Intelligent driving test method and device, vehicle and medium

Abstract

The invention relates to the technical field of computer information processing, and discloses an intelligent driving test method and device, a vehicle and a medium, and the method comprises the steps: obtaining the own vehicle positioning information, perception data and map data of a road test vehicle in the driving process, carrying out the timestamp alignment of other data based on the timestamp of the own vehicle positioning information, converting the data after timestamp alignment to the same coordinate system, and constructing drive test scene data; constructing a scene file based on the corresponding relationship between the self-vehicle positioning information of each frame and the sensing data and the map data; the simulation engine calculates current self-vehicle positioning information in real time according to the vehicle dynamics model, matches sensing data in a scene file based on the current self-vehicle positioning information, sends the sensing data to the test end of the intelligent driving algorithm to be tested, and sends the current self-vehicle positioning information to the map engine; and the map engine matches the map data in the scene file and sends the map data to the test end for simulation test. According to the invention, the accuracy and authenticity of the simulation test are effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of computer information processing technology, and in particular to an intelligent driving test method, device, vehicle and medium. Background Art

[0002] Before intelligent driving technology can be put into large-scale application, it needs to go through a rigorous and complete testing and verification process. Traditional road testing needs to screen out scenarios that cause functional failures from a large amount of test mileage. Considering test efficiency and cost, it faces problems such as long cycle, high cost, strong randomness of test scenarios, and the inability to regress and verify problems. It is difficult to meet the test requirements of high-level intelligent driving system safety and reliability, which is one of the key factors restricting the rapid development and upgrading of intelligent driving technology. Simulation testing has become an indispensable and important part of the field of intelligent driving technology testing due to its high efficiency, scenario repeatability and high coverage. Simulation testing conducts comprehensive testing and verification of specific functions, performance or behaviors of intelligent driving systems through various simulation models of intelligent driving systems based on preset simulation scenarios. The coverage and authenticity of simulation scenarios will directly affect the effect of simulation testing. At present, there are many solutions for the coverage and authenticity of simulation scenarios. The core idea of ​​the mainstream technical solutions is to convert road test data into dynamic and static simulation files of the standard OPEN series, and then combine them into simulation scenarios. This method can solve the reusability of simulation scenarios and the authenticity of traffic flow, but in the process of data format conversion, especially the map data needs to be abstracted, so some data content will be lost, resulting in some differences from the actual road test scenarios, affecting the accuracy of the intelligent driving test results. Summary of the invention

[0003] In view of this, the present invention provides an intelligent driving test method, device, vehicle and medium to solve the problem that the existing intelligent driving test scheme for new energy vehicles cannot perform targeted tests on power interruption and stall problems caused by battery power cutting during the test phase.

[0004] In a first aspect, the present invention provides an intelligent driving test method, the method comprising:

[0005] Acquire the vehicle positioning information, perception data and map data of the road test vehicle during driving, wherein the perception data is collected by various sensors carried by the road test vehicle;

[0006] Performing time stamp alignment on the perception data and the map data based on the time stamp corresponding to the vehicle positioning information, and converting the perception data and the map data after the time stamp alignment to the same coordinate system to construct the drive test scene data;

[0007] Based on the vehicle positioning information of each frame, extracting corresponding first perception data and first map data from the drive test scene data;

[0008] Based on the correspondence between the vehicle positioning information of each frame and the corresponding extracted first perception data and first map data, a scene file is constructed;

[0009] The simulation engine calculates the current vehicle positioning information in real time according to the vehicle dynamics model, matches the corresponding second perception data in the scenario file based on the current vehicle positioning information, and sends it to the test end of the intelligent driving algorithm to be tested, and sends the current vehicle positioning information to the map engine, so that the map engine matches the corresponding second map data in the scenario file based on the current vehicle positioning information and sends it to the test end of the intelligent driving algorithm to be tested, so that the test end performs a simulation test on the intelligent driving algorithm to be tested.

[0010] The present invention uses the vehicle positioning information, perception data and map data of the road test vehicle during driving, and uses the vehicle positioning information timestamp as a reference to align the timestamps of all data, and unify the coordinates to obtain road test scene data with consistent time sequence and the same coordinate parameters, and then constructs a scene file by separating and extracting the road test scene data according to the vehicle positioning information of each frame. Since the scene file performs wireless data format conversion during the generation process, the scene file retains the real road test data to the maximum extent, meets the requirements of high fidelity, high reusability and generalization of the intelligent driving algorithm, and respectively processes the two different data types of perception data and map data through the simulation engine and the map engine, and provides them to the test end of the intelligent driving algorithm for simulation testing, thereby providing an accurate data basis for the testing of the intelligent driving algorithm and subsequent optimization and product application, and effectively improving the accuracy and authenticity of the simulation test results.

[0011] In an optional implementation, the scene file is constructed based on the correspondence between the vehicle positioning information of each frame and the corresponding extracted first perception data and first map data, including:

[0012] Based on the correspondence between the vehicle positioning information of each frame and the corresponding extracted first perception data and first map data, a key value structure of each frame is formed with the vehicle positioning information as a key value and the corresponding extracted first perception data and first map data as values;

[0013] The key-value structure of each frame is serialized in the order of the corresponding timestamps of each frame to form a scene file.

[0014] The present invention forms a key-value structure with the corresponding perception data and map data using the vehicle positioning information as the key value, thereby facilitating subsequent accurate search of data in the scene file, and serializing the key-value structure in a timestamp manner, thereby facilitating long-term storage and transmission of the scene file.

[0015] In an optional implementation, after constructing a scene file based on the correspondence between the vehicle positioning information of each frame and the corresponding extracted first perception data and first map data, the method further includes:

[0016] Loading the scene file into the visual interface for quality inspection and review;

[0017] After passing the quality inspection, anti-collision processing is performed on each traffic participant in the scenario file;

[0018] Add attribute information to the scene file after anti-collision processing to obtain an updated scene file.

