Data processing method, vehicle communication device, and storage medium

Abstract

The present invention discloses a data processing method, a vehicle communication device, and a storage medium. The method comprises: obtaining raw data, where the raw data is generated from a preset data set; invoking a perception fusion algorithm to process the raw data to obtain a fusion result; and generating evaluation metrics corresponding to the perception fusion algorithm based on the fusion result, where the evaluation metrics are used to evaluate the performance of the perception fusion algorithm. This invention solves the technical problems of low test accuracy and the inability to evaluate test results using metrics due to insufficient data volume during testing.

Description

Technical Field

[0001] The present invention relates to the field of software testing, and in particular to a data processing method, a vehicle communication device and a storage medium. Background Art

[0002] In the autonomous driving system, it can be roughly divided into five functional modules: perception, positioning, prediction, planning and control. Among them, the perception module is the core of autonomous driving and the prerequisite for other modules. It needs to use sensors to perceive the surrounding environment and then fuse the data collected by different sensors.

[0003] In the prior art, the evaluation of perception fusion algorithms usually adopts unit testing and real vehicle testing. Among them, unit testing is generally carried out after the perception fusion algorithm is written, and it simply verifies whether the code is functional by fabricating data; the real vehicle test controls the real vehicle verification through code, and evaluates based on the actual effect. It is easy to notice that in the above-mentioned unit test, since data needs to be artificially fabricated, the test scale is usually very small, and in the real vehicle test, there is also the problem of not being able to perform an indicator evaluation of the actual results. In summary, there are problems in vehicle testing such as low test data volume, low test result accuracy, and inability to perform an indicator evaluation of the real vehicle test results.

[0004] To address the above-mentioned problems, no effective solutions have been proposed so far. Summary of the Invention

[0005] Embodiments of the present invention provide a data processing method, a vehicle communication device, and a storage medium to at least solve the technical problems of low test accuracy and inability to perform indexed evaluation of test results due to too small a data volume during the test process.

[0006] According to a first aspect of an embodiment of the present invention, a data processing method is provided, including: obtaining original data, wherein the original data is generated by a preset data set; calling a perceptual fusion algorithm to process the original data to obtain a fusion result; and generating an evaluation index corresponding to the perceptual fusion algorithm based on the fusion result, wherein the evaluation index is used to evaluate the performance of the perceptual fusion algorithm.

[0007] Optionally, obtaining raw data includes: obtaining a preset data set, wherein the preset data set includes: detection results detected by different sensors during the vehicle's driving in different scenarios; processing the preset data set to generate raw data, wherein the raw data includes at least: yaw angle, speed, obstacle polygon and obstacle detection box.

[0008] Optionally, processing the preset data set to generate original data includes: processing the detection results in different scenarios to obtain the original data in different scenarios.

[0009] Optionally, a perception fusion algorithm is called to process the original data in different scenarios to obtain a fusion result.

[0010] Optionally, calling the perception fusion algorithm to process the original data to obtain the fusion result includes: parsing the original data to obtain data fields; and processing the data fields using the perception fusion algorithm to obtain the fusion result.

[0011] Optionally, processing the data field using the perception fusion algorithm to obtain a fusion result includes: initializing the perception fusion algorithm; traversing the data field in sequence and calling the perception fusion algorithm to obtain a fusion result.

[0012] Optionally, the method further includes: storing the fusion result in a preset format.

[0013] Optionally, based on the fusion result, generating an evaluation index corresponding to the perception fusion algorithm includes: storing the fusion result to a preset location of the evaluation code; and running the evaluation code to generate the evaluation index.

[0014] According to a second aspect of an embodiment of the present invention, a vehicle communication device is provided, comprising: an acquisition module for acquiring target code data in an autonomous driving system, wherein the target code data is used for data communication between other modules; a construction module for constructing a target library based on the target code data; and a communication module for invoking the target library for data communication using an intermediate communication layer.

[0015] Optionally, the device is also used to obtain a preset data set, wherein the preset data set includes: detection results detected by different sensors during the vehicle's driving in different scenarios; the preset data set is processed to generate raw data, wherein the raw data includes at least: yaw angle, speed, obstacle polygon and obstacle detection box.

[0016] Optionally, the device is further configured to process a preset data set, and generating raw data includes: processing detection results in different scenarios to obtain raw data in different scenarios.

