Data correction method and device, equipment, storage medium and program product

Abstract

The invention discloses a data correction method and device, equipment, a storage medium and a program product, and relates to the technical field of intelligent driving, and the data correction method comprises the steps: obtaining an intelligent driving scene data set which comprises a plurality of frames of scene data; scene object recognition is carried out according to the intelligent driving scene data set, intelligent driving scene objects are obtained, and the intelligent driving scene objects comprise traffic participants and / or static map elements; and correcting the scene data corresponding to the intelligent driving scene object through a preset data smoothing algorithm to obtain a corrected intelligent driving scene data set. The intelligent driving scene data correction based on the data smoothing algorithm is realized, the problem of how to effectively correct the abnormal data in the intelligent driving scene data set is solved, and the smoothness and effectiveness of the intelligent driving scene data are ensured.

Description

Technical Field

[0001] The present application relates to the field of intelligent driving technology, and in particular to a data correction method, device, equipment, storage medium and program product. Background Art

[0002] Intelligent driving datasets are the basis for training various intelligent driving algorithm models. The diversity and scale of intelligent driving datasets are important factors that determine the performance of the models. In order to improve the effectiveness of model training, data is usually collected through different channels.

[0003] The working conditions of different vehicles vary greatly, and there are differences in positioning and perception. There may be coordinate outliers in the data set. For data collection scenarios where data is difficult to obtain, these data sets with coordinate outliers are particularly important and need to be corrected.

[0004] The above content is only used to assist in understanding the technical solution of this application and does not constitute an admission that the above content is prior art. Summary of the Invention

[0005] The main purpose of this application is to provide a data correction method, device, equipment, storage medium and program product, aiming to solve the technical problem of how to effectively correct abnormal data in intelligent driving scenario data sets.

[0006] To achieve the above objectives, the present application proposes a data correction method, which includes:

[0007] Acquire an intelligent driving scene dataset, where the intelligent driving scene dataset includes several frames of scene data;

[0008] Performing scene object recognition according to the intelligent driving scene dataset to obtain intelligent driving scene objects, wherein the intelligent driving scene objects include traffic participants and / or static map elements;

[0009] The scene data corresponding to the intelligent driving scene object is corrected by a preset data smoothing algorithm to obtain a corrected intelligent driving scene data set.

[0010] In one embodiment, before the step of correcting the scene data corresponding to the intelligent driving scene object using a preset data smoothing algorithm to obtain a corrected intelligent driving scene data set, the step further includes:

[0011] Determining whether the scene data meets a preset data abnormality condition, where the data abnormality condition includes one or more of a missing coordinate value, an infinite coordinate value, and an abnormal coordinate value range;

[0012] If the scene data meets the data abnormality condition, a step of correcting the scene data is performed.

[0013] In one embodiment, when the intelligent driving scene object is a traffic participant, the scene data at least includes position coordinate data.

[0014] The step of correcting the scene data corresponding to the intelligent driving scene object by a preset data smoothing algorithm to obtain a corrected intelligent driving scene data set includes:

[0015] According to a preset data abnormality condition, an abnormality detection is performed on the position coordinate data of the traffic participant to obtain abnormal position coordinates of the position coordinate data;

[0016] According to the data abnormality condition, abnormal traversal detection is performed on the position coordinates of the historical frames and / or future frames corresponding to the current frame in the position coordinate data;

[0017] If the presence of non-abnormal position coordinates is detected, weighted average calculation is performed on the abnormal position coordinates based on the non-abnormal position coordinates to obtain corrected position coordinates;

[0018] According to the corrected position coordinates, the position coordinate data of the traffic participant is corrected to obtain a corrected intelligent driving scene data set.

[0019] In one embodiment, after the step of performing abnormal traversal detection on the position coordinates of the historical frames and / or future frames corresponding to the current frame in the position coordinate data according to the data abnormality condition, the method further includes:

[0020] If abnormal position coordinates are detected, abnormal traversal detection is stopped and the position coordinate data is marked as abnormal data.

[0021] In one embodiment, when the intelligent driving scene object is a static map element, the scene data at least includes a reference point coordinate sequence.

[0022] The step of correcting the scene data corresponding to the intelligent driving scene object by a preset data smoothing algorithm to obtain a corrected intelligent driving scene data set includes:

[0023] performing anomaly detection on the reference point coordinate sequence of the static map element according to a preset data anomaly condition to obtain the abnormal reference point coordinates of the reference point coordinate sequence;

[0024] According to the data abnormality condition, abnormal traversal detection is performed on the reference point coordinates in the starting point and / or end point direction corresponding to the current reference point in the reference point coordinate sequence;

[0025] If a non-abnormal reference point coordinate is detected, a weighted average calculation is performed on the abnormal reference point coordinate based on the non-abnormal reference point coordinate to obtain a corrected reference point coordinate;

[0026] According to the corrected reference point coordinates, the reference point coordinate sequence of the static map element is corrected to obtain a corrected intelligent driving scene dataset.

[0027] In one embodiment, the step of performing scene object recognition based on the intelligent driving scene dataset to obtain the intelligent driving scene object includes:

[0028] Identify the scene object corresponding to each frame of scene data in the intelligent driving scene dataset;

[0029] According to the time sequence of each frame of scene data, the scene objects corresponding to each frame of scene data are associated to obtain the intelligent driving scene object.

[0030] In addition, to achieve the above-mentioned purpose, the present application also proposes a data correction device, which includes:

[0031] An acquisition module is used to acquire an intelligent driving scene data set, where the intelligent driving scene data set includes a plurality of frames of scene data;

[0032] an identification module, configured to identify scene objects based on the intelligent driving scene dataset to obtain intelligent driving scene objects, wherein the intelligent driving scene objects include traffic participants and / or static map elements;

[0033] The correction module is used to correct the scene data corresponding to the intelligent driving scene object through a preset data smoothing algorithm to obtain a corrected intelligent driving scene data set.

[0034] In addition, to achieve the above-mentioned purpose, the present application also proposes a data correction device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the data correction method described above.

[0035] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium and stores a computer program. When the computer program is executed by a processor, the steps of the data correction method described above are implemented.

[0036] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the steps of the data correction method described above are implemented.

