Automatic driving simulation test method, device, equipment and storage medium

Abstract

The present invention relates to the field of autonomous driving technology, and discloses an autonomous driving simulation test method, device, equipment and storage medium for improving the safety and stability of autonomous driving scene testing. The autonomous driving simulation test method comprises: obtaining a vehicle simulation test scene data set and a vehicle actual road test full data set; preprocessing and labeling the vehicle simulation test scene data set and the vehicle actual road test full data set to obtain a target test scene labeling data set, wherein the target test scene labeling data set is used to label test scenes related to the target; performing an autonomous driving simulation test according to the target test scene labeling data set to obtain an autonomous driving simulation test result.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to an autonomous driving simulation test method, device, equipment and storage medium. Background Art

[0002] With the development of autonomous driving technology, the safety of autonomous driving technology has attracted more and more attention. Existing autonomous driving vehicles are generally tested in the autonomous driving operation area, and autonomous driving vehicles can also avoid obstacles during the autonomous driving process under normal operation.

[0003] However, in the prior art, in order to ensure the safety of autonomous driving tests, the driving locations and driving times of autonomous driving vehicles are limited so that the autonomous driving vehicles can be tested within the operating area drawn by the map. This leads to a lack of test scenarios for interaction with targets (e.g., pedestrians), resulting in poor safety and stability of autonomous driving tests. Summary of the invention

[0004] The present invention provides an autonomous driving simulation test method, device, equipment and storage medium for improving the safety and stability of autonomous driving scenario testing.

[0005] To achieve the above-mentioned objectives, the first aspect of the present invention provides an autonomous driving simulation test method, comprising: obtaining a vehicle simulation test scenario dataset and a full dataset of a vehicle actual road test; preprocessing and labeling the vehicle simulation test scenario dataset and the full dataset of the vehicle actual road test to obtain a target test scene labeling dataset, wherein the target test scene labeling dataset is used to label test scenes related to the target; performing an autonomous driving simulation test according to the target test scene labeling dataset to obtain an autonomous driving simulation test result.

[0006] In a feasible implementation, the acquisition of a vehicle simulation test scenario data set and a vehicle actual road test full data set includes: acquiring autonomous driving literature data, and generating a vehicle simulation test scenario data set based on the autonomous driving literature data; performing full data extraction, cleaning and clustering operations on a preset data source in sequence through a preset data mining task to obtain a vehicle actual road test full data set, wherein the preset data source includes vehicle log information, vehicle sensor information and vehicle positioning information.

[0007] In a feasible implementation, the obtaining of autonomous driving literature data and generating a vehicle simulation test scenario dataset based on the autonomous driving literature data includes: reading autonomous driving literature data from a preset literature database, the autonomous driving literature data being used to indicate descriptive information related to a target in an autonomous driving scenario; performing named entity recognition on the autonomous driving literature data according to preset rules to obtain an entity recognition dataset, the entity recognition dataset including a vehicle initial speed, a vehicle deceleration, and a driver reaction time; and generating a vehicle simulation test scenario dataset based on the entity recognition dataset.

[0008] In a feasible implementation manner, the vehicle simulation test scenario dataset and the vehicle actual road test full dataset are preprocessed and labeled to obtain a target test scene labeled dataset, including: merging the vehicle simulation test scenario dataset and the vehicle actual road test full dataset to obtain a merged scenario dataset; calling a preset data screening model to perform data screening on the merged scenario dataset to obtain an initial test scenario dataset; performing scene annotation processing on the initial test scenario dataset through a preset data annotation model to obtain a target test scene labeled dataset, and storing the target test scene labeled dataset in a preset scene database.