[0017] Optionally, the device is also used to call a perception fusion algorithm to process raw data in different scenarios to obtain a fusion result.

[0018] Optionally, the device is also used to call the perception fusion algorithm to process the original data, and obtain the fusion result including: parsing the original data to obtain data fields; and using the perception fusion algorithm to process the data fields to obtain the fusion result.

[0019] Optionally, the device is also used to process the data field using the perception fusion algorithm to obtain the fusion result, including: initializing the perception fusion algorithm; traversing the data field in sequence and calling the perception fusion algorithm to obtain the fusion result.

[0020] Optionally, the device is also used to generate evaluation indicators corresponding to the perception fusion algorithm based on the fusion results, including: storing the fusion results to a preset location of the evaluation code; running the evaluation code to generate the evaluation indicators.

[0021] According to a third aspect of an embodiment of the present invention, a computer-readable storage medium is provided. The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute any data processing method.

[0022] According to a fourth aspect of an embodiment of the present invention, a processor is provided, characterized in that the processor is used to run a program, wherein any one of the data processing methods is executed when the program is run.

[0023] In an embodiment of the present invention, for testing and evaluating a fusion perception algorithm, a large amount of raw data is obtained from a preset data set for perception fusion, a fusion result is obtained, and then corresponding evaluation indicators are generated based on the fusion result, thereby achieving the purpose of testing and evaluating the performance of the perception fusion algorithm. It is easy to notice that in this application, sufficient raw data is generated from a preset data set for perception fusion, and evaluation indicators for evaluating the performance of the perception fusion algorithm are generated. There is no need to artificially fabricate data or conduct actual vehicle verification, thereby achieving the technical effect of improving the accuracy of the perception fusion algorithm test results and evaluating the test results in an indicator-based manner, solving the technical problem that insufficient test data leads to low test result accuracy and the inability to perform indicator-based evaluation of the test results. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0025] Figure 1 is a flow chart of a data processing method according to an embodiment of the present invention;

[0026] Figure 2 is a schematic diagram of some optional basic indicators according to an embodiment of the present invention;

[0027] Figure 3 is a schematic diagram of an optional perception fusion algorithm evaluation process according to an embodiment of the present invention;

[0028] Figure 41 is a schematic diagram of a detection result format of a Radar sensor in a Nusenes dataset according to an optional embodiment of the present invention;

[0029] Figure 5 is a schematic diagram of an optional fusion offline tool flow according to an embodiment of the present invention;

[0030] Figure 6 FIG. 4 is a schematic structural diagram of a vehicle communication device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0031] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0032] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0033] According to an embodiment of the present invention, a method embodiment of a data processing method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0034] Figure 1 FIG. 1 is a flow chart of a data processing method according to an embodiment of the present invention. Figure 1 As shown, the method includes the following steps:

[0035] Step S101: obtaining original data, wherein the original data is generated from a preset data set.

[0036] Specifically, the preset dataset may be a large dataset including different driving scenarios, for example, an open source dataset based on Nuscenes on the Internet, including detection results of Lidar, Camera, and Radar sensors.

[0037] In an optional embodiment, required data may be obtained from a preset data set, and then the obtained data may be formatted according to actual needs to obtain original data.

[0038] Step S102: calling the perception fusion algorithm to process the original data to obtain a fusion result.

[0039] Specifically, the above-mentioned perception fusion algorithm can be any algorithm used for multi-sensor fusion in the perception module in autonomous driving, such as the weighted average method and Kalman filter method in random methods; neural networks, genetic algorithms, fuzzy theory and their improvements or combinations in artificial intelligence methods.

[0040] In an optional embodiment, the original data may be processed using an existing perception fusion algorithm to obtain a fusion result, which is not specifically limited in the present invention.

[0041] Step S103: generating an evaluation index corresponding to the perception fusion algorithm based on the fusion result, wherein the evaluation index is used to evaluate the performance of the perception fusion algorithm.

[0042] Specifically, the above-mentioned evaluation indicators include basic indicators and fusion indicators, among which the basic indicators may include missed detection, false alarm, false association, etc., and the fusion indicators may include MOTA (Multiple Object Tracking Accuracy), MOTP (Multiple Object Tracking Precision), AMOTA (Average Multiple Object Tracking Accuracy), etc., but are not limited to these.