[0037] The present application provides a data correction method, which first obtains an intelligent driving scene data set containing several frames of scene data in the intelligent driving scene to provide a data basis for subsequent scene object recognition and data correction; then, scene object recognition is performed based on the acquired intelligent driving scene data set to perform data correction processing on all scene objects in the intelligent driving scene; finally, the scene data of different intelligent driving scene objects are corrected by a data smoothing algorithm to obtain a corrected intelligent driving scene data set to correct abnormal data of different scene objects in the intelligent driving scene data set, thereby realizing intelligent driving scene data correction based on the data smoothing algorithm, solving the problem of how to effectively correct abnormal data in the intelligent driving scene data set, ensuring the smoothness and effectiveness of the intelligent driving scene data, and thus improving the accuracy of intelligent driving in complex driving environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0039] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0040] Figure 1 A flowchart of the first embodiment of the data correction method of this application is provided;

[0041] Figure 2 A flowchart of the second embodiment of the data correction method of this application is provided;

[0042] Figure 3 This is a schematic diagram of the module structure of the data correction device according to an embodiment of the present application;

[0043] Figure 4 Schematic diagram of the device structure of the hardware operating environment involved in the data correction method in the embodiment of the present application.

[0044] The purpose, features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0045] It should be understood that the specific embodiments described herein are merely used to explain the technical solutions of the present application and are not intended to limit the present application.

[0046] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0047] The main solution of the embodiment of the present application is: obtaining an intelligent driving scene data set, which includes several frames of scene data; performing scene object recognition based on the intelligent driving scene data set to obtain intelligent driving scene objects, which include traffic participants and / or static map elements; and correcting the scene data corresponding to the intelligent driving scene objects through a preset data smoothing algorithm to obtain a corrected intelligent driving scene data set.

[0048] In existing technologies, intelligent driving datasets are the foundation for training various intelligent driving algorithm models. The diversity and scale of these datasets are crucial factors influencing model performance. To improve model training, data is typically collected through various channels, such as through data collection vehicles or using data sent back by users.

[0049] Due to the diverse data sources and the significant differences in the vehicle environments and operating conditions, data quality cannot be fully guaranteed. This necessitates data inspection and correction across various dimensions. Due to differences in positioning and perception between vehicles, the data may contain some coordinate anomalies, such as the vehicle's position coordinates or the coordinates representing lane lines.

[0050] Outliers are typically characterized by missing coordinate values, infinite coordinate values, or abnormal coordinate values. Data from some intelligent driving scenarios is difficult to obtain, and the data from these scenarios is extremely valuable. To fully utilize the data and increase its value, it is necessary to repair outliers in important scenarios to ensure the scale of data for important core scenarios.

[0051] This application solves the problem of how to effectively correct abnormal data in the intelligent driving scenario data set based on the data smoothing algorithm, ensures the smoothness and effectiveness of the intelligent driving scenario data, and thus improves the accuracy of intelligent driving in complex driving environments.

[0052] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, mobile phone, etc., or an electronic device or data correction device capable of performing the above functions. The following uses a data correction device as an example to illustrate this embodiment and the following embodiments.

[0053] Based on this, the embodiment of the present application provides a data correction method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the data correction method of this application.

[0054] In this embodiment, the data correction method includes steps S10 to S30:

[0055] Step S10: obtaining an intelligent driving scene data set, where the intelligent driving scene data set includes a plurality of frames of scene data;

[0056] It should be noted that the intelligent driving scene dataset refers to the continuous multi-frame scene data monitored by the vehicle during driving in the intelligent driving scenario or the dataset related to the intelligent driving map. It is the basis for the intelligent driving system to make decisions. It can be collected by the vehicle that collects the data through on-board sensors (such as cameras, millimeter-wave radars and lidars, etc.), and obtained by the data sent back by the collection vehicle or the user.

[0057] It can be understood that collecting intelligent driving scene data collected during vehicle driving and constructing an intelligent driving scene dataset containing several frames of scene data provide a basis for subsequent scene object recognition and data correction.

[0058] Step S20: performing scene object recognition based on the intelligent driving scene dataset to obtain intelligent driving scene objects, where the intelligent driving scene objects include traffic participants and / or static map elements;

[0059] It should be noted that intelligent driving scene objects refer to key entities in intelligent driving scenes. These entities can be dynamically moving objects in traffic scenes, i.e., traffic participants (such as vehicles and people), or static objects in the traffic environment that correspond to intelligent driving maps, i.e., static map elements (such as lanes, traffic signs, fences, and streetlights). Intelligent driving maps can be maps constructed by real-time vehicle perception or maps acquired through vehicle networking. Intelligent driving scene objects reflect the dynamically moving and statically changing entities in intelligent driving scenes and are crucial for vehicles to understand the environment and make decisions.

[0060] It is understandable that technologies such as computer vision and machine learning can be used to analyze the collected data and identify traffic participants in the intelligent driving scenario, so as to determine the intelligent driving scene objects that need data repair, and perform data repair processing on different intelligent driving scene objects separately, thereby providing accurate environmental information for the intelligent driving system to make correct decisions.

[0061] Step S30: Correct the scene data corresponding to the intelligent driving scene object using a preset data smoothing algorithm to obtain a corrected intelligent driving scene data set.

[0062] It should be noted that the data smoothing algorithm refers to an algorithm that reduces noise and outliers in the data, making the data more regular and stable, thereby improving data quality.

[0063] It can be understood that the data smoothing algorithm is used to perform data correction processing on the scene data corresponding to the identified intelligent driving scene objects to obtain a corrected intelligent driving scene data set, so as to reduce abnormal data in the intelligent driving scene data set, improve the accuracy and reliability of the data, and ensure that the intelligent driving system can make decisions based on high-quality data.

[0064] In a feasible implementation, the step of performing scene object recognition based on the intelligent driving scene dataset to obtain the intelligent driving scene object includes:

[0065] Step S201, identifying the scene object corresponding to each frame of scene data in the intelligent driving scene dataset;

[0066] It should be noted that when the intelligent driving scene dataset contains data in image format, the scene objects can be identified by deep learning-based image recognition, which uses convolutional neural networks (CNN) to extract and classify features from images; or they can be identified using methods such as edge detection and color segmentation.