[0019] The present invention ensures that the scene files meet the data requirements of the test data required by the intelligent driving algorithm and the actual driving environment by performing quality inspection and anti-collision processing on the scene files, thereby further improving the accuracy and effectiveness of the intelligent driving algorithm test, and adding attribute information to the scene files to facilitate the subsequent classification management and application of the scene files.

[0020] In an optional implementation, the performing anti-collision processing on each traffic participant in the scene file includes:

[0021] Predicting the trajectory of each traffic participant in the scenario file;

[0022] Determining whether the predicted driving trajectory of each traffic participant conflicts with the predicted driving trajectory of the road test vehicle;

[0023] When the predicted driving trajectory of the current traffic participant conflicts with the predicted driving trajectory of the road test vehicle, data related to the current traffic participant in the scenario file is removed.

[0024] The present invention predicts the trajectory of each traffic participant in the scenario file, screens out traffic participants that conflict with the predicted driving trajectory of the road test vehicle, and deletes their related data from the scenario file to avoid such abnormal driving scenarios from interfering with the simulation test of the intelligent driving algorithm, thereby further improving the accuracy and effectiveness of the simulation test of the intelligent driving algorithm.

[0025] In an optional implementation, before the simulation engine calculates the current vehicle positioning information in real time according to the vehicle dynamics model, and matches the corresponding second perception data in the scenario file based on the current vehicle positioning information and sends it to the test end of the intelligent driving algorithm to be tested, the method further includes:

[0026] Loading the scene file, deserializing the scene file, and obtaining perception data and map data based on the key-value separation;

[0027] The perception data is input into the simulation engine, and the map data is input into the map engine.

[0028] The present invention loads and deserializes the scene file before applying it, separates the perception data and map data accordingly, and sends them to the simulation engine and the map engine respectively, so as to facilitate the simulation engine and the map engine to process the data in the simulation file and improve the data processing efficiency.

[0029] In an optional implementation, the simulation engine matches the corresponding second perception data in the scenario file based on the current vehicle positioning information and sends the data to the test end of the intelligent driving algorithm to be tested, including:

[0030] The simulation engine matches the third perception data closest to the position corresponding to the current vehicle positioning information in the scenario file;

[0031] Calculating current dynamic driving data of the road test vehicle using the vehicle dynamics model, and replacing corresponding dynamic driving data in the third perception data based on the current dynamic driving data to obtain the second perception data;

[0032] The second perception data is sent to the test end of the intelligent driving algorithm to be tested according to the preset engine task scheduling sequence.

[0033] After using the simulation engine to match the perception data closest to the position corresponding to the current vehicle positioning information, the present invention further ensures the accuracy of the perception data provided to the intelligent driving algorithm by using the vehicle dynamics model to calculate the current dynamic driving data of the road test vehicle to replace the corresponding dynamic driving data in the perception data. By sending the perception data according to the preset engine task scheduling sequence, the consistency of the perception data input of the intelligent driving algorithm and the actual road test scenario is ensured, further improving the accuracy and effectiveness of the simulation test of the intelligent driving algorithm.

[0034] In an optional implementation, the map engine matches the corresponding second map data in the scene file based on the current vehicle positioning information and sends the second map data to the test end of the intelligent driving algorithm to be tested, including:

[0035] The simulation engine matches the second map data closest to the position corresponding to the current vehicle positioning information in the scenario file;

[0036] The second map data is sent to the test end of the intelligent driving algorithm to be tested according to the preset engine task scheduling sequence.

[0037] After using the map engine to match the map data closest to the position corresponding to the current vehicle positioning information, the present invention sends the perception data according to the preset engine task scheduling sequence, thereby ensuring the consistency of the map data input of the intelligent driving algorithm with the actual road test scenario, and further improving the accuracy and effectiveness of the simulation test of the intelligent driving algorithm.

[0038] In a second aspect, the present invention provides an intelligent driving test device, the device comprising:

[0039] An acquisition module is used to acquire the vehicle positioning information, perception data and map data of the road test vehicle during driving, wherein the perception data is collected by various sensors carried by the road test vehicle;

[0040] A first processing module, configured to perform timestamp alignment on the perception data and the map data based on the timestamp corresponding to the vehicle positioning information, and convert the perception data and the map data after the timestamp alignment into the same coordinate system to construct the drive test scene data;

[0041] A second processing module, configured to extract corresponding first perception data and first map data from the drive test scene data based on the vehicle positioning information of each frame;

[0042] A third processing module is used to construct a scene file based on the correspondence between the vehicle positioning information of each frame and the corresponding extracted first perception data and first map data;

[0043] The fourth processing module is used for the simulation engine to calculate the current vehicle positioning information in real time according to the vehicle dynamics model, match the corresponding second perception data in the scene file based on the current vehicle positioning information, and send it to the test end of the intelligent driving algorithm to be tested, and send the current vehicle positioning information to the map engine, so that the map engine matches the corresponding second map data in the scene file based on the current vehicle positioning information and sends it to the test end of the intelligent driving algorithm to be tested, so that the test end performs a simulation test on the intelligent driving algorithm to be tested.

[0044] In a third aspect, the present invention provides a vehicle, comprising:

[0045] The memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method provided by the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0046] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the method provided in the first aspect or any corresponding embodiment thereof.