[0043] In an optional embodiment, as Figure 2 As shown, basic indicators include but are not limited to: missed detection (FN), which indicates that there is no true value that matches any hypothesis; false alarm (FP), which indicates that there is no hypothesis that matches any true value; wrong association (Switch), which indicates that the matching relationship between the true value and the hypothesis of the previous and next frames is inconsistent; correct association (Match), which indicates that the matching relationship between the true value and the hypothesis of the previous and next frames is consistent; recall (Recall), which indicates the ratio of the true value that is successfully matched to the total number of true values, and is calculated as follows:

[0044] Recall=TP / (TP+FN)=(Match+Switch) / GT,

[0045] Among them, GT represents the total number of true targets.

[0046] In an optional embodiment, the comprehensive indicators include but are not limited to MOTA, MOTP, AMOTA, and AMOTP, wherein the calculation formula of MOTA is as follows:

[0047]

[0048] Among them, GT t Indicates the total number of true value targets in the t-th frame, FN t Indicates the number of missed detections in the t-th frame, FP t represents the number of false alarms in the t-th frame, and IDSt represents the number of incorrect associations in the t-th frame (ie, the number of IDS).

[0049] The calculation formula for MOT is as follows:

[0050]

[0051] Among them, TP t Indicates the number of successful matches in the t-th frame, d i t Indicates the distance between the true value i and the hypothesis value that are successfully matched in the t-th frame, i = 1, ..., TP t .

[0052] The calculation formula for AMOTA is as follows:

[0053]

[0054]

[0055] Among them, GT is the comprehensive number of true value targets in each frame up to the t-th frame; FN r is the sum of the number of missed detections in the previous t frames when Recall is r; FP r is the sum of the number of false alarms in the previous t frames when Recall is r; IDS r is the sum of the number of error associations in the previous t frames when Recall is r, that is, IDS r The sum of the numbers, n represents the number of Recall divisions.

[0056] The calculation formula for AMOTP is as follows:

[0057]

[0058]

[0059] Among them, TPt represents the number of successful matches in the tth frame when Recall is r, d i t Indicates the distance between the true value i and the hypothesis value that are successfully matched in the t-th frame, i = 1, ..., TP t .

[0060] The above embodiments can be used to test and evaluate fusion perception algorithms. Existing evaluation methods can be used to perform performance evaluation based on evaluation indicators. For example, the generated evaluation indicators can be compared with pre-set thresholds, and the performance of the perception fusion algorithm can be determined based on the comparison results.

[0061] Through the above-mentioned embodiments of the present invention, the purpose of testing and evaluating the performance of the perception fusion algorithm is achieved by obtaining a large amount of raw data from a preset data set for perception fusion, obtaining a fusion result, and then generating corresponding evaluation indicators based on the fusion result. It is easy to note that in this application, sufficient raw data is generated from a preset data set for perception fusion, and evaluation indicators for evaluating the performance of the perception fusion algorithm are generated. There is no need for artificially fabricated data or actual vehicle verification, thereby achieving the technical effect of improving the accuracy of the perception fusion algorithm test results and evaluating the test results in an indicator-based manner, and solving the technical problem that the test results are low in accuracy due to insufficient test data and the test results cannot be evaluated in an indicator-based manner.

[0062] Optionally, obtaining raw data includes: obtaining a preset data set, wherein the preset data set includes: detection results detected by different sensors during the vehicle's driving in different scenarios; processing the preset data set to generate raw data, wherein the raw data includes at least: yaw angle, speed, obstacle polygon and obstacle detection box.

[0063] Specifically, the above-mentioned sensors may include cameras, optical radars (Lidar for short), millimeter-wave radars (Radar) and other sensors. The above-mentioned preset data sets may be real-time monitoring data uploaded by different vehicles already existing in the network. Different vehicles may travel in different scenarios, so that multiple sensors installed on the vehicle can collect driving data.

[0064] In an optional embodiment, the preset dataset may be the public dataset Nuscense, which includes driving data collected in cities such as Boston and Singapore under multiple different scenarios. The aforementioned processing of the preset dataset to generate raw data includes: first obtaining the required preliminary dataset based on actual needs; formatting the data in the preset dataset to obtain a formatted dataset, which may be a proto file; and then obtaining raw data based on the formatted dataset based on actual needs. It should be noted that some raw data cannot be obtained directly from the preset dataset but can be obtained indirectly. For example, the yaw angle can be obtained from position, length, width, height, and orientation.

[0065] Optionally, processing the preset data set to generate original data includes: processing the detection results in different scenarios to obtain the original data in different scenarios.