[0009] In a feasible implementation manner, the autonomous driving simulation test is performed according to the target test scene annotated data set to obtain the autonomous driving simulation test result, including: performing scene screening according to the target test scene annotated data set to obtain a screened scene data set, wherein the screened scene data set is used to indicate a real scene of interaction with the target; performing scene segmentation according to the start time of each interaction annotation and the end time of each interaction annotation in the screened scene data set to obtain multiple simulation scene fragment information; converting the multiple simulation scene fragment information into multiple target simulation scenes through a preset simulation tool; performing autonomous driving simulation tests on the multiple target simulation scenes to obtain autonomous driving simulation test results.

[0010] In a feasible implementation, the autonomous driving simulation test is performed on the multiple target simulation scenarios to obtain the autonomous driving simulation test results, including: performing scene simulation evaluation on the multiple target simulation scenarios according to a preset target prediction model to obtain multiple scene safety scores, multiple scene comfort scores and multiple driver position difference values; determining the autonomous driving simulation test results according to the multiple scene safety scores, the multiple scene comfort scores and the multiple driver position difference values.

[0011] In a feasible implementation, after performing the autonomous driving simulation test according to the target test scene labeled data set and obtaining the autonomous driving simulation test results, the autonomous driving simulation test method further includes: generating an autonomous driving simulation test report according to the autonomous driving simulation test results and a preset test report template; readjusting the target test scene labeled data set according to the autonomous driving simulation test report to obtain an updated test scene labeled data set, and retesting based on the updated test scene labeled data set.

[0012] A second aspect of the present invention provides an autonomous driving simulation test device, comprising: an acquisition module, used to acquire a vehicle simulation test scenario data set and a full vehicle actual road test data set; a processing module, used to preprocess and label the vehicle simulation test scenario data set and the full vehicle actual road test data set to obtain a target test scene labeled data set, wherein the target test scene labeled data set is used to label test scenes related to the target; a testing module, used to perform an autonomous driving simulation test based on the target test scene labeled data set to obtain an autonomous driving simulation test result.

[0013] In a feasible implementation, the acquisition module also includes: a generation unit, used to acquire autonomous driving literature data, and generate a vehicle simulation test scenario data set based on the autonomous driving literature data; a processing unit, used to perform full data extraction, cleaning and clustering operations on a preset data source in sequence through a preset data mining task to obtain a full data set of actual vehicle road tests, wherein the preset data source includes vehicle log information, vehicle sensor information and vehicle positioning information.

[0014] In a feasible implementation, the generation unit is specifically used to: read autonomous driving literature data from a preset literature database, wherein the autonomous driving literature data is used to indicate descriptive information related to a target in an autonomous driving scenario; perform named entity recognition on the autonomous driving literature data according to preset rules to obtain an entity recognition data set, wherein the entity recognition data set includes vehicle initial speed, vehicle deceleration, and driver reaction time; and generate a vehicle simulation test scenario data set based on the entity recognition data set.

[0015] In a feasible implementation, the processing module is specifically used to: merge the vehicle simulation test scenario data set and the vehicle actual road test full data set to obtain a merged scenario data set; call a preset data screening model to perform data screening on the merged scenario data set to obtain an initial test scenario data set; perform scene annotation processing on the initial test scenario data set through a preset data annotation model to obtain a target test scene annotation data set, and store the target test scene annotation data set in a preset scene database.

[0016] In a feasible implementation, the test module also includes: a screening unit, used to perform scene screening according to the target test scene annotation data set to obtain a filtered scene data set, and the filtered scene data set is used to indicate a real scene of interaction with the target; a segmentation unit, used to perform scene segmentation according to the start time of each interaction annotation and the end time of each interaction annotation in the filtered scene data set to obtain multiple simulation scene fragment information; a conversion unit, used to convert the multiple simulation scene fragment information into multiple target simulation scenes through a preset simulation tool; a testing unit, used to perform autonomous driving simulation tests on the multiple target simulation scenes to obtain autonomous driving simulation test results.