[0047] Beneficial effects of the present invention:

[0048] The intelligent driving test solution provided by the embodiment of the present invention uses the self-vehicle positioning information, perception data and map data of the road test vehicle during driving, and uses the self-vehicle positioning information timestamp as a reference to align the timestamps of all data, and unify the coordinates to obtain road test scene data with consistent time sequence and the same coordinate parameters, and then constructs a scene file by separating and extracting the road test scene data according to the self-vehicle positioning information of each frame. Since the scene file converts the data format wirelessly during the generation process, the scene file retains the real road test data to the maximum extent, meets the requirements of high fidelity, high reusability and generalization of the intelligent driving algorithm, and respectively processes the two different data types of perception data and map data through the simulation engine and the map engine, and provides them to the test end of the intelligent driving algorithm for simulation testing, thereby providing an accurate data basis for the testing of the intelligent driving algorithm and subsequent optimization and product application, and effectively improving the accuracy and authenticity of the simulation test results. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0050] Figure 1 is a flow chart of an intelligent driving test method according to an embodiment of the present invention;

[0051] Figure 2 is a flow chart of another intelligent driving test method according to an embodiment of the present invention;

[0052] Figure 3 An overall operation schematic diagram of the intelligent driving test is shown;

[0053] Figure 4 A specific process diagram of intelligent driving test is shown;

[0054] Figure 5 The specific process diagram of data collection and preprocessing is shown;

[0055] Figure 6 A schematic diagram of the structure of the data model is shown;

[0056] Figure 7 A schematic diagram showing the specific process of scene data generation and storage is shown;

[0057] Figure 8 A schematic diagram of the application of scene data is shown;

[0058] Fig. 9 shows a schematic diagram of engine task scheduling timing;

[0059] Fig.10 is a structural schematic diagram of an intelligent driving test device according to an embodiment of the present invention;

[0060] Fig.11 is a schematic structural diagram of a vehicle according to an embodiment of the present invention. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution 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 described embodiments are 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0062] Simulation testing has become an indispensable and important part of the intelligent driving technology testing field due to its high efficiency, scenario repeatability and high coverage. Simulation testing conducts comprehensive testing and verification of specific functions, performance or behaviors of intelligent driving systems through various simulation models of intelligent driving systems based on preset simulation scenarios. The coverage and authenticity of simulation scenarios will directly affect the effect of simulation testing. At present, there are many solutions for the coverage and authenticity of simulation scenarios. The core idea of ​​the mainstream technical solutions is to convert road test data into dynamic and static simulation files of the standard OPEN series, and then combine them into simulation scenarios. This method can solve the reusability of simulation scenarios and the authenticity of traffic flow, but in the process of data format conversion, especially map data needs to be abstracted, so some data content will be lost, resulting in some differences from the actual road test scenarios, affecting the accuracy of intelligent driving test results.

[0063] Based on the above problems, an embodiment of the present invention provides a solution for automatically constructing a generalizable high-fidelity simulation scenario based on road test data to simulate and test the intelligent driving algorithm. The main process of the solution is as follows:

[0064] 1. Record the sensor data (camera, lidar, millimeter-wave radar, GPS, IMU, etc.) and map data during driving, and filter, compensate, interpolate and align timestamps on the sensor data, vehicle posture data and map data. Combine the vehicle positioning information to convert the sensor data and map data into coordinate systems, so that all dynamic and static targets and map data are in the same UTM coordinate system. Finally, fuse the data to generate a data model.

[0065] 2. Extract the vehicle positioning information, perception data and map data from the data model. The positioning information includes the lateral and longitudinal displacement, pitch angle, roll angle and heading angle of the vehicle in the UTM coordinate system. The map data contains all map elements, such as lanes, crosswalks, intersections and traffic lights. The perception data can be subdivided into static elements and dynamic elements. Static elements are lane lines, speed limit signs, traffic signals and their status, and dynamic elements are vehicles, pedestrians and non-motor vehicles. Perception data and map data are collectively referred to as scene data. Scene data is recorded in frames according to the original data. After data preprocessing, the positioning information of each frame can be matched to a corresponding set of scene data. The key-value structure of each frame is composed of the positioning information as the key value and the scene data as the value. The key-value structure data is serialized and written into the scene file. The scene file can be imported into the scene editor. The trajectory driving points of some traffic participants can be changed manually or automatically in the scene editor. The changed trajectory points and nearby trajectory points are regenerated by multiple curve fitting, and the vehicle dynamics are used as constraints. This supports the generalization of scene data.

[0066] 3. Before entering the scene library, the scene file needs to be reviewed for scene data, scene label, and anti-collision strategy. During simulation, when the traffic participant prediction conflicts with the vehicle's driving trajectory (time and position), the corresponding traffic participant information will be blocked and no data will be output to the algorithm to be tested;

[0067] 4. The scene data application loads the scene file, separates the perception data and the map data, and inputs them into the simulation engine and the map engine respectively. The simulation engine calculates the vehicle positioning information in real time according to the vehicle dynamics model, and matches the perception data with the closest position in the perception data, and replaces the relevant elements in the source data with the dynamic data calculated by the vehicle dynamics model. Finally, the perception data is sent to the intelligent driving algorithm (hereinafter referred to as the intelligent driving algorithm) according to the engine scheduling sequence. At the same time, the simulation engine passes the real-time calculated positioning information to the map engine. The map engine matches the nearest map data in real time according to the input positioning information, and inputs the matched map data to the intelligent driving algorithm according to the engine scheduling sequence. The intelligent driving algorithm makes appropriate planning and control instructions based on the input of the simulation engine and the map engine.

[0068] In this way, under the premise of maximizing the preservation of original data and special anti-collision strategy processing, the requirements of high fidelity, high reusability and generalization of simulation scenes are met, and the dependence on map vendors and surveying and mapping qualifications is eliminated, effectively improving the accuracy and authenticity of simulation tests.

[0069] According to an embodiment of the present invention, an embodiment of an intelligent driving test 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 may be executed in an order different from that shown here.