[0067] It can be understood that each frame of scene data in the intelligent driving scene data set is analyzed to identify the scene objects therein, such as vehicles, pedestrians, traffic signs, etc., and all scene objects appearing in each frame of scene data are extracted to provide basic data for subsequent association and analysis.

[0068] Step S202: Associating the scene objects corresponding to each frame of scene data according to the time sequence of each frame of scene data to obtain an intelligent driving scene object.

[0069] It should be noted that time sequence refers to the temporal sequence between frames, which is used to understand the dynamic changes of the scene and the motion trajectory of the object.

[0070] It can be understood that according to the time sequence of each frame of scene data, the scene objects identified in consecutive frames are associated to obtain the intelligent driving scene objects that appear continuously in different frames in the corresponding intelligent driving scene dataset, so as to construct the continuity of scene objects in the time series, and convert isolated in-frame information into continuous and dynamic scene understanding, thereby identifying the dynamic changes and states of different objects in the scene.

[0071] In this implementation, scene objects in each frame are identified from the intelligent driving scene dataset, and these objects are associated according to the time sequence to construct the dynamic changes of objects in the intelligent driving scene. This not only provides static information of the scene, but also provides dynamic information of object changes over time, which is used to effectively correct abnormal data in the intelligent driving scene dataset, and then train and optimize the intelligent driving algorithm, improve the system's understanding and response capabilities to complex traffic environments, and thus enhance the safety and efficiency of intelligent driving.

[0072] In a feasible implementation manner, before the step of correcting the scene data corresponding to the intelligent driving scene object by a preset data smoothing algorithm to obtain a corrected intelligent driving scene data set, the step further includes:

[0073] Step S31, determining whether the scene data meets a preset data abnormality condition, wherein the data abnormality condition includes one or more of missing coordinate values, infinite coordinate values, and abnormal coordinate value range;

[0074] It should be noted that data anomaly conditions refer to abnormal phenomena that may exist in the data set, such as missing, infinite or out-of-normal coordinate values. These anomalies will affect the accuracy and availability of the data.

[0075] It is understandable that before making data corrections, it is necessary to use pre-set data anomaly conditions to determine whether there are any anomalies in the scene data, such as missing coordinate values, infinity, or values ​​outside the normal range, to identify abnormal data in the data set so that subsequent corrections can be made to ensure the quality and reliability of the data set.

[0076] Step S32: If the scene data meets the data abnormality condition, the step of correcting the scene data is executed.

[0077] It is understandable that when data anomalies are detected in the scene data, the preset data smoothing algorithm is used to correct the abnormal data to correct the abnormal data in the scene data, improve the accuracy and availability of the data set, and provide more reliable input data for the intelligent driving system.

[0078] In this embodiment, after object recognition is performed on the intelligent driving scenario data set, anomalies in the data set are further detected and corrected. First, the abnormal points in the data set are identified through preset data anomaly conditions, and then these abnormal points are corrected by applying a data smoothing algorithm. This not only improves the accuracy of the data, but also enhances the robustness of the data set, enabling it to better serve the intelligent driving system. The corrected data set can provide more reliable input for the intelligent driving algorithm, thereby improving the system's decision-making quality and driving safety.

[0079] This embodiment provides a data correction method, which first obtains an intelligent driving scene data set containing several frames of scene data in the intelligent driving scene to provide a data basis for subsequent scene object recognition and data correction; then, scene object recognition is performed based on the obtained intelligent driving scene data set to perform data correction processing on all scene objects in the intelligent driving scene; finally, the scene data of different scene objects in the intelligent driving scene data set are corrected by a data smoothing algorithm to obtain a corrected intelligent driving scene data set to correct abnormal data of different scene objects in the intelligent driving scene data set, thereby realizing intelligent driving scene data correction based on the data smoothing algorithm, solving the problem of how to effectively correct abnormal data in the intelligent driving scene data set, ensuring the smoothness and effectiveness of the intelligent driving scene data, and thus improving the accuracy of intelligent driving in complex driving environments.

[0080] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the first embodiment can be referred to the above introduction and will not be described in detail later. Figure 2 , Figure 2 This is a flow chart of the second embodiment of the data correction method of this application.

[0081] In this embodiment, when the intelligent driving scene object is a traffic participant, the scene data includes at least position coordinate data. The step of correcting the scene data corresponding to the intelligent driving scene object using a preset data smoothing algorithm to obtain a corrected intelligent driving scene data set includes:

[0082] Step S301, performing anomaly detection on the position coordinate data of the traffic participant according to a preset data anomaly condition to obtain abnormal position coordinates of the position coordinate data;

[0083] It should be noted that location coordinate data refers to the geographic location of traffic participants in intelligent driving scenarios, and can include x, y, and z coordinates. Abnormal location coordinates refer to those that do not conform to normal driving or physical laws, which may be caused by sensor errors, data transmission errors, and other factors. Data anomalies can include missing coordinate values, infinity, or values ​​outside the normal range.

[0084] It is understandable that the preset data anomaly conditions are used to identify abnormal points in the vehicle position coordinate data, so as to identify abnormal values ​​in the position coordinate data and provide a processing target for subsequent data correction.

[0085] Step S302, performing abnormal traversal detection on the position coordinates of the historical frames and / or future frames corresponding to the current frame in the position coordinate data according to the data abnormality condition;

[0086] It should be noted that the position coordinate data of traffic participants in the intelligent driving scenario dataset usually exist in the form of a multi-frame data sequence that is continuous in time. It is necessary to associate the dynamic changes in the positions of traffic participants at different times to perform abnormal data repair processing.

[0087] It can be understood that the abnormal position coordinates are taken as the current frame, and the position coordinates corresponding to the current frame in the historical frame and / or future frame in the position coordinate data are traversed to identify whether there are non-abnormal values ​​in the time period before and after the current frame, and by checking the frame data before and after in the time series, the reference coordinates that can be used to correct the abnormal values ​​are found.