[0017] In a feasible implementation, the test unit is specifically used to: perform scenario simulation evaluation on the multiple target simulation scenarios according to a preset target prediction model to obtain multiple scenario safety scores, multiple scenario comfort scores and multiple driver position difference values; determine the autonomous driving simulation test results according to the multiple scenario safety scores, the multiple scenario comfort scores and the multiple driver position difference values.

[0018] In a feasible implementation, the autonomous driving simulation test device also includes: a generation module, used to generate an autonomous driving simulation test report according to the autonomous driving simulation test results and a preset test report template; an adjustment module, used to readjust the target test scene annotation data set according to the autonomous driving simulation test report, obtain an updated test scene annotation data set, and retest based on the updated test scene annotation data set.

[0019] The third aspect of the present invention provides an autonomous driving simulation test device, comprising: a memory and at least one processor, wherein instructions are stored in the memory, and the memory and the at least one processor are interconnected through lines; the at least one processor calls the instructions in the memory so that the autonomous driving simulation test device executes the above-mentioned autonomous driving simulation test method.

[0020] A fourth aspect of the present invention provides a computer-readable storage medium, which stores instructions that, when executed on a computer, enable the computer to execute the above-mentioned autonomous driving simulation test method.

[0021] In the technical solution provided by the present invention, a vehicle simulation test scenario data set and a full data set of vehicle actual road tests are obtained; the vehicle simulation test scenario data set and the full data set of vehicle actual road tests are preprocessed and labeled to obtain a target test scenario labeled data set, and the target test scenario labeled data set is used to label test scenes related to the target; an autonomous driving simulation test is performed according to the target test scenario labeled data set to obtain an autonomous driving simulation test result. In an embodiment of the present invention, by preprocessing and labeling the vehicle simulation test scenario data set and the full data set of vehicle actual road tests to obtain a target test scenario labeled data set, and performing an autonomous driving simulation test according to the target test scenario labeled data set to obtain an autonomous driving simulation test result, the safety and stability of autonomous driving scenario testing is improved, and the cost of autonomous driving testing is reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A schematic diagram of an embodiment of an automatic driving simulation test method in an embodiment of the present invention;

[0023] Figure 2 A schematic diagram of another embodiment of the autonomous driving simulation test method in an embodiment of the present invention;

[0024] Figure 3 A schematic diagram of an embodiment of an automatic driving simulation test device in an embodiment of the present invention;

[0025] Figure 4 A schematic diagram of another embodiment of the automatic driving simulation test device in an embodiment of the present invention;

[0026] Figure 5 Schematic diagram of an embodiment of an autonomous driving simulation test device in an embodiment of the present invention. DETAILED DESCRIPTION

[0027] Embodiments of the present invention provide an autonomous driving simulation test method, apparatus, equipment and storage medium, which are used to improve the safety and stability of autonomous driving scenario testing and reduce the cost of autonomous driving testing through a vehicle simulation test scenario data set and a full data set of vehicle actual road tests.

[0028] The terms "first", "second", "third", "fourth", etc. (if any) in the specification 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 data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "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 that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0029] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , an embodiment of the automatic driving simulation test method in the embodiment of the present invention includes:

[0030] 101. Obtain a vehicle simulation test scenario dataset and a full dataset of vehicle actual road tests.

[0031] Among them, the vehicle simulation test scenario data set includes a simulation scenario data set associated with the target, and the vehicle actual road test full data set includes an actual scenario data set associated with the target. The target may include pedestrians and animals, and may also include other movable objects, which are not specifically limited here.

[0032] Specifically, the server receives an autonomous driving simulation test request sent by a target terminal; the server parses and verifies the autonomous driving simulation test request to obtain a scenario test identifier; the server sets the scenario test identifier as a target key, and queries a preset scenario database based on the target key to obtain a target value; the server determines whether the target value is a null value; if the target value is not a null value, the server generates scenario test identifier exception information, and sends the scenario test identifier exception information to the target terminal, so that the target terminal prompts the target user to reset the scenario test identifier according to the scenario test identifier exception information; if the target value is not a null value, the server extracts the field value corresponding to the preset field name from the target value according to the preset field name, and obtains a vehicle simulation test scenario data set and a full data set of vehicle actual road tests.