[0070] In an embodiment of the present invention, an intelligent driving test method is provided, which is applied to computer equipment such as a single chip microcomputer, a CPU, etc. Figure 1 is a flow chart of an intelligent driving test method according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0071] Step S101, obtaining the vehicle positioning information, perception data and map data of the road test vehicle during driving.

[0072] The above-mentioned perception data is collected by various sensors carried by the road test vehicle, such as cameras, radars, etc. The vehicle positioning information can be obtained through positioning components such as GPS, or through the fusion processing of GPS, IMU and camera data on the road test vehicle to calculate the precise positioning information of the vehicle, and the map data can be obtained through the map application software carried by the road test vehicle, but the present invention is not limited to this.

[0073] Specifically, in actual applications, the original road test data collected by the road test vehicle during the road test is collected by various acquisition devices on the vehicle, including the vehicle's posture data (including the vehicle's positioning information), sensor perception data and map data, etc. The acquisition equipment needs to ensure that each frame of data has corresponding timestamp information, try to avoid frame loss or delay of the signal, and the sampling period of the vehicle's posture data is less than or equal to the period of the sensor and map data.

[0074] Step S102 , aligning the timestamps of the perception data and the map data based on the timestamp corresponding to the vehicle positioning information, and converting the perception data and the map data after the timestamp alignment to the same coordinate system to construct the drive test scene data.

[0075] Specifically, based on the timestamp of the vehicle positioning information, various sensor data and map data are filtered, compensated, interpolated, and timestamp aligned. In addition, the sensor data and map data are converted into coordinates based on the vehicle posture information, so that all dynamic and static targets and map data are in the same UTM coordinate system.

[0076] In practical applications, since the data formats collected by the equipment on the road test vehicle, i.e. the road test collection vehicle, are different and the signal periods are inconsistent, it is necessary to pre-process the original real vehicle data, including the removal of abnormal values, interpolation compensation of frame loss signals, sampling point and period alignment of various types of signals, signal unit conversion and coordinate system conversion. For example, some sensor data is based on the VCS coordinate system, and the map data is the global coordinate system. All data signals need to be converted to the UTM coordinate system before unified modeling and processing. For the original road test data with a frequency higher than the predetermined sampling frequency, data is extracted from the original road test data after format conversion according to the predetermined sampling frequency; for the original road test data with a frequency lower than the predetermined sampling frequency, data is filled in the original road test data after format conversion according to the predetermined sampling frequency, such as by linear interpolation. The adjusted original road test data is filtered and denoised to obtain standard data content. The purpose of denoising is to make the constructed scene data closer to the road test scene data at that time.

[0077] Step S103: extracting corresponding first perception data and first map data from the drive test scene data based on the vehicle positioning information of each frame.

[0078] Specifically, the positioning information includes the lateral and longitudinal displacements, pitch angles, roll angles and heading angles of the vehicle in the UTM coordinate system. The perception data can be subdivided into static elements and dynamic elements. Static elements include lane lines, speed limit signs, traffic signals and their status, etc., and dynamic elements include vehicles, pedestrians and non-motor vehicles, etc. The map data contains all map elements, such as lanes, crosswalks, intersections and traffic lights, etc. The perception data can be subdivided into static elements and dynamic elements. Static elements include lane lines, speed limit signs, traffic signals and their status, etc., and dynamic elements include vehicles, pedestrians and non-motor vehicles, etc. The present invention takes this as an example only and is not limited to this.

[0079] Step S104: construct a scene file based on the correspondence between the vehicle positioning information of each frame and the corresponding extracted first perception data and first map data.

[0080] Specifically, the perception data and map data extracted above are collectively referred to as scene data, and the scene data is recorded in frames according to the original data. After data preprocessing, the positioning information of each frame can be matched to a corresponding set of scene data, and then the key-value structure of each frame is composed of the positioning information as the key value and the scene data as the value. The key-value structure data is then serialized and written into the scene file. The scene file can be imported into the scene editor, and the trajectory driving points of some traffic participants can be changed manually or automatically in the scene editor. The changed trajectory points and nearby trajectory points are regenerated by multiple curve fitting methods, and vehicle dynamics are used as constraints, which supports the generalization of scene data.

[0081] In step S105, the simulation engine calculates the current vehicle positioning information in real time according to the vehicle dynamics model, matches the corresponding second perception data in the scene file based on the current vehicle positioning information, and sends it to the test end of the intelligent driving algorithm to be tested, and sends the current vehicle positioning information to the map engine, so that the map engine matches the corresponding second map data in the scene file based on the current vehicle positioning information and sends it to the test end of the intelligent driving algorithm to be tested, so that the test end performs a simulation test on the intelligent driving algorithm to be tested.

[0082] Among them, the test end is a terminal device running the intelligent driving algorithm to be tested, which can be a computer device installed on and executing the above steps S101 to S105, or it can be other external terminal devices, such as an external computer device specially used for intelligent driving tests, etc. When the intelligent driving algorithm to be tested is installed on a vehicle, the test end can also be the vehicle's computer. This is just an example, and the present invention is not limited to this.

[0083] Specifically, the simulation engine calculates the vehicle positioning information in real time based on the vehicle dynamics model, matches the perception data with the closest position in the perception data, replaces the relevant elements in the source data with the dynamic data calculated by the vehicle dynamics model, and finally sends the perception data to the intelligent driving algorithm according to the engine scheduling sequence. At the same time, the simulation engine passes the real-time calculated positioning information to the map engine. The map engine matches the nearest map data in real time according to the input positioning information, and inputs the matched map data to the test end according to the engine scheduling sequence to run the intelligent driving algorithm. The intelligent driving algorithm makes appropriate planning and control instructions based on the input of the simulation engine and the map engine.

