182] This application provides a method for evaluating position accuracy, which obtains trajectory data and visual data collected by a vehicle during driving; based on the trajectory data and visual data, generates vector data corresponding to the vehicle's driving environment; the vector data contains at least one vector element; based on the trajectory data, visual data, and vector data, the position accuracy of the vector element is evaluated to obtain a position accuracy evaluation result. This application implements an internal evaluation of position accuracy based on the inherent characteristics of the trajectory data, visual data, and vector data. It can evaluate position accuracy without the need for high-precision maps, thereby improving the accuracy and efficiency of position accuracy evaluation.

[0183] The embodiment of the present application provides a position accuracy evaluation device 13, such as Figure 13 As shown, including:

[0184] An acquisition module 131 is used to acquire trajectory data and visual data collected by the vehicle during driving;

[0185] A generating module 132 is configured to generate vector data corresponding to the vehicle's driving environment based on the trajectory data and the visual data; the vector data includes at least one vector element;

[0186] The evaluation module 133 is used to evaluate the position accuracy of the vector element based on the trajectory data, the visual data, and the vector data to obtain a position accuracy evaluation result.

[0187] In one embodiment of the present application, the evaluation module 133 is also used to extract features from vector data, trajectory data, and visual data to obtain multiple feature information; perform feature mapping on the multiple feature information, predict the position accuracy corresponding to the vector elements, and obtain a position accuracy evaluation result.

[0188] In one embodiment of the present application, multiple feature information includes trajectory feature information, positioning feature information, visual feature information, and geometric feature information and semantic feature information of vector elements; the evaluation module 133 is also used to perform feature extraction on vector data and trajectory data to obtain geometric feature information and semantic feature information; perform feature extraction on trajectory data to obtain trajectory feature information and positioning feature information; and perform feature extraction on visual data to obtain visual feature information.

[0189] In one embodiment of the present application, the semantic feature information includes the field of view distance; the field of view distance is the distance between the center point of the vector element and the driving trajectory of the vehicle.

[0190] In one embodiment of the present application, the position accuracy evaluation result includes the accuracy evaluation result of the vector element and the position deviation; the evaluation module 133 is also used to adjust the position of the vector element based on the position deviation of the vector element in response to the accuracy evaluation result of the vector element being inaccurate, to obtain adjusted vector data; based on the adjusted vector data, construct a map corresponding to the driving environment.

[0191] In one embodiment of the present application, the evaluation module 133 is also used to identify the scene of the driving environment based on the trajectory feature information in the multiple feature information to obtain the scene category; based on the scene category, feature mapping is performed on the multiple feature information to predict the position accuracy corresponding to the vector element to obtain the position accuracy evaluation result.

[0192] In one embodiment of the present application, the trajectory feature information includes at least one trajectory point, each trajectory point corresponds to a star search number and a positioning state; the evaluation module 133 is further used to determine the star search number mean of at least one star search number, and to statistically calculate the discrete distribution of at least one positioning state; in response to the star search number mean being greater than a first threshold value, and the discrete distribution of at least one positioning state being located in a first preset distribution set, it is determined that the driving environment is in an open scene; in response to the star search number mean being less than a second threshold value, and the discrete distribution of at least one positioning state being located in a second preset distribution set, the ratio is greater than a preset ratio, it is determined that the driving environment is in a satellite denial scene.

[0193] In one embodiment of the present application, the evaluation module 133 is further configured to evaluate the position accuracy of the vector element based on the trajectory data, the visual data, and the vector data using a preset position accuracy evaluation model to obtain a position accuracy evaluation result.

[0194] In one embodiment of the present application, the preset position accuracy evaluation model is set with corresponding model parameters for different scenarios; the evaluation module 133 is also used to determine the scene category of the vehicle's driving environment; using the preset position accuracy evaluation model, the model parameters matching the scene category are called, and the position accuracy of the vector elements is evaluated based on the trajectory data, visual data, and vector data to obtain the position accuracy evaluation result.

[0195] In one embodiment of the present application, the evaluation module 133 is also used to obtain sample data; the sample data includes trajectory sample data, visual sample data, and vector sample data matching the trajectory sample data and the visual sample data; using the position accuracy evaluation model to be trained, based on the sample data, the sample position accuracy of the sample vector element is predicted to obtain a sample position prediction result; based on the sample prediction result and the target prediction result corresponding to the sample data, the model parameters of the position accuracy evaluation model to be trained are adjusted to obtain a preset position accuracy evaluation model.

[0196] An embodiment of the present application provides a position accuracy assessment device, which obtains trajectory data and visual data collected by a vehicle during driving; based on the trajectory data and visual data, generates vector data corresponding to the vehicle's driving environment; the vector data contains at least one vector element; based on the trajectory data, visual data, and vector data, the position accuracy of the vector element is evaluated to obtain a position accuracy assessment result; using the above implementation scheme, the present application realizes internal evaluation of position accuracy based on the inherent characteristics of the trajectory data, visual data, and vector data, and can realize position accuracy assessment without the aid of high-precision maps, thereby improving the accuracy and efficiency of position accuracy assessment.

[0197] Figure 14 This is a structural diagram of an electronic device 14 provided in an embodiment of the present application. In practical applications, based on the same disclosed concept of the above embodiments, as Figure 14 As shown, the electronic device 14 of this embodiment includes: a processor 141 , a memory 142 and a communication bus 143 , and the memory 142 stores application code 143 .

[0198] The embodiment of the present application provides a vehicle, including Figure 14 The electronic device.

[0199] An embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon. The computer-readable storage medium stores one or more programs, which can be executed by one or more processors and applied to a parking space positioning device for a parking scenario. The computer program implements the position accuracy assessment method as described above.