[0033] It should be noted that both the vehicle simulation test scenario dataset and the vehicle actual road test full dataset include target-related test scenarios. The vehicle simulation test scenario dataset is used to indicate a set of target-related test scenarios of simulated unmanned vehicles in different areas (e.g., commercial areas, residential areas, development areas, etc.), road features (e.g., road grade, road structure, road edges, etc.), traffic facilities (e.g., traffic signs, signal lights, etc.), and environmental conditions (e.g., weather, particulate matter, lighting, etc.) formed in advance based on autonomous driving literature data. The vehicle actual road test full dataset is used to indicate the full dataset obtained by conducting multiple scheduled actual road tests in advance for the autonomous driving pre-operation design area, and then filtering out target-related test scenarios from the full dataset, and synthesizing the test data of different test instances and different test days to obtain at least one target's actual test scenarios in different areas, road features, traffic facilities and / or environmental conditions. For example, actual test scenarios of 1 actor, 2 pedestrians or multiple actors in different areas, road features, traffic facilities and / or environmental conditions.

[0034] It is understandable that the execution subject of the present invention may be an autonomous driving simulation test device, or a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0035] 102. Preprocess and annotate the vehicle simulation test scenario dataset and the vehicle actual road test full dataset to obtain a target test scenario annotated dataset, where the target test scenario annotated dataset is used to annotate test scenarios related to the target.

[0036] It can be understood that the server performs intersection processing on the vehicle simulation test scenario data set (simulated extreme test scenarios) and the vehicle actual road test full data set (real road test scenarios within the design operation area), and filters out scenario data sets that do not meet the requirements to obtain the initial test scenario data set. The server annotates the initial test scenario data set with information in order to further subdivide the verification algorithm indicators and thereby obtain test scenarios relevant to the target.

[0037] Specifically, the server obtains data screening rules and data annotation rules; the server merges the vehicle simulation test scene data set and the vehicle actual road test full data set to obtain a merged scene data set; the server performs data screening on the merged scene data set according to the data screening rules to obtain an initial test scene data set; the server performs data annotation on the initial test scene data set according to the data annotation rules to obtain a target test scene annotation data set, and the target test scene annotation data set is used to annotate test scenes related to the target. For example, a test scene in which the target crosses the motor vehicle lane, a test scene in which the target and the unmanned vehicle drive in the opposite direction, etc. In some embodiments, the server sequentially performs data screening and data annotation processing on the merged scene data set through a preset data screening model and a preset data annotation model to obtain a target test scene annotation data set, wherein the preset data screening model and the preset data annotation model are both pre-trained models, the preset data screening model includes data screening rules, and the preset data annotation model includes data annotation rules. Further, the server writes the target test scene annotation data set into the field corresponding to the preset scene database to facilitate the execution of step 103.

[0038] It should be noted that the data labeling rules include labeling scene data of whether interaction with the target occurs on the path, labeling whether the driver's behavior is reasonable driving behavior (that is, the driver's behavior cannot be reckless and cannot be overly conservative, for example, labeling the driver's behavior as conservative driving behavior, normal driving behavior or reckless driving behavior), labeling the driver's reaction time to the target, labeling the start time of the interaction labeling between the driver and the target, and labeling the end time of the interaction labeling between the driver and the target, and other rule data.

[0039] 103. Perform an autonomous driving simulation test based on the target test scenario labeled data set to obtain an autonomous driving simulation test result.

[0040] It should be noted that the server filters out real scenes of interaction with the target based on the target test scene annotation dataset, and performs scene segmentation and simulation scene construction according to the start time and end time of each interaction annotation in the real scene of interaction with the target, so as to avoid scene failure due to excessive length of the scene, or false interaction with the target.