[0084] The embodiment of the present invention uses the self-vehicle positioning information, perception data and map data of the road test vehicle during driving, and uses the self-vehicle positioning information timestamp as a reference to align the timestamps of all data, and unify the coordinates to obtain road test scene data with consistent time sequence and the same coordinate parameters, and then constructs a scene file by separating and extracting the road test scene data according to the self-vehicle positioning information of each frame. Since the scene file performs wireless data format conversion during the generation process, the scene file retains the real road test data to the maximum extent, meets the requirements of high fidelity, high reusability and generalization of the intelligent driving algorithm, and respectively processes the two different data types of perception data and map data through the simulation engine and the map engine, and provides them to the test end of the intelligent driving algorithm for simulation testing, thereby providing an accurate data basis for the testing of the intelligent driving algorithm and subsequent optimization and product application, and effectively improving the accuracy and authenticity of the simulation test results.

[0085] In an embodiment of the present invention, an intelligent driving test method is also provided, which is applied to computer equipment such as a single chip microcomputer, a CPU, etc. Figure 2 is a flow chart of an intelligent driving test method according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps:

[0086] Step S201, obtaining the vehicle positioning information, perception data and map data of the road test vehicle during driving, where the perception data is collected by various sensors mounted on the road test vehicle. Figure 1 The relevant description of step S101 is not repeated here.

[0087] Step S202: align the timestamps of the perception data and the map data based on the timestamp corresponding to the vehicle positioning information, and convert the perception data and the map data after the timestamp alignment to the same coordinate system to construct the drive test scene data. Figure 1 The description of step S102 is not repeated here.

[0088] Step S203: extract corresponding first perception data and first map data from the drive test scene data based on the vehicle positioning information of each frame. Figure 1 The relevant description of step S103 is not repeated here.

[0089] Step S204: construct a scene file based on the correspondence between the vehicle positioning information of each frame and the corresponding extracted first perception data and first map data.

[0090] Specifically, the above step S204 includes:

[0091] Step S2041, based on the correspondence between the vehicle positioning information of each frame and the corresponding extracted first perception data and first map data, a key value structure of each frame is formed with the vehicle positioning information as the key value and the corresponding extracted first perception data and first map data as the value.

[0092] Step S2042, serializing the key value structure of each frame in the order of the corresponding timestamps of each frame to form a scene file.

[0093] Specifically, the positioning information of each frame can be matched to a corresponding set of scene data (perception data and map data), and then the key-value structure of each frame is composed of the positioning information as the key value and the scene data as the value. The key-value structure data is then serialized and written into the scene file.

[0094] The embodiment of the present invention forms a key-value structure with the corresponding perception data and map data using the vehicle positioning information as the key value, thereby facilitating subsequent accurate search of data in the scene file, and serializing the key-value structure in a timestamp manner, thereby facilitating long-term storage and transmission of the scene file.

[0095] Step S205: Load the scene file into the visual interface for quality inspection.

[0096] Specifically, the scene files of the visualization interface can be manually reviewed to manually eliminate data that does not meet the requirements of the intelligent driving algorithm test.

[0097] Step S206, after passing the quality inspection, anti-collision processing is performed on each traffic participant in the scene file.

[0098] Specifically, the above step S206 includes: performing trajectory prediction for each traffic participant in the scene file; determining whether the predicted driving trajectory of each traffic participant conflicts with the predicted driving trajectory of the road test vehicle; when the predicted driving trajectory of the current traffic participant conflicts with the predicted driving trajectory of the road test vehicle, removing the data related to the current traffic participant in the scene file.

[0099] In actual applications, each traffic participant in the scenario file needs special collision avoidance strategy processing. During simulation, when it is predicted that the vehicle will conflict with the vehicle's driving trajectory (time and position), the corresponding traffic participant data will be blocked and no data will be output to the intelligent driving algorithm. It should be noted that the specific process of predicting the driving trajectory is a prior art and can be implemented using the relevant trajectory prediction algorithm in the prior art, which will not be described in detail here.

[0100] The embodiment of the present invention predicts the trajectory of each traffic participant in the scenario file, screens out traffic participants that conflict with the predicted driving trajectory of the road test vehicle, and deletes their related data from the scenario file, so as to avoid such abnormal driving scenarios from interfering with the simulation test of the intelligent driving algorithm, and further improve the accuracy and effectiveness of the simulation test of the intelligent driving algorithm.

[0101] Step S207, adding attribute information to the scene file after the anti-collision processing to obtain an updated scene file.

[0102] Specifically, the attribute information includes information such as scene labels and judgment criteria to form a complete test case file that can be used for intelligent driving algorithm testing and is stored in the scene library.

[0103] The embodiment of the present invention ensures that the scene files meet the data requirements of the test data required by the intelligent driving algorithm and the actual driving environment by performing quality inspection and anti-collision processing on the scene files, thereby further improving the accuracy and effectiveness of the intelligent driving algorithm test, and adding attribute information to the scene files to facilitate the subsequent classification management and application of the scene files.

[0104] Step S208, loading the scene file, deserializing the scene file, and obtaining perception data and map data based on key-value separation.

[0105] Step S209, input the perception data into the simulation engine, and input the map data into the map engine.

[0106] The embodiment of the present invention loads and deserializes the scenario file before applying it, thereby separating the perception data and map data accordingly, and sending them to the simulation engine and the map engine respectively, so as to facilitate the simulation engine and the map engine to process the data in the simulation file and improve data processing efficiency.

[0107] In step S210, the simulation engine calculates the current vehicle positioning information in real time according to the vehicle dynamics model, matches the corresponding second perception data in the scene file based on the current vehicle positioning information, and sends it to the test end of the intelligent driving algorithm to be tested, and sends the current vehicle positioning information to the map engine, so that the map engine matches the corresponding second map data in the scene file based on the current vehicle positioning information and sends it to the test end of the intelligent driving algorithm to be tested, so that the test end performs a simulation test on the intelligent driving algorithm to be tested.