[0088] Step S303: If the presence of non-abnormal position coordinates is detected, weighted average calculation is performed on the abnormal position coordinates based on the non-abnormal position coordinates to obtain corrected position coordinates;

[0089] It can be understood that when the traversal detects the existence of non-abnormal position coordinates in the previous and next frame data in the time series, this non-abnormal position coordinate is used as the reference coordinate for correcting the abnormal value. By assigning different weights to the position coordinates of different frames and performing weighted averaging calculations, the position coordinate value for correction is obtained. This value is used to correct the abnormal data in the vehicle position coordinate data, so as to reduce the impact of the abnormal value through the weighted averaging method, ensure the smoothness of the vehicle position coordinate data, and improve the accuracy of the intelligent driving scenario data set.

[0090] Step S304: Correct the position coordinate data of the traffic participant according to the corrected position coordinates to obtain a corrected intelligent driving scenario data set.

[0091] It can be understood that the position coordinate data of traffic participants in the intelligent driving scenario dataset are updated based on the corrected position coordinates obtained by weighted average calculation to obtain a data-corrected intelligent driving scenario dataset, so as to gradually adjust the abnormal position coordinates of traffic participants in the intelligent driving scenario dataset until they are close to the position and status of traffic participants in the real environment that change over time, thereby improving the quality and availability of the entire dataset.

[0092] In this embodiment, anomaly detection is first used to identify outliers in the position coordinate data. Then, anomaly traversal detection is used to find non-outliers in the time series as a correction reference. Next, a weighted average calculation method is used to correct the outlier position coordinates, and finally, the vehicle position coordinate data in the intelligent driving scenario dataset is updated. Data repair processing based on the temporal relationship between vehicle position changes can significantly improve the accuracy and reliability of position data, reduce intelligent driving system misjudgments caused by anomalies, and thus enhance the safety and efficiency of the intelligent driving system.

[0093] In a feasible implementation manner, after the step of performing abnormal traversal detection on the position coordinates of the historical frames and / or future frames corresponding to the current frame in the position coordinate data according to the data abnormality condition, the step further includes:

[0094] Step S3021: If abnormal position coordinates are detected, abnormal traversal detection is stopped, and the position coordinate data is marked as abnormal data.

[0095] It should be noted that abnormal data refers to data points that do not conform to the expected pattern or range. They may be caused by sensor failure, data transmission errors or other abnormal factors, and do not meet the conditions to serve as reference data for optimizing intelligent driving algorithms.

[0096] It is understandable that when the abnormal traversal detection fails to find non-abnormal position coordinates in the historical frame and / or future frame, it indicates that the position coordinate data of the traffic participant may have abnormalities that cannot be corrected by the existing data. The data points that cannot be corrected by the data smoothing algorithm are identified and marked as abnormal data so that these data can be specially processed or excluded when the intelligent driving algorithm is optimized in the future to prevent them from causing adverse effects on the intelligent driving system.

[0097] In this embodiment, when abnormal traversal detection cannot find non-abnormal position coordinates in historical frames and / or future frames, the system will stop further traversal detection and mark these position coordinate data as abnormal, which helps to maintain the integrity and reliability of the data set and ensure that the intelligent driving system does not make wrong decisions due to abnormal data. After marking the abnormal data, you can choose to further analyze the data to find the cause of the abnormality, or exclude them in the data preprocessing stage, thereby improving the performance and safety of the intelligent driving system.

[0098] For example, the process of performing a smoothing algorithm on abnormal points of the traffic participant position coordinates in the data includes the following steps:

[0099] Step 1. Identify and process each frame in time, extract information about all traffic participants in each frame, and process each traffic participant;

[0100] Step 2. Let the coordinates of the traffic participant be (x, y), where x is the horizontal coordinate and y is the vertical coordinate. First, perform anomaly detection on the horizontal coordinate x:

[0101] (1) Whether the x value is missing (None);

[0102] (2) Whether the value of x is infinite (-inf, +inf);

[0103] (3) Whether the x value is an anomaly (the difference between the values ​​of two adjacent identical elements is large, which is determined by comparing with the threshold).

[0104] If the x-coordinate is one of the above cases, it means that the x-coordinate is abnormal. For abnormal x-coordinates, do the following:

[0105] 2.1 Set the left and right pointers left_i and right_i, and initialize them to left_i = 1 and right_i = 1. The left pointer represents the historical frame data left_i frames away from the current frame, and the right pointer represents the future frame data right_i frames away from the current frame. Set the storage variables left_value and right_value to record the key values ​​in the frames corresponding to the left and right pointers, and initialize them to left_value = -1 and right_value = -1. Where left_value represents the valid x-coordinate value in the frame corresponding to the left pointer (-1 represents an invalid value), and right_value represents the valid x-coordinate value in the frame corresponding to the right pointer (-1 represents an invalid value).

[0106] 2.2 Search for the first non-abnormal value of the x-coordinate of the traffic participant from the current frame to the historical frame, including: extracting the x-coordinate value of the traffic participant in the historical frame corresponding to the left pointer, and according to the above three conditions for abnormality determination, if the x-coordinate value is non-abnormal, assign the x-coordinate value to left_value and stop searching. If the x-coordinate is determined to be abnormal, add 1 to the left pointer left_i to obtain a new historical frame, and determine whether the x-coordinate of the traffic participant in the historical frame is abnormal. If the x-coordinate is not abnormal, assign the x-coordinate to left_value, so as to assign the first normal value found from the current frame to the historical frame to left_value, and exit the search. If the x-coordinate is an abnormal value, add 1 to left_i and continue searching until the x-coordinate of the traffic participant is found to be a non-abnormal value or the first frame is found, then exit the search.

[0107] 2.3 Search for the first non-abnormal x-coordinate value of the traffic participant from the current frame to the future frame, including: extracting the x-coordinate value of the traffic participant in the future frame corresponding to the right pointer, and according to the three conditions for abnormality determination mentioned above, if the x-coordinate value is non-abnormal, assigning the x-coordinate value to right_value and stopping the search. If the x-coordinate is determined to be abnormal, incrementing the right pointer right_i by 1 to obtain a new future frame, and determining whether the x-coordinate of the traffic participant in the future frame is abnormal. If the x-coordinate is non-abnormal, assigning the x-coordinate to right_value and exiting the search. If the x-coordinate is an abnormal value, incrementing right_i by 1 and continuing the search until a non-abnormal x-coordinate value of the traffic participant is found or the last frame is found, then exiting the search.