[0041] Specifically, the server generates a three-dimensional scene model based on a preset high-precision semantic map; the server performs scene segmentation according to the start time of each interactive annotation and the end time of each interactive annotation in the target test scene annotation data set to obtain multiple simulation scene fragment information; the server builds multiple target simulation scenes according to the multiple simulation scene fragment information through the three-dimensional scene model, and each target simulation scene includes information such as targets, vehicles, static obstacles, road networks and traffic lights, which increases the richness of each target simulation scene. Further, the server generates multiple scene files according to the multiple simulation scene fragment information, and the server stores the multiple scene files in a preset file storage system. The server performs automatic driving simulation tests on the multiple target simulation scenes to obtain automatic driving simulation test results. Specifically, the server performs automatic driving simulation tests on the multiple target simulation scenes respectively through a preset automatic driving algorithm to obtain multiple initial simulation test results, and the server determines the automatic driving simulation test results based on the multiple initial simulation test results, wherein the preset automatic driving algorithm includes a preset scene safety degree scoring algorithm, a preset scene comfort degree scoring algorithm and / or a preset driver position difference algorithm. For example, in a simulation scenario where a target suddenly emerges from a truck parked on the roadside when the vehicle is traveling straight (when the target is blocked by the parked truck before passing through, the sensors of the running vehicle cannot detect it in advance), the server detects whether the driver's preset reaction time to the target in the simulation scenario is reasonable, obtains the autonomous driving simulation test results, and adjusts the driver's preset reaction time to the target based on the autonomous driving simulation test results to improve the safety of the autonomous driving scenario test.

[0042] It should be noted that the results of the autonomous driving simulation test are used to indicate the scene safety score, scene comfort score, and the position difference between the target and the driver. For example, if the scene safety score is 3 points, the scene comfort score is 5 points, and the position difference between the target and the driver is 0.01 meters, the server determines that the scene safety level is low, the scene comfort level is low, and the driver's preset reaction time to the target is unreasonable. The server needs to reset the preset reaction time to update the simulation scene and re-run the simulation test.

[0043] In an embodiment of the present invention, a target test scene labeled data set is obtained by preprocessing and labeling a vehicle simulation test scenario data set and a full data set of vehicle actual road tests, and an autonomous driving simulation test is performed based on the target test scene labeled data set to obtain an autonomous driving simulation test result, thereby improving the safety and stability of the autonomous driving scenario test and reducing the cost of the autonomous driving test.

[0044] See also Figure 2 Another embodiment of the automatic driving simulation test method in the embodiment of the present invention includes:

[0045] 201. Obtain a vehicle simulation test scenario dataset and a vehicle actual road test full dataset.

[0046] Among them, the vehicle simulation test scenario dataset is used to indicate the simulation scenario dataset obtained from the autonomous driving literature data, and the vehicle actual road test full dataset is used to indicate the actual scenario dataset obtained from the road sections within the preset map range at different traffic times. The vehicle simulation test scenario dataset and the vehicle actual road test full dataset combine the simulation test scenario and the actual test scenario, which improves the coverage of the scenario dataset.