[0108] Specifically, in the above step S210, the simulation engine matches the corresponding second perception data in the scene file based on the current vehicle positioning information and sends it to the test end of the intelligent driving algorithm to be tested. The specific process is as follows:

[0109] The simulation engine matches the third perception data closest to the position corresponding to the current vehicle positioning information in the scenario file; uses the vehicle dynamics model to calculate the current dynamic driving data of the road test vehicle, and replaces the corresponding dynamic driving data in the third perception data based on the current dynamic driving data to obtain the second perception data; sends the second perception data to the test end of the intelligent driving algorithm to be tested according to the preset engine task scheduling sequence.

[0110] After using the simulation engine to match the perception data closest to the position corresponding to the current vehicle positioning information, the embodiment of the present invention further ensures the accuracy of the perception data provided to the intelligent driving algorithm by using the vehicle dynamics model to calculate the current dynamic driving data of the road test vehicle to replace the corresponding dynamic driving data in the perception data. By sending the perception data according to the preset engine task scheduling sequence, the consistency of the perception data input of the intelligent driving algorithm and the actual road test scenario is ensured, further improving the accuracy and effectiveness of the simulation test of the intelligent driving algorithm.

[0111] Furthermore, in the above step S210, the specific process of the map engine matching the corresponding second map data in the scene file based on the current vehicle positioning information and sending it to the test end of the intelligent driving algorithm to be tested is as follows:

[0112] The simulation engine matches the second map data closest to the position corresponding to the current vehicle positioning information in the scene file; and sends the second map data to the test end of the intelligent driving algorithm to be tested according to the preset engine task scheduling sequence.

[0113] After using the map engine to match the map data closest to the position corresponding to the current vehicle positioning information, the embodiment of the present invention sends the perception data according to the preset engine task scheduling sequence, thereby ensuring the consistency of the map data input of the intelligent driving algorithm with the actual road test scenario, and further improving the accuracy and effectiveness of the simulation test of the intelligent driving algorithm.

[0114] In actual applications, the simulation engine calculates the vehicle positioning information in real time based on the vehicle dynamics model, matches the perception data and map data with the closest position in the perception data, and uses the dynamic data calculated by the vehicle dynamics model to replace the relevant elements in the source data. Finally, the perception data is sent to the intelligent driving algorithm according to the engine task scheduling sequence. In addition, the simulation engine passes the real-time calculated positioning information to the map engine, and the map engine matches the nearest map data in real time according to the input positioning information, and inputs the matched map data to the test end of the intelligent driving algorithm according to the engine scheduling sequence. The intelligent driving algorithm running in the test end makes appropriate planning and control instructions based on the input of the simulation engine and the map engine, which can ensure that the dynamic and static data of the simulation output are consistent with those during the road test, making the realism of the simulation test closer to the road test, but maintaining the flexibility, convenience and high repeatability of the simulation test.

[0115] The specific working process and working principle of the intelligent driving test solution provided by the embodiment of the present invention will be described in detail below in conjunction with specific application examples.

[0116] In this example, the intelligent driving algorithm is simulated and tested by automatically building a generalizable high-fidelity simulation scenario based on the road test data. The main purpose is to make the intelligent driving simulation scenario closer to the real road conditions and facilitate subsequent scene generalization and other operations. The overall solution is referenced Figure 3 and Figure 4 As shown, the specific process includes the following:

[0117] 1. Data recording and preprocessing, processing flow reference Figure 5 :

[0118] (1) Collect various types of sensor data and map data. Each frame of each data type needs to have an accurate timestamp.

[0119] (2) The GPS, IMU and camera data are integrated and processed to calculate the precise positioning information of the vehicle.

[0120] (3) Based on the timestamp of the vehicle positioning information, filter, compensate, interpolate, and align the timestamps of various sensor data and map data, and transform the sensor data and map data based on the vehicle posture information so that all dynamic and static targets and map data are in the same UTM coordinate system. Finally, the lane and target data are fused to construct a data model, which is the above-mentioned road test scenario data. The data model structure refers to Figure 6 .

[0121] 2. Scene data generation and storage, processing flow reference Figure 7 :

[0122] (1) Extract the ego-vehicle positioning information of each frame from the data model. The positioning information includes the lateral and longitudinal displacements, pitch angle, roll angle, and heading angle of the ego-vehicle in the UTM coordinate system.

[0123] (2) Map data is extracted from the data model. Map data includes all map elements, such as lanes, crosswalks, intersections, and traffic lights. Perception data can be subdivided into static elements and dynamic elements. Static elements include lane lines, speed limit signs, traffic signals and their status, and dynamic elements include vehicles, pedestrians, and non-motor vehicles.

[0124] (3) Extracting perception data from the data model. Perception data can be subdivided into static elements and dynamic elements. Static elements include lane lines, speed limit signs, traffic signals and their status, etc., while dynamic elements include vehicles, pedestrians, and non-motor vehicles, etc.

[0125] (4) Perception data and map data are collectively referred to as scene data, and scene data is recorded in frames based on the original data.

[0126] (5) After data preprocessing, the positioning information of each frame can be matched to a corresponding set of scene data, and then the key-value structure of each frame is composed of the positioning information as the key value and the scene data as the value. The key-value structure data is serialized and written into the scene file.

[0127] (6) The scenario file can be imported into the scenario editor, where the trajectory points of some traffic participants can be changed manually or automatically. The changed trajectory points and nearby trajectory points are regenerated by multiple curve fitting, and vehicle dynamics are used as constraints. This supports the generalization of scenario data.