[0108] 2.4 Correct the x coordinate by following the steps below:

[0109] 2.4.1 If left_value is not equal to -1 and right_value is not equal to -1, then use x_curr to correct the x coordinate of the traffic participant in the current frame. The calculation method is as follows:

[0110] x_curr=[left_i / (left_i+right_i)]*(right_value-left_value)+left_value

[0111] 2.4.2 If left_value is not equal to -1 and right_value is equal to -1, indicating that a non-outlier value exists in the current historical frame and no non-outlier value exists in the current future frame, then left_i is incremented by 1 to obtain a new historical frame. Following the three conditions for abnormality determination described in 2.2, the search continues from the new historical frame to the next historical frame, searching for the second non-outlier x-coordinate value of the traffic participant (i.e., the second normal value found in the historical frame). If such a value is found, the x-coordinate is recorded as left_value2, and the corresponding left pointer is recorded as left_i_2. If no second non-outlier value is found in the first frame, left_value2 is set to -1.

[0112] If left_value2 is not equal to -1, it means that the second non-outlier value has been found. Use x_curr to correct the x coordinate of the traffic participant in the current frame. The calculation method is as follows:

[0113] x_curr=(left_value-left_value2)*(left_i / left_i_2)+left_value

[0114] If left_value2 is equal to -1, then x_curr=left_value.

[0115] 2.4.3 If left_value is -1 and right_value is not -1, indicating that there is no non-outlier value in the current historical frame, but there is a non-outlier value in the current future frame, then right_i is incremented by 1 to obtain a new future frame. Following the three conditions for anomaly determination described in 2.2, the search continues from the new future frame to the next future frame to find the second non-outlier x-coordinate value of the traffic participant. If such a value is found, the x-coordinate is recorded as right_value2, and the corresponding right pointer is recorded as right_i_2 (the number of frames from the future frame to the current frame). If no second non-outlier value is found in the last frame, right_value2 is set to -1.

[0116] If right_value2 is not equal to -1, then use x_curr to correct the x coordinate of the traffic participant in the current frame. The calculation method is as follows:

[0117] x_curr=right_value-(right_value2-right_value)*(right_i / right_i_2)

[0118] If right_value2 is equal to -1, then x_curr=right_value.

[0119] 2.4.4 If both left_value and right_value are -1, indicating that no non-outlier values ​​exist in any historical or future frames, the entire lifecycle of the traffic participant is anomalous. In this case, we set x_curr to -1, indicating that the traffic participant is anomalous throughout its entire lifecycle. This method not only repairs the outlier x-coordinate value in the current frame, but also ensures smoothness across frames.

[0120] Step 3. Use the method in Step 2 to detect the vertical coordinate y of the traffic participant. If the inspection result is abnormal, you can use the same method as above to repair it.

[0121] In a feasible implementation, when the intelligent driving scene object is a static map element, the scene data includes at least a reference point coordinate sequence, and the step of correcting the scene data corresponding to the intelligent driving scene object using a preset data smoothing algorithm to obtain a corrected intelligent driving scene dataset includes:

[0122] Step S305: performing anomaly detection on the reference point coordinate sequence of the static map element according to a preset data anomaly condition to obtain abnormal reference point coordinates of the reference point coordinate sequence;

[0123] It should be noted that the reference point coordinate sequence refers to the coordinates of a series of reference points that are related in spatial position to constitute static map elements. In intelligent driving, reference points usually refer to fixed or dynamic landmarks or feature points used for positioning, navigation or path planning. For example, the reference point coordinate sequence of a lane can include all the reference point coordinates contained in a lane in the intelligent driving map (including lane line coordinates, lane direction coordinates, lane divider coordinates, road marking coordinates, etc.). Abnormal reference point coordinates refer to reference point coordinates that do not conform to the normal map element change pattern or range, which may be caused by sensor errors, data transmission errors, etc. Data anomaly conditions can be the absence, infinity or exceeding the normal range of coordinate values.

[0124] It can be understood that when the identified intelligent driving scene object is a static map element, the pre-set data anomaly conditions are used to detect the abnormal reference point coordinates in the reference point coordinate sequence, providing a target for subsequent data correction of the static map elements.

[0125] Step S306: performing abnormal traversal detection on the reference point coordinates in the starting point and / or ending direction corresponding to the current reference point in the reference point coordinate sequence according to the data abnormality condition;

[0126] It can be understood that the detected abnormal reference point coordinates are set as the current reference point, and the reference point coordinates of the current abnormal reference point in the starting point and / or end point direction of the space in the reference point coordinate sequence of the same frame are traversed to detect non-abnormal values ​​in the directions of both ends of the current abnormal reference point, so as to analyze the spatial relationship between the current abnormal reference point and other non-abnormal reference points and find reference points that can be used to correct abnormal values.

[0127] Step S307: If the presence of non-abnormal reference point coordinates is detected, weighted average calculation is performed on the abnormal reference point coordinates based on the non-abnormal reference point coordinates to obtain corrected reference point coordinates;

[0128] It can be understood that when the abnormal traversal detects that there are non-abnormal reference point coordinates in the reference point coordinates of the current abnormal reference point in the starting point and / or end point direction of the space in the reference point coordinate sequence of the same frame, this non-abnormal reference point coordinate is used as a reference value for correcting the abnormal value. By assigning different weights to the position coordinates of different reference points and performing weighted averaging calculation, the reference point coordinate value for correction is obtained, and this value is used to correct the abnormal data in the reference point coordinate sequence of the static map element, so as to reduce the impact of the abnormal value through the weighted averaging method, ensure the smoothness of the reference point coordinate data of the static map element, and improve the accuracy of the intelligent driving scene data set.

[0129] Step S308: Correct the reference point coordinate sequence of the static map element according to the corrected reference point coordinates to obtain a corrected intelligent driving scene dataset.

[0130] It can be understood that the reference point coordinate sequence of the static map elements in the intelligent driving scenario dataset is updated according to the corrected reference point coordinates obtained by weighted average calculation to obtain the intelligent driving scenario dataset after data correction, so as to gradually adjust the abnormal reference point coordinates of the lane lines in the intelligent driving scenario dataset until they are close to the position and state of the static map elements in the real environment as they change with space, thereby improving the quality and availability of the entire dataset.