[0047] In some embodiments, the server obtains autonomous driving literature data, and generates a vehicle simulation test scenario data set based on the autonomous driving literature data. Specifically, the server reads the autonomous driving literature data from a preset literature database, and performs data mining processing on the autonomous driving literature data to obtain a vehicle simulation test scenario data set. The server extracts, cleans and clusters the preset data source in sequence through a preset data mining task to obtain a full data set of the vehicle's actual road test. The preset data source includes vehicle log information, vehicle sensor information and vehicle positioning information. The preset data mining task is executed once every preset time period (for example, one minute or one hour), and obtains the full data set of the vehicle's actual road test from the vehicle log information, vehicle sensor information and vehicle positioning information. The vehicle log information is used to record the vehicle's autonomous driving information, the vehicle sensor information is used to indicate the recording of various operating condition information of the vehicle during the autonomous driving process (for example, vehicle speed, temperature of various media and engine operating conditions, etc.), and the vehicle positioning information is used to indicate the tracking and positioning of the vehicle through the global satellite positioning system and the geographic information system. The server stores the vehicle simulation test scenario data set and the full data set of the vehicle's actual road test in the preset scenario database. Furthermore, the server generates a scenario test identifier, uses the scenario test identifier as a key, sets the vehicle simulation test scenario data set and the vehicle actual road test full data set as values, and stores the vehicle simulation test scenario data set and the vehicle actual road test full data set in a preset scenario database according to the mapping relationship of the key-value pairs.

[0048] In some embodiments, when the server executes the steps of obtaining the autonomous driving literature data and generating the vehicle simulation test scenario data set according to the autonomous driving literature data, the server reads the autonomous driving literature data from the preset literature database, and the autonomous driving literature data is used to indicate the description information related to the target in the autonomous driving scenario; the server performs named entity recognition on the autonomous driving literature data according to the preset rules to obtain the entity recognition data set, and the entity recognition data set includes the vehicle initial speed, the vehicle deceleration and the driver's reaction time; the server generates the vehicle simulation test scenario data set according to the entity recognition data set. Among them, the vehicle simulation test scenario data set is used to indicate the simulation test scenario related to the target. For example, the simulation test scenario is a test scenario of pedestrians crossing the road. The server obtains the vehicle initial speed v, the vehicle deceleration a and the driver's reaction time reaction_time, and calculates the braking distance Distance according to the preset formula Distance = v*reaction_time+v*v / 2*a. The server sets the target crossing time t according to the braking distance Distance, and then constructs the vehicle simulation test scenario data set.

[0049] 202. Preprocess and annotate the vehicle simulation test scenario dataset and the vehicle actual road test full dataset to obtain a target test scenario annotated dataset, where the target test scenario annotated dataset is used to annotate test scenarios related to the target.

[0050] It should be noted that the target test scene annotation dataset is used to annotate test scenes related to the target. Taking pedestrians as an example, the test scenes related to pedestrians include: test scenes of pedestrians recklessly crossing sidewalks, test scenes of pedestrians recklessly crossing non-sidewalks, test scenes of pedestrians walking along the roadside in the same direction or in the opposite direction of the unmanned vehicle, test scenes of pedestrians rushing out of obstructions, test scenes of pedestrians suddenly twisting their direction while walking, and test scenes of pedestrians stopping to pick up items while walking, etc. The specifics are not limited here. The server needs to annotate the test scenes related to the target to obtain the target test scene annotation dataset.

[0051] In some embodiments, the server merges the vehicle simulation test scene data set and the vehicle actual road test full data set to obtain a merged scene data set; the server calls a preset data screening model to perform data screening on the merged scene data set to obtain an initial test scene data set; the server performs scene annotation processing on the initial test scene data set through a preset data annotation model to obtain a target test scene annotation data set, and stores the target test scene annotation data set in a preset scene database. Wherein, the preset data screening model and the preset data annotation model are both pre-trained models, the preset data screening model is used to indicate the screening of scene data related to the target from the merged scene data set, the preset data annotation model is used to indicate the annotation of scene data related to the target from the initial test scene data set, and the server implements target-related data screening and data annotation through the preset data screening model and the preset data annotation model, thereby improving data processing efficiency and reducing data processing costs. For example, the target test scene annotation data set is used to indicate whether there is an interactive behavior on the path with the target, the evaluation data of the driver's driving behavior (for example, conservative driving behavior, normal driving behavior or reckless driving behavior), the driver's reaction time to the target, the start time of the driver's interaction annotation with the target, and the end time of the driver's interaction annotation with the target.