[0128] 3. Scenario review and special strategy processing:

[0129] (1) The scene file is loaded into the visual interface to facilitate manual quality inspection and review operations.

[0130] (2) Each traffic participant in the scenario file requires special collision avoidance strategy processing. During simulation, when it is predicted that the vehicle will collide with the vehicle's driving trajectory, the corresponding traffic participant data will be blocked and no data will be output to the intelligent driving algorithm.

[0131] (3) Finally, add scenario labels, judgment criteria and other information to form a complete test case file and store it in the scenario library.

[0132] 4. Scene data application, processing flow reference Figure 8 :

[0133] (1) Load the scene file and perform deserialization processing to separate the perception data and map data, and input them into the simulation engine and map engine respectively.

[0134] (2) The simulation engine calculates the vehicle positioning information in real time based on the vehicle dynamics model, matches the perception data and map data with the closest location in the perception data, replaces the relevant elements in the source data with the dynamic data calculated by the vehicle dynamics model, and finally inputs the perception data into the intelligent driving algorithm according to the engine task scheduling sequence. The task scheduling sequence is shown in the figure below: Fig. 9 shown.

[0135] (3) The simulation engine passes the real-time calculated positioning information to the map engine. The map engine then matches the nearest map data in real time based on the input positioning information, and inputs the matched map data into the intelligent driving algorithm according to the engine's scheduling sequence.

[0136] (4) The intelligent driving algorithm makes appropriate planning and control instructions based on the input of the simulation engine and the map engine. This ensures that the dynamic and static data of the simulation output are consistent with those during the road test, making the realism of the simulation test closer to the road test, while maintaining the flexibility, convenience and high repeatability of the simulation test.

[0137] In this embodiment, an intelligent driving test device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0138] This embodiment provides an intelligent driving test device, such as Fig.10 As shown, the device comprises:

[0139] The acquisition module 1001 is used to acquire the vehicle positioning information, perception data and map data of the road test vehicle during driving. The perception data is collected by various sensors carried by the road test vehicle.

[0140] The first processing module 1002 is used to align the timestamps of the perception data and the map data based on the timestamp corresponding to the vehicle positioning information, and convert the perception data and the map data after the timestamp alignment to the same coordinate system to construct the drive test scene data;

[0141] The second processing module 1003 is used to extract corresponding first perception data and first map data from the drive test scene data based on the vehicle positioning information of each frame;

[0142] The third processing module 1004 is used to construct a scene file based on the correspondence between the vehicle positioning information of each frame and the corresponding extracted first perception data and first map data;

[0143] The fourth processing module 1005 is used for the simulation engine to calculate the current vehicle positioning information in real time according to the vehicle dynamics model, match the corresponding second perception data in the scene file based on the current vehicle positioning information and send it to the test end of the intelligent driving algorithm to be tested, and send the current vehicle positioning information to the map engine, so that the map engine matches the corresponding second map data in the scene file based on the current vehicle positioning information and sends it to the test end of the intelligent driving algorithm to be tested, so that the test end performs simulation testing on the intelligent driving algorithm to be tested.

[0144] In some optional implementations, the third processing module 1004 includes:

[0145] A first processing unit is used to form a key value structure of each frame by using the self-vehicle positioning information as a key value and the corresponding extracted first perception data and first map data as values ​​based on the correspondence between the self-vehicle positioning information of each frame and the corresponding extracted first perception data and first map data;

[0146] The second processing unit is used to serialize the key value structure of each frame according to the order of the timestamps corresponding to each frame to form a scene file.

[0147] In some optional implementations, the intelligent driving test device further includes:

[0148] The fifth processing module is used to load the scene file into the visual interface for quality inspection and review;

[0149] The sixth processing module is used to perform anti-collision processing on each traffic participant in the scene file after passing the quality inspection review;

[0150] The seventh processing module is used to add attribute information to the scene file after the anti-collision processing to obtain an updated scene file.

[0151] In some optional implementations, the sixth processing module includes:

[0152] A third processing unit is used to predict the trajectory of each traffic participant in the scene file;

[0153] a fourth processing unit, configured to determine whether the predicted driving trajectory of each traffic participant conflicts with the predicted driving trajectory of the road test vehicle;

[0154] The fifth processing unit is used to remove data related to the current traffic participant in the scene file when the predicted driving trajectory of the current traffic participant conflicts with the predicted driving trajectory of the road test vehicle.

[0155] In some optional implementations, the intelligent driving test device further includes:

[0156] An eighth processing module is used to load the scene file, deserialize the scene file, and obtain perception data and map data based on key-value separation;

[0157] The ninth processing module is used to input the perception data into the simulation engine and input the map data into the map engine.

[0158] In some optional implementations, the fourth processing module 1005 includes:

[0159] A sixth processing unit, configured to enable the simulation engine to match the third perception data closest to the position corresponding to the current vehicle positioning information in the scene file;

[0160] a seventh processing unit, configured to calculate current dynamic driving data of the road test vehicle using the vehicle dynamics model, and replace corresponding dynamic driving data in the third perception data based on the current dynamic driving data to obtain second perception data;

[0161] The eighth processing unit is used to send the second perception data to the test end of the intelligent driving algorithm to be tested according to a preset engine task scheduling sequence.

[0162] In some optional implementations, the fourth processing module 1005 further includes:

[0163] A ninth processing unit, configured for the simulation engine to match the second map data closest to the position corresponding to the current vehicle positioning information in the scene file;

[0164] The tenth processing unit is used to send the second map data to the test end of the intelligent driving algorithm to be tested according to a preset engine task scheduling sequence.

[0165] The intelligent driving test device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0166] The further functional description of each of the above modules and units is the same as that of the above corresponding method embodiments and will not be repeated here.