[0131] In this implementation, anomaly detection is first used to identify outliers in the reference point coordinate sequence. Then, anomaly traversal detection is used to search for non-outlier values ​​in the direction of the starting point and / or end point as correction references. Next, a weighted average calculation method is used to correct the outlier reference point coordinates. Finally, the static map element reference point coordinate sequence in the intelligent driving scenario dataset is updated. Data repair, which utilizes the changing spatial relationships between static map element reference points, significantly improves the accuracy and reliability of reference point coordinates, reduces misjudgments caused by anomalies in the intelligent driving system, and thus enhances the safety and efficiency of the intelligent driving system.

[0132] For example, since the map under LCC (pure perception) is constructed in real time by perception, and the coordinates of the same lane in two adjacent frames are usually different, their coordinates are prone to anomalies. Therefore, under LCC perception mapping, it is necessary to detect and correct each frame of the map. The process of applying a smoothing algorithm to the abnormal points of the lane line reference point coordinates in the map data includes the following steps:

[0133] Step 1. Extract all lanes in the map corresponding to the current frame. Lanes are usually composed of a series of reference coordinate points, and are processed sequentially from the starting reference coordinate point of the lane.

[0134] Step 2. Let the reference coordinate point of the lane be (x, y), where x is the horizontal coordinate and y is the vertical coordinate. First, perform anomaly detection on the horizontal coordinate x:

[0135] (1) Whether the x value is missing (None);

[0136] (2) Whether the value of x is infinite (-inf, +inf);

[0137] (3) Whether the x value is abnormal (the value difference between two adjacent identical elements is large, which is determined by comparing with the threshold).

[0138] If x is one of the above cases, it means that the coordinate x is abnormal. For abnormal x, do the following:

[0139] 2.1 Set the left and right pointers left_i and right_i, and initialize them to left_i = 1 and right_i = 1. The left pointer indicates the reference coordinate point that is left_i points away from the current reference point in the direction of the starting point, and the right pointer indicates the reference coordinate point that is right_i points away from the current reference point in the direction of the end point. Set the values ​​left_value and right_value of the reference points corresponding to the left and right pointers, and initialize them to left_value = -1 and right_value = -1. Among them, left_value indicates the valid value of the x-coordinate of the coordinate point indicated by the left pointer (-1 indicates an invalid value), and right_value indicates the valid value of the x-coordinate of the coordinate point indicated by the right pointer (-1 indicates an invalid value).

[0140] 2.2 Search for the first non-outlier x-coordinate value from the current reference point toward the starting reference point, including: extracting the x-coordinate value of the reference point corresponding to the left pointer, and according to the three conditions for abnormality determination described above, if the x-coordinate value is non-outlier, assigning the x-coordinate value to left_value and stopping the search. If the x-coordinate value is determined to be abnormal, increment the left pointer left_i by 1 to obtain a new reference point, and determine whether the x-coordinate of the reference point is abnormal. If the x-coordinate is non-outlier, assign the x-coordinate value to left_value and exit the search. If it is an abnormal value, increment left_i by 1 and continue searching until the x-coordinate of the reference point is found to be non-outlier or the starting coordinate point of the lane is found, then exit the search.

[0141] 2.3 Search for the first non-abnormal x-coordinate value from the current reference point to the end reference point, including: extracting the x-coordinate value of the reference point corresponding to the right pointer, and according to the three conditions for abnormality determination mentioned above, if the x-coordinate value is non-abnormal, assigning the x-coordinate value to right_value and stopping the search. If the x-coordinate value is determined to be abnormal, incrementing the right pointer right_i by 1 to obtain a new reference point, and determining whether the x-coordinate of the reference point is abnormal. If the x-coordinate is non-abnormal, assigning the x-coordinate value to right_value and exiting the search. If it is an abnormal value, incrementing right_i by 1 and continuing the search until the x-coordinate of the reference point is found to be non-abnormal or the coordinates of the end reference point of the lane are found, then exiting the search.

[0142] 2.4. Correct the x-coordinate by the following conditions:

[0143] 2.4.1 If left_value is not equal to -1 and right_value is not equal to -1, it means that there are non-abnormal reference point coordinates in the direction of the current reference point to the starting reference point and the end reference point. Then use x_curr to correct the x coordinate of the current reference point. The calculation method is as follows:

[0144] x_curr=[left_i / (left_i+right_i)]*(right_value-left_value)+left_value

[0145] 2.4.2 If left_value is not equal to -1 and right_value is equal to -1, indicating that there is a non-outlier reference point coordinate in the direction from the current reference point to the starting reference point, and no non-outlier reference point coordinate in the direction from the current reference point to the ending reference point, then add 1 to left_i to obtain a new reference point. Based on the three conditions for abnormality determination described in 2.2, continue searching from this new reference point toward the starting reference point for a second non-outlier x-coordinate value. If such a value is found, record that x-coordinate as left_value2 and the corresponding left pointer as left_i_2 (the number of distances from the current reference point to the starting point). If no second non-outlier value is found at the starting point of the lane, set left_value2 to -1.

[0146] If left_value2 is not equal to -1, it means that there is a second non-abnormal reference point coordinate in the direction of the current reference point to the starting reference point. Then use x_curr to correct the x coordinate of the current reference point. The calculation method is as follows:

[0147] x_curr=(left_value-left_value2)*(left_i / left_i_2)+left_value

[0148] If left_value2 is equal to -1, then x_curr=left_value.

[0149] 2.4.3 If left_value is -1 and right_value is not -1, indicating that there are no non-outlier reference point coordinates from the current reference point toward the starting reference point, but there are non-outlier reference point coordinates from the current reference point toward the end reference point, then right_i is incremented by 1 to obtain a new reference point. Following the three outlier determination criteria described in 2.2, the search continues from this new reference point toward the lane's end reference point, searching for a second non-outlier x-coordinate value. If such a value is found, the x-coordinate is recorded as right_value2, and the corresponding right pointer is recorded as right_i_2 (the number of distances from the current reference point to the current reference point). If no second non-outlier value is found at the end of the lane, right_value2 is set to -1.