[0066] Specifically, a preset dataset contains data collected in multiple different scenarios. The initial dataset required for the subsequent perception fusion algorithm is obtained from this preset dataset. This initial dataset is then formatted to obtain a formatted dataset, and raw data is generated based on this formatted dataset. Some of the raw data may not be directly available from the preset dataset but can be obtained indirectly through calculation. It should be noted that in actual applications, the data required for the perception fusion algorithm can also be obtained based on the preset dataset before formatting.

[0067] In an optional embodiment, after obtaining the raw data, the scene name is read from the folder of the raw data. For example, the detection results of the Lidar sensor in the preset data set are stored in json format. The detection results of the Lidar sensor are converted from json format to proto format for storage, and each scene has a corresponding folder, which can be named with the scene label.

[0068] Optionally, a perception fusion algorithm is called to process the original data in different scenarios to obtain a fusion result.

[0069] In an alternative embodiment, if Figure 3 As shown in the figure, the detection result of the optical radar sensor represents the Nuscenes Lidar tracking result obtained from the Nusenes dataset; the detection result of the millimeter wave radar sensor represents the Nuscenes radar tracking result obtained from the Nusenes dataset; and the detection result of the camera sensor represents the Nuscenes camera tracking result obtained from the Nusenes dataset. Figure 4The following figure shows the detection result format of the Radar sensor in the Nusenes dataset. These detection results are stored in the JSON (JavaScript Object Notation) format. These JSON-formatted detection results need to be formatted and saved as a proto file. It should be noted that the detection results of different sensors vary. For example, the detection results of the Radar sensor are in the pcd format, which also needs to be formatted and saved as a proto file. FusionOffline Tools are then used to perform perception fusion to obtain the fusion result. This fusion result is then placed in the specified location in the evaluation code in the Nuscenes-devkit, and the evaluation code is run. Nuscenes-devkit is the development kit for Nuscenes.

[0070] Optionally, calling the perception fusion algorithm to process the original data to obtain the fusion result includes: parsing the original data to obtain data fields; and processing the data fields using the perception fusion algorithm to obtain the fusion result.

[0071] Specifically, after obtaining the raw data, the scene name is read from the folder of the raw data, and then the proto files of multiple sensors are read in according to the scene name and in a preset order. The proto files are parsed based on the proto files to obtain the data field, which can be a sensor data frame. After initializing the fusion algorithm plug-in, the data field is processed using the perception fusion algorithm to obtain the fusion result.

[0072] Optionally, processing the data field using the perception fusion algorithm to obtain a fusion result includes: initializing the perception fusion algorithm; traversing the data field in sequence and calling the perception fusion algorithm to obtain a fusion result.

[0073] Specifically, the initialization of the perception fusion algorithm can be to parse the sensor data and assign values ​​to the objects in the perception fusion algorithm, sequentially traverse the sensor data fields and call the perception fusion algorithm, and after obtaining the fusion result, save the fusion result in a preset format.

[0074] In an optional embodiment, if Figure 5As shown in the figure, after obtaining the raw data, a folder list is retrieved from the raw data, and the scene name is read from this folder list. The proto files for the three sensors are then read in the specified order based on the scene name. After reading the proto files, they are converted into sensor data frames. After initializing the fusion algorithm plugin, the sensor data frames are sequentially traversed and the algorithm function is called. Finally, the algorithm results are saved in the specified JSON format. It should be noted that the three sensors mentioned above are radar, lidar, and camera, and the algorithm function is a fusion algorithm. The results of the perception fusion algorithm are stored in JSON format in a protobuf file.

[0075] Optionally, the method further includes: storing the fusion result in a preset format.

[0076] Specifically, the preset format may be a protobuf format.

[0077] Optionally, based on the fusion result, generating an evaluation index corresponding to the perception fusion algorithm includes: storing the fusion result to a preset location of the evaluation code; and running the evaluation code to generate the evaluation index.

[0078] Specifically, the evaluation code may be an evaluation code for generating evaluation indicators, and there is a preset location in the evaluation code for storing the fusion result. After the fusion result is placed in the preset location of the evaluation code, the evaluation code is run to obtain the evaluation indicator corresponding to the fusion result.