[0052] It is understandable that the data formats corresponding to the vehicle simulation test scenario data set and the vehicle actual road test full data set are the same, which facilitates operations such as data merging, data screening and data annotation processing, thereby improving data processing efficiency.

[0053] 203. Perform an autonomous driving simulation test based on the target test scenario labeled data set to obtain an autonomous driving simulation test result.

[0054] It should be noted that the autonomous driving simulation test is used to indicate the specific process of conducting autonomous driving testing in a target simulation scenario. In some embodiments, the server performs scene screening according to the target test scene annotation data set to obtain a screened scene data set, and the screened scene data set is used to indicate the real scene of interaction with the target; the server performs scene segmentation according to the start time of each interaction annotation and the end time of each interaction annotation in the screened scene data set to obtain multiple simulation scene fragment information, wherein there is a one-to-one correspondence between the start time of each interaction annotation and the start time of each interaction annotation; the server converts the multiple simulation scene fragment information into multiple target simulation scenes through a preset simulation tool, and the multiple target simulation scenes are used to indicate complex test scenes for interacting with the target under different areas, road features, traffic facilities and environmental conditions. The preset simulation tool can be a three-dimensional simulation software or a smart driving complex traffic scene simulation software, which is not limited here; the server performs an autonomous driving simulation test on multiple target simulation scenes to obtain an autonomous driving simulation test result, that is, the server runs a preset target prediction model under multiple target simulation scenes to obtain an initial test result corresponding to each target simulation scene, and determines the autonomous driving simulation test result according to the initial test result corresponding to each target simulation scene, wherein the preset target prediction model includes a preset scene safety degree scoring algorithm, a preset scene comfort degree scoring algorithm and a preset driver position difference algorithm. For example, the target simulation scenario is: the driver's reaction time to the target is 1.2s (1.2s is the driver's reaction time without prior information), the start time of the target simulation scenario is set to the interaction marking start time A, and the end time of the target simulation scenario is the driver's interaction marking end time B. The server performs an autonomous driving simulation test in the target simulation scenario to obtain the autonomous driving simulation test results.

[0055] In some embodiments, when the server performs the step of performing an autonomous driving simulation test on multiple target simulation scenarios to obtain the autonomous driving simulation test results, the server performs a scene simulation evaluation on multiple target simulation scenarios according to a preset target prediction model to obtain multiple scene safety scores, multiple scene comfort scores, and multiple driver position difference values; the server determines the autonomous driving simulation test results according to the multiple scene safety scores, multiple scene comfort scores, and multiple driver position difference values. Further, the server records the autonomous driving simulation test results according to the scene test identifier. For example, if the scene safety score is 8 points, the scene comfort score is 9 points, and the position difference value between the target and the driver is 0.50 meters, the server determines that the scene safety level is high, the scene comfort level is high, and the driver's preset reaction time to the target is reasonable.

[0056] 204. Generate an autonomous driving simulation test report according to the autonomous driving simulation test results and a preset test report template.

[0057] Specifically, the server receives a request for generating an autonomous driving simulation test report; the server parses the request for generating an autonomous driving simulation test report to obtain a parsing result; the server performs parameter name and parameter value verification on the parsing result according to a preset parameter verification rule to obtain a verification result; when the verification result is that the verification passes, the server extracts a scenario test identifier and a report type from the parsing result; the server obtains a preset test report template according to the report type, and different report types correspond to different test report templates; the server extracts multiple replacement identifiers from the preset test report template; the server obtains the autonomous driving simulation test result according to the scenario test identifier, and searches for multiple target test data from the autonomous driving simulation test result according to the multiple replacement identifiers, and updates the multiple target test data to the preset test report template to obtain an autonomous driving simulation test report.

[0058] Furthermore, the server backs up the autonomous driving simulation test report to a preset file service storage system, and sends the autonomous driving simulation test report to the target terminal so that the target terminal can download or display the autonomous driving simulation test report.