[0150] If right_value2 is not equal to -1, it means that there is a second non-abnormal reference point coordinate in the direction from the current reference point to the end reference point. Then use x_curr to correct the x coordinate of the current reference point. The calculation method is as follows:

[0151] x_curr=right_value-(right_value2-right_value)*(right_i / right_i_2)

[0152] If right_value2 is equal to -1, then x_curr=right_value.

[0153] 2.4.4 If both left_value and right_value are -1, this indicates that there are no non-anomalous reference point coordinates from the current reference point toward either the starting or ending reference point. Therefore, the entire lane's reference points are considered abnormal. In this case, set x_curr to -1 to indicate that the lane is abnormal. This method can be used to correct the abnormal x-coordinate value of the lane's current reference point, ensuring smoothness.

[0154] Step 3. Similarly, use the method in Step 2 to detect the vertical coordinate y of the reference point. If the inspection result is abnormal, you can use the same method as above to repair it.

[0155] This embodiment uses a data smoothing algorithm combined with time series analysis and spatial series analysis to perform data correction, effectively correcting outliers in vehicle position coordinate data and lane reference point coordinate data. For vehicle position coordinates, their temporal continuity is leveraged to identify and correct outliers, ensuring a smooth and continuous vehicle trajectory. For lane reference point coordinates, their spatial geometric relationships are leveraged to identify and correct outliers, ensuring the accuracy and consistency of lane data. This improves the intelligent driving system's ability to understand the environment, reduces misjudgments due to data anomalies, and thus enhances the safety and reliability of the intelligent driving system.

[0156] This application also provides a data correction device, please refer to Figure 3 , the data correction device includes:

[0157] An acquisition module 10 is configured to acquire an intelligent driving scene data set, where the intelligent driving scene data set includes a plurality of frames of scene data;

[0158] an identification module 20 for performing scene object recognition based on the intelligent driving scene dataset to obtain intelligent driving scene objects, wherein the intelligent driving scene objects include traffic participants and / or static map elements;

[0159] The correction module 30 is used to correct the scene data corresponding to the intelligent driving scene object through a preset data smoothing algorithm to obtain a corrected intelligent driving scene data set.

[0160] Optionally, the correction module 30 is further configured to:

[0161] Determining whether the scene data meets a preset data abnormality condition, where the data abnormality condition includes one or more of a missing coordinate value, an infinite coordinate value, and an abnormal coordinate value range;

[0162] If the scene data meets the data abnormality condition, a step of correcting the scene data is performed.

[0163] Optionally, when the intelligent driving scene object is a traffic participant, the scene data at least includes position coordinate data, and the correction module 30 is further configured to:

[0164] According to a preset data abnormality condition, an abnormality detection is performed on the position coordinate data of the traffic participant to obtain abnormal position coordinates of the position coordinate data;

[0165] According to the data abnormality condition, abnormal traversal detection is performed on the position coordinates of the historical frames and / or future frames corresponding to the current frame in the position coordinate data;

[0166] If the presence of non-abnormal position coordinates is detected, weighted average calculation is performed on the abnormal position coordinates based on the non-abnormal position coordinates to obtain corrected position coordinates;

[0167] According to the corrected position coordinates, the position coordinate data of the traffic participant is corrected to obtain a corrected intelligent driving scene data set.

[0168] Optionally, the correction module 30 is further configured to:

[0169] If abnormal position coordinates are detected, abnormal traversal detection is stopped and the position coordinate data is marked as abnormal data.

[0170] Optionally, when the intelligent driving scene object is a static map element, the scene data at least includes a reference point coordinate sequence, and the correction module 30 is further configured to:

[0171] performing anomaly detection on the reference point coordinate sequence of the static map element according to a preset data anomaly condition to obtain the abnormal reference point coordinates of the reference point coordinate sequence;

[0172] According to the data abnormality condition, abnormal traversal detection is performed on the reference point coordinates in the starting point and / or end point direction corresponding to the current reference point in the reference point coordinate sequence;

[0173] If a non-abnormal reference point coordinate is detected, a weighted average calculation is performed on the abnormal reference point coordinate based on the non-abnormal reference point coordinate to obtain a corrected reference point coordinate;

[0174] According to the corrected reference point coordinates, the reference point coordinate sequence of the static map element is corrected to obtain a corrected intelligent driving scene dataset.

[0175] Optionally, the identification module 20 is further configured to:

[0176] Identify the scene object corresponding to each frame of scene data in the intelligent driving scene dataset;

[0177] According to the time sequence of each frame of scene data, the scene objects corresponding to each frame of scene data are associated to obtain the intelligent driving scene object.

[0178] The data correction device provided in this application, employing the data correction method described in the aforementioned embodiments, can address the technical problem of effectively correcting abnormal data in intelligent driving scenario datasets. Compared to the prior art, the data correction device provided in this application achieves the same beneficial effects as the data correction method described in the aforementioned embodiments. Other technical features of the data correction device are the same as those disclosed in the aforementioned embodiments and are not further elaborated here.

[0179] The present application provides a data correction device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the data correction method in the above-mentioned first embodiment.

[0180] Reference below Figure 4 , which shows a schematic diagram of the structure of a data correction device suitable for implementing an embodiment of the present application. The data correction device in the embodiment of the present application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 4 The data correction device shown is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0181] like Figure 4As shown, the data correction device may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory 1002 or a program loaded from a storage device 1003 into a random access memory 1004. Random access memory 1004 also stores various programs and data required for the operation of the data correction device. Processing device 1001, read-only memory 1002, and random access memory 1004 are interconnected via a bus 1005. An input / output interface 1006 is also connected to the bus. Typically, the following systems can be connected to the input / output interface 1006: an input device 1007 including, for example, a touch screen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, hard disk, etc.; and a communication device 1009. Communication device 1009 can allow the data modification device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a data modification device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have instead.

[0182] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device 1003, or installed from a read-only memory 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are performed.

[0183] The data correction device provided in this application, employing the data correction method described in the aforementioned embodiment, can address the technical problem of effectively correcting abnormal data in intelligent driving scenario datasets. Compared to the prior art, the data correction device provided in this application achieves the same beneficial effects as the data correction method described in the aforementioned embodiment. Other technical features of this data correction device are the same as those disclosed in the aforementioned embodiment and are not further elaborated here.