[0079] Figure 6 FIG. 1 is a schematic structural diagram of a vehicle communication device according to an embodiment of the present invention. Figure 6 As shown, the device includes:

[0080] The acquisition module 61 is used to acquire target code data in the automatic driving system, wherein the target code data is used for data communication between other modules.

[0081] The construction module 62 is used to construct a target library based on the target code data.

[0082] The communication module 63 is used to use the intermediate communication layer to call the target library for data communication.

[0083] In the above-mentioned embodiment, the purpose of testing and evaluating the performance of the perception fusion algorithm is achieved by obtaining a large amount of raw data from a preset data set for perception fusion, obtaining a fusion result, and then generating corresponding evaluation indicators based on the fusion result. It is easy to note that in this application, sufficient raw data is generated from a preset data set for perception fusion, and evaluation indicators for evaluating the performance of the perception fusion algorithm are generated. There is no need for artificial data fabrication or actual vehicle verification, thereby achieving the technical effect of improving the accuracy of the perception fusion algorithm test results and evaluating the test results in an indicator-based manner, and solving the technical problem that insufficient test data leads to low test result accuracy and the inability to perform indicator-based evaluation of the test results.

[0084] Optionally, the device is also used to obtain raw data including: obtaining a preset data set, wherein the preset data set includes: detection results detected by different sensors during the vehicle's driving in different scenarios; processing the preset data set to generate raw data, wherein the raw data includes at least: yaw angle, speed, obstacle polygon and obstacle detection box.

[0085] Specifically, the above-mentioned sensors may include cameras, optical radars (Lidar for short), millimeter-wave radars (Radar) and other sensors. The above-mentioned preset data sets may be real-time monitoring data uploaded by different vehicles already existing in the network. Different vehicles may travel in different scenarios, so that multiple sensors installed on the vehicle can collect driving data.

[0086] In an optional embodiment, the preset dataset may be the public dataset Nuscense, which includes driving data collected in cities such as Boston and Singapore under multiple different scenarios. The aforementioned processing of the preset dataset to generate raw data includes: first obtaining the required preliminary dataset based on actual needs; formatting the data in the preset dataset to obtain a formatted dataset, which may be a proto file; and then obtaining raw data based on the formatted dataset based on actual needs. It should be noted that some raw data cannot be obtained directly from the preset dataset but can be obtained indirectly. For example, the yaw angle can be obtained from position, length, width, height, and orientation.

[0087] Optionally, the device is further configured to process a preset data set, and generating raw data includes: processing detection results in different scenarios to obtain raw data in different scenarios.

[0088] Specifically, a preset dataset contains data collected in multiple different scenarios. The initial dataset required for the subsequent perception fusion algorithm is obtained from this preset dataset. This initial dataset is then formatted to obtain a formatted dataset, and raw data is generated based on this formatted dataset. Some of the raw data may not be directly available from the preset dataset but can be obtained indirectly through calculation. It should be noted that in actual applications, the data required for the perception fusion algorithm can also be obtained based on the preset dataset before formatting.

[0089] In an optional embodiment, after obtaining the raw data, the scene name is read from the folder of the raw data. For example, the detection results of the Lidar sensor in the preset data set are stored in json format. The detection results of the Lidar sensor are converted from json format to proto format for storage, and each scene has a corresponding folder, which can be named with the scene label.

[0090] Optionally, the device is also used to process raw data in different scenarios using a perception fusion algorithm to obtain a fusion result.

[0091] Optionally, calling the perception fusion algorithm to process the original data to obtain the fusion result includes: parsing the original data to obtain data fields; and processing the data fields using the perception fusion algorithm to obtain the fusion result.

[0092] Specifically, after obtaining the raw data, the scene name is read from the folder of the raw data, and then the proto files of multiple sensors are read in according to the scene name and in a preset order. The proto files are parsed based on the proto files to obtain the data field, which can be a sensor data frame. After initializing the fusion algorithm plug-in, the data field is processed using the perception fusion algorithm to obtain the fusion result.

[0093] Optionally, processing the data field using the perception fusion algorithm to obtain a fusion result includes: initializing the perception fusion algorithm; traversing the data field in sequence and calling the perception fusion algorithm to obtain a fusion result.

[0094] Specifically, the initialization of the perception fusion algorithm can be to parse the sensor data and assign values ​​to the objects in the perception fusion algorithm, sequentially traverse the sensor data fields and call the perception fusion algorithm, and after obtaining the fusion result, save the fusion result in a preset format.