[0059] It should be noted that the server also records the simulation test details data during the simulation test process; the server converts the autonomous driving simulation test results and simulation test details data into an autonomous driving simulation test report based on a preset test report template.

[0060] 205. Re-adjust the target test scene annotated dataset according to the autonomous driving simulation test report to obtain an updated test scene annotated dataset, and retest based on the updated test scene annotated dataset.

[0061] Specifically, the server obtains multiple statements to be parsed from the autonomous driving simulation test report; the server performs semantic parsing on the multiple statements to be parsed according to preset keywords to obtain parsing results corresponding to each statement to be parsed, and the server sets data labeling rules based on the parsing results corresponding to each statement to be parsed; the server readjusts the target test scene labeling data set according to the data labeling rules to obtain an updated test scene labeling data set, and retests based on the updated test scene labeling data set, that is, the server executes steps 203 to 204, thereby improving the accuracy of the autonomous driving simulation test and the accuracy of the generation of the autonomous driving simulation test report.

[0062] In an embodiment of the present invention, a target test scene labeled data set is obtained by preprocessing and labeling a vehicle simulation test scenario data set and a full data set of vehicle actual road tests, and an autonomous driving simulation test is performed based on the target test scene labeled data set to obtain an autonomous driving simulation test result, thereby improving the safety and stability of the autonomous driving scenario test and reducing the cost of the autonomous driving test.

[0063] The above describes the automatic driving simulation test method in the embodiment of the present invention. The following describes the automatic driving simulation test device in the embodiment of the present invention. Figure 3 , an embodiment of the automatic driving simulation test device in the embodiment of the present invention includes:

[0064] An acquisition module 301 is used to acquire a vehicle simulation test scenario data set and a vehicle actual road test full data set;

[0065] A processing module 302 is used to preprocess and annotate the vehicle simulation test scenario dataset and the vehicle actual road test full dataset to obtain a target test scenario annotated dataset, where the target test scenario annotated dataset is used to annotate test scenarios related to the target;

[0066] The testing module 303 is used to perform an autonomous driving simulation test based on the target test scenario labeled data set to obtain an autonomous driving simulation test result.

[0067] In an embodiment of the present invention, a target test scene labeled data set is obtained by preprocessing and labeling a vehicle simulation test scenario data set and a full data set of vehicle actual road tests, and an autonomous driving simulation test is performed based on the target test scene labeled data set to obtain an autonomous driving simulation test result, thereby improving the safety and stability of the autonomous driving scenario test and reducing the cost of the autonomous driving test.

[0068] See also Figure 4 Another embodiment of the automatic driving simulation test device in the embodiment of the present invention includes:

[0069] An acquisition module 301 is used to acquire a vehicle simulation test scenario data set and a vehicle actual road test full data set;

[0070] A processing module 302 is used to preprocess and annotate the vehicle simulation test scenario dataset and the vehicle actual road test full dataset to obtain a target test scenario annotated dataset, where the target test scenario annotated dataset is used to annotate test scenarios related to the target;

[0071] The testing module 303 is used to perform an autonomous driving simulation test based on the target test scenario labeled data set to obtain an autonomous driving simulation test result.

[0072] Optionally, the acquisition module 301 further includes:

[0073] A generating unit 3011 is used to obtain autonomous driving literature data and generate a vehicle simulation test scenario data set according to the autonomous driving literature data;

[0074] The processing unit 3012 is used to perform full data extraction, cleaning and clustering operations on the preset data source in sequence through the preset data mining tasks to obtain the full data set of the actual vehicle road test. The preset data source includes vehicle log information, vehicle sensor information and vehicle positioning information.

[0075] Optionally, the generating unit 3011 is specifically used for:

[0076] Reading autonomous driving literature data from a preset literature database, where the autonomous driving literature data is used to indicate descriptive information related to a target in an autonomous driving scenario;

[0077] Performing named entity