[0184] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0185] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0186] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer program) stored thereon, wherein the computer-readable program instructions are used to execute the data correction method in the above embodiment.

[0187] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared or semiconductor systems, systems or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0188] The computer-readable storage medium may be included in the data correction device, or may exist independently without being assembled into the data correction device.

[0189] The above-mentioned computer-readable storage medium carries one or more programs. When the above-mentioned one or more programs are executed by the data correction device, the data correction device: obtains an intelligent driving scene data set, and the intelligent driving scene data set includes several frames of scene data; performs scene object recognition based on the intelligent driving scene data set to obtain an intelligent driving scene object, and the intelligent driving scene object includes traffic participants and / or static map elements; corrects the scene data corresponding to the intelligent driving scene object through a preset data smoothing algorithm to obtain a corrected intelligent driving scene data set.

[0190] Computer program code for performing the operations of the present application may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on the remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0191] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, program segment or part of code, and the module, program segment or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, as well as the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or can be implemented using a combination of dedicated hardware and computer instructions.

[0192] The modules described in the embodiments of the present application may be implemented in software or hardware, wherein the name of a module does not necessarily limit the unit itself.

[0193] The computer-readable storage medium provided in this application stores computer-readable program instructions (i.e., a computer program) for executing the aforementioned data correction method. This computer-readable storage medium can address the technical problem of effectively correcting abnormal data in intelligent driving scenario datasets. Compared to the prior art, the beneficial effects of the computer-readable storage medium provided in this application are similar to those of the data correction method provided in the aforementioned embodiments and are not further elaborated here.

[0194] The present application also provides a computer program product, comprising a computer program, which implements the steps of the above-mentioned data correction method when executed by a processor.

[0195] The computer program product provided in this application can solve the technical problem of effectively correcting abnormal data in intelligent driving scenario datasets. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the data correction method provided in the above embodiment, and will not be elaborated here.

[0196] The above description is only part of the embodiments of the present application and does not limit the patent scope of the present application. All equivalent structural transformations made by using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect application in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A data correction method, characterized in that: The method comprises: Acquire an intelligent driving scene dataset, where the intelligent driving scene dataset includes several frames of scene data; Performing scene object recognition according to the intelligent driving scene dataset to obtain intelligent driving scene objects, wherein the intelligent driving scene objects include traffic participants and / or static map elements; The scene data corresponding to the intelligent driving scene object is corrected by a preset data smoothing algorithm to obtain a corrected intelligent driving scene data set.

2. The method according to claim 1, wherein: Before the step of correcting the scene data corresponding to the intelligent driving scene object by a preset data smoothing algorithm to obtain a corrected intelligent driving scene data set, the method further includes: Determining whether the scene data meets a preset data abnormality condition, where the data abnormality condition includes one or more of a missing coordinate value, an infinite coordinate value, and an abnormal coordinate value range; If the scene data meets the data abnormality condition, a step of correcting the scene data is performed.

3. The method according to claim 1, wherein: When the intelligent driving scene object is a traffic participant, the scene data at least includes position coordinate data, The step of correcting the scene data corresponding to the intelligent driving scene object by a preset data smoothing algorithm to obtain a corrected intelligent driving scene data set includes: According to a preset data abnormality condition, an abnormality detection is performed on the position coordinate data of the traffic participant to obtain abnormal position coordinates of the position coordinate data; According to the data abnormality condition, abnormal traversal detection is performed on the position coordinates of the historical frames and / or future frames corresponding to the current frame in the position coordinate data; If the presence of non-abnormal position coordinates is detected, weighted average calculation is performed on the abnormal position coordinates based on the non-abnormal position coordinates to obtain corrected position coordinates; According to the corrected position coordinates, the position coordinate data of the traffic participant is corrected to obtain a corrected intelligent driving scene data set.

4. The method according to claim 3, wherein: After the step of performing abnormal traversal detection on the position coordinates of the historical frames and / or future frames corresponding to the current frame in the position coordinate data according to the data abnormality condition, the method further includes: If abnormal position coordinates are detected, abnormal traversal detection is stopped and the position coordinate data is marked as abnormal data.

5. The method according to claim 1, wherein: When the intelligent driving scene object is a static map element, the scene data at least includes a reference point coordinate sequence, The step of correcting the scene data corresponding to the intelligent driving scene object by a preset data smoothing algorithm to obtain a corrected intelligent driving scene data set includes: performing anomaly detection on the reference point coordinate sequence of the static map element according to a preset data anomaly condition to obtain the abnormal reference point coordinates of the reference point coordinate sequence; According to the data abnormality condition, abnormal traversal detection is performed on the reference point coordinates in the starting point and / or end point direction corresponding to the current reference point in the reference point coordinate sequence; If a non-abnormal reference point coordinate is detected, a weighted average calculation is performed on the abnormal reference point coordinate based on the non-abnormal reference point coordinate to obtain a corrected reference point coordinate; According to the corrected reference point coordinates, the reference point coordinate sequence of the static map element is corrected to obtain a corrected intelligent driving scene dataset.

6. The method according to claim 1, wherein: The step of performing scene object recognition according to the intelligent driving scene dataset to obtain the intelligent driving scene object includes: Identify the scene object corresponding to each frame of scene data in the intelligent driving scene dataset; According to the time sequence of each frame of scene data, the scene objects corresponding to each frame of scene data are associated to obtain the intelligent driving scene object.

7. A data correction device, characterized in that: The device comprises: An acquisition module is used to acquire an intelligent driving scene data set, where the intelligent driving scene data set includes a plurality of frames of scene data; an identification module, configured to identify scene objects based on the intelligent driving scene dataset to obtain intelligent driving scene objects, wherein the intelligent driving scene objects include traffic participants and / or static map elements; The correction module is used to correct the scene data corresponding to the intelligent driving scene object through a preset data smoothing algorithm to obtain a corrected intelligent driving scene data set.

8. A data correction device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the data correction method according to any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the data correction method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the data correction method according to any one of claims 1 to 6 are implemented.

Images