[0095] Optionally, based on the fusion result, generating an evaluation index corresponding to the perception fusion algorithm includes: storing the fusion result to a preset location of the evaluation code; and running the evaluation code to generate the evaluation index.

[0096] Specifically, the evaluation code may be an evaluation code for generating evaluation indicators, and there is a preset location in the evaluation code for storing the fusion result. After the fusion result is placed in the preset location of the evaluation code, the evaluation code is run to obtain the evaluation indicator corresponding to the fusion result.

[0097] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is further provided. The computer-readable storage medium includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the data processing methods in the above embodiments.

[0098] According to another aspect of an embodiment of the present invention, a processor is further provided, and the processor is used to run a program, wherein when the program is run, any one of the data processing methods in the above embodiments is executed.

[0099] The serial numbers of the above embodiments of the present invention are for description only and do not represent the advantages or disadvantages of the embodiments.

[0100] In the above embodiments of the present invention, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0101] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only exemplary. For example, the division of the units can be a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0102] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0103] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0104] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk, etc. Various media that can store program codes.

[0105] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A data processing method, characterized in that: include: Obtaining raw data, wherein the raw data is generated from a preset data set, the raw data is obtained based on a formatted data set, the formatted data set is generated by formatting an initial data set, the initial data set is obtained from the preset data set, the preset data set is real-time monitoring data uploaded by different vehicles, the preset data set includes data collected in multiple different scenarios, and the raw data includes at least data obtained based on the position, length, width, height, and orientation of the vehicle; Calling a perception fusion algorithm to process the raw data to obtain a fusion result; Based on the fusion result, generating evaluation indicators corresponding to the perception fusion algorithm, wherein the evaluation indicators are used to evaluate the performance of the perception fusion algorithm, and the evaluation indicators are obtained by placing the fusion result into a preset location of the evaluation code and then running the evaluation code, the evaluation code is used to represent the code that generates the evaluation indicators, and the evaluation code has the preset location where the fusion result is stored, the evaluation indicators include basic indicators and fusion indicators, the basic indicators include at least: missed detection, false alarm, incorrect association, correct association, and recall rate, the fusion indicators include at least multi-target tracking accuracy, multi-target tracking precision, and average multi-target tracking accuracy, and the perception fusion algorithm is an algorithm that performs multi-sensor fusion in the autonomous driving perception module; The calling of the perception fusion algorithm to process the raw data to obtain a fusion result includes: reading a scene name from a folder of the raw data; reading formatted data sets of different sensors in a preset order according to the scene name; parsing the raw data based on the formatted data set to obtain a data field; and processing the data field using the perception fusion algorithm to obtain a fusion result, wherein the data field is a sensor data frame.

2. The method according to claim 1, characterized in that Obtaining raw data includes: Acquiring the preset data set, wherein the preset data set includes: detection results detected by different sensors during the vehicle's driving in different scenarios; The preset data set is processed to generate the original data, wherein the original data may further include: yaw angle, speed, obstacle polygon and obstacle detection frame.

3. The method according to claim 2, characterized in that Processing the preset data set to generate the original data includes: The detection results under different scenarios are processed to obtain the original data under the different scenarios.

4. The method according to claim 3, characterized in that The perception fusion algorithm is called to process the original data under the different scenarios to obtain the fusion result.

5. The method according to any one of claims 1 to 4, characterized in that The perception fusion algorithm is called to process the raw data to obtain the fusion results including: Parsing the original data to obtain data fields; The data field is processed using the perception fusion algorithm to obtain the fusion result.

6. The method according to claim 5, characterized in that Processing the data field using the perception fusion algorithm to obtain the fusion result includes: Initializing the perception fusion algorithm; The data fields are traversed in sequence and the perception fusion algorithm is called to obtain the fusion result.

7. The method according to claim 3, characterized in that The fusion result is stored in a preset format.

8. The method according to claim 1, characterized in that Based on the fusion result, the evaluation index corresponding to the perception fusion algorithm is generated, including: Storing the fusion result in a preset location of the evaluation code; The evaluation code is run to generate the evaluation index.

9. A vehicle communication device, characterized in that: include: an acquisition module, configured to acquire target code data in the autonomous driving system, wherein the target code data is used for data communication between other modules; A construction module, configured to construct a target library based on the target code data; The communication module is used to call the target library to perform data communication using the intermediate communication layer.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the data processing method according to any one of claims 1 to 8.

Images