[0167] See also Fig.11 , Fig.11 is a schematic diagram of the structure of a vehicle provided by an optional embodiment of the present invention, such as Fig.11 As shown, the vehicle includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. The various components are connected to each other using different buses for communication, and can be installed on a common motherboard or installed in other ways as needed. The processor can process instructions executed in the computer device, including instructions stored in or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Fig.11 A processor 10 is taken as an example.

[0168] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0169] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0170] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created by the use of a computer device based on the presentation of a small program landing page, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0171] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0172] The vehicle further comprises a communication interface 30 for the control unit to communicate with other devices or a communication network.

[0173] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0174] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. An intelligent driving test method, characterized in that: The method comprises: Acquire the vehicle positioning information, perception data and map data of the road test vehicle during driving, wherein the perception data is collected by various sensors carried by the road test vehicle; Performing time stamp alignment on the perception data and the map data based on the time stamp corresponding to the vehicle positioning information, and converting the perception data and the map data after the time stamp alignment to the same coordinate system to construct the drive test scene data; Based on the vehicle positioning information of each frame, extracting corresponding first perception data and first map data from the drive test scene data; Based on the correspondence between the vehicle positioning information of each frame and the corresponding extracted first perception data and first map data, a scene file is constructed; The simulation engine calculates the current vehicle positioning information in real time according to the vehicle dynamics model, matches the corresponding second perception data in the scenario file based on the current vehicle positioning information, and sends it to the test end of the intelligent driving algorithm to be tested, and sends the current vehicle positioning information to the map engine, so that the map engine matches the corresponding second map data in the scenario file based on the current vehicle positioning information and sends it to the test end of the intelligent driving algorithm to be tested, so that the test end performs a simulation test on the intelligent driving algorithm to be tested.

2. The method according to claim 1, characterized in that The scene file is constructed based on the correspondence between the vehicle positioning information of each frame and the corresponding extracted first perception data and first map data, including: Based on the correspondence between the vehicle positioning information of each frame and the corresponding extracted first perception data and first map data, a key value structure of each frame is formed with the vehicle positioning information as a key value and the corresponding extracted first perception data and first map data as values; The key-value structure of each frame is serialized in the order of the corresponding timestamps of each frame to form a scene file.

3. The method according to claim 1, characterized in that After constructing a scene file based on the correspondence between the vehicle positioning information of each frame and the corresponding extracted first perception data and first map data, the method further includes: Loading the scene file into the visual interface for quality inspection and review; After passing the quality inspection, anti-collision processing is performed on each traffic participant in the scenario file; Add attribute information to the scene file after anti-collision processing to obtain an updated scene file.

4. The method according to claim 3, characterized in that The anti-collision processing for each traffic participant in the scene file includes: Predicting the trajectory of each traffic participant in the scenario file; Determining whether the predicted driving trajectory of each traffic participant conflicts with the predicted driving trajectory of the road test vehicle; When the predicted driving trajectory of the current traffic participant conflicts with the predicted driving trajectory of the road test vehicle, data related to the current traffic participant in the scenario file is removed.

5. The method according to claim 2, characterized in that: Before the simulation engine calculates the current vehicle positioning information in real time according to the vehicle dynamics model, and matches the corresponding second perception data in the scenario file based on the current vehicle positioning information and sends it to the test end of the intelligent driving algorithm to be tested, the method further includes: Loading the scene file, deserializing the scene file, and obtaining perception data and map data based on the key-value separation; The perception data is input into the simulation engine, and the map data is input into the map engine.

6. The method according to claim 1, characterized in that The simulation engine matches the corresponding second perception data in the scenario file based on the current vehicle positioning information and sends the data to the test end of the intelligent driving algorithm to be tested, including: The simulation engine matches the third perception data closest to the position corresponding to the current vehicle positioning information in the scenario file; Calculating current dynamic driving data of the road test vehicle using the vehicle dynamics model, and replacing corresponding dynamic driving data in the third perception data based on the current dynamic driving data to obtain the second perception data; The second perception data is sent to the test end of the intelligent driving algorithm to be tested according to the preset engine task scheduling sequence.

7. The method according to claim 1, characterized in that The map engine matches the corresponding second map data in the scene file based on the current vehicle positioning information and sends the second map data to the test end of the intelligent driving algorithm to be tested, including: The simulation engine matches the second map data closest to the position corresponding to the current vehicle positioning information in the scenario file; The second map data is sent to the test end of the intelligent driving algorithm to be tested according to the preset engine task scheduling sequence.

8. An intelligent driving test device, characterized in that: The device comprises: An acquisition module is used to acquire the vehicle positioning information, perception data and map data of the road test vehicle during driving, wherein the perception data is collected by various sensors carried by the road test vehicle; A first processing module is used to align the perception data and the map data with the timestamps based on the timestamp corresponding to the vehicle positioning information, and convert the perception data and the map data after the timestamp alignment into the same coordinate system to construct the drive test scene data; A second processing module, configured to extract corresponding first perception data and first map data from the drive test scene data based on the vehicle positioning information of each frame; A third processing module is used to construct a scene file based on the correspondence between the vehicle positioning information of each frame and the corresponding extracted first perception data and first map data; The fourth processing module is used for the simulation engine to calculate the current vehicle positioning information in real time according to the vehicle dynamics model, match the corresponding second perception data in the scene file based on the current vehicle positioning information, and send it to the test end of the intelligent driving algorithm to be tested, and send the current vehicle positioning information to the map engine, so that the map engine matches the corresponding second map data in the scene file based on the current vehicle positioning information and sends it to the test end of the intelligent driving algorithm to be tested, so that the test end performs a simulation test on the intelligent driving algorithm to be tested.

9. A vehicle, characterized in that: The vehicle comprises: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method according to any one of claims 1 to 7 by executing the computer instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method according to any one of claims 1 to 7.

Images