[0200] An embodiment of the present application provides a computer program product, including a computer program or instructions. When the computer program or instructions are executed by the processor 141, the position accuracy assessment method as described above is implemented.

[0201] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage and optical storage) containing computer-usable program code.

[0202] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0203] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0204] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0205] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any changes or replacements that can be easily thought of by any technician familiar with this technical field within the technical scope disclosed in this application should be covered by the scope of protection of the present application.

Claims

1. A method for evaluating position accuracy, characterized in that: The method comprises: Obtain trajectory data and visual data collected by the vehicle during driving; Generate vector data corresponding to the vehicle's driving environment based on the trajectory data and the visual data using a simultaneous positioning and mapping algorithm; the vector data includes at least one vector element; the vector element includes at least one of an arrow, a lane boundary, a road sign, and a traffic facility; Using a preset position accuracy evaluation model, model parameters that match the scene category of the vehicle's driving environment are called, and based on the trajectory data, the visual data, and the vector data, the position accuracy of the vector elements is evaluated to obtain a position accuracy evaluation result; the preset position accuracy evaluation model is set with corresponding model parameters for different scenes.

2. The position accuracy evaluation method according to claim 1, characterized in that: The evaluating the position accuracy of the vector element based on the trajectory data, the visual data, and the vector data to obtain a position accuracy evaluation result includes: performing feature extraction on the vector data, the trajectory data, and the visual data to obtain a plurality of feature information; Feature mapping is performed on the multiple feature information, and the position accuracy corresponding to the vector elements is predicted to obtain the position accuracy evaluation result.

3. The position accuracy evaluation method according to claim 2, characterized in that: The plurality of feature information includes trajectory feature information, positioning feature information, visual feature information, and geometric feature information and semantic feature information of the vector elements; the feature extraction of the vector data, the trajectory data, and the visual data to obtain the plurality of feature information includes: Performing feature extraction on the vector data and the trajectory data to obtain the geometric feature information and the semantic feature information; Performing feature extraction on the trajectory data to obtain the trajectory feature information and the positioning feature information; Feature extraction is performed on the visual data to obtain the visual feature information.

4. The position accuracy evaluation method according to claim 3, characterized in that: The semantic feature information includes a field of view distance; the field of view distance is the distance between the center point of the vector element and the driving track of the vehicle.

5. The position accuracy evaluation method according to claim 1, wherein: The position accuracy evaluation result includes the accuracy evaluation result of the vector element and the position deviation; the method further includes: In response to an accuracy evaluation result of the vector element being inaccurate, adjusting the position of the vector element based on the position deviation of the vector element to obtain adjusted vector data; A map corresponding to the driving environment is constructed based on the adjusted vector data.

6. The position accuracy evaluation method according to claim 3, characterized in that: Before performing feature mapping on the plurality of feature information, predicting the position accuracy corresponding to the vector elements, and obtaining the position accuracy evaluation result, the method further includes: Based on the trajectory feature information among the plurality of feature information, the scene of the driving environment is identified to obtain a scene category; Based on the scene category, feature mapping is performed on the multiple feature information, and the position accuracy corresponding to the vector element is predicted to obtain the position accuracy evaluation result.

7. The position accuracy evaluation method according to claim 6, characterized in that: The trajectory feature information includes at least one trajectory point, each trajectory point corresponds to a search number and a positioning state; the scene of the driving environment is identified based on the trajectory feature information in the plurality of feature information to obtain a scene category, including: determining a mean value of at least one of the star search numbers, and calculating a discrete distribution of at least one of the positioning states; In response to the average value of the number of satellite searches being greater than a first threshold, and the discrete distribution of at least one of the positioning states being within a first preset distribution set, determining that the driving environment is in an open scene; In response to the mean value of the number of satellite searches being less than a second threshold value, and a ratio of the discrete distribution of at least one of the positioning states being located in a second preset distribution set being greater than a preset ratio, it is determined that the driving environment is in a satellite denial scenario.

8. The position accuracy evaluation method according to claim 1, wherein: The method further comprises: Acquire sample data; the sample data includes trajectory sample data, visual sample data, and vector sample data matching the trajectory sample data and the visual sample data; Using the position accuracy evaluation model to be trained, based on the sample data, predicting the sample position accuracy of the sample vector element to obtain a sample position prediction result; Based on the sample prediction results and the target prediction results corresponding to the sample data, the model parameters of the position accuracy evaluation model to be trained are adjusted to obtain the preset position accuracy evaluation model.

9. A position accuracy evaluation device, characterized in that: The device comprises: An acquisition module is used to acquire trajectory data and visual data collected by the vehicle during driving; a generation module configured to generate vector data corresponding to the driving environment of the vehicle based on the trajectory data and the visual data using a simultaneous positioning and mapping algorithm; the vector data comprising at least one vector element; the vector element comprising at least one of an arrow, a lane boundary, a road sign, and a traffic facility; An evaluation module is used to use a preset position accuracy evaluation model to call model parameters that match the scene category of the vehicle's driving environment, evaluate the position accuracy of the vector elements based on the trajectory data, the visual data, and the vector data, and obtain a position accuracy evaluation result; the preset position accuracy evaluation model is set with corresponding model parameters for different scenes.

10. An electronic device, characterized in that: The electronic device includes: a processor, a memory and a communication bus; when the processor executes the running program stored in the memory, the position accuracy assessment method according to any one of claims 1 to 8 is implemented.

11. A vehicle, characterized in that: Comprising the electronic device as claimed in claim 10.

12. A computer-readable storage medium, characterized in that A computer program is stored thereon, and when the computer program is executed by a processor, the position accuracy evaluation method according to any one of claims 1 to 8 is implemented.

13. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the position accuracy assessment method according to any one of claims 1 to 8 is implemented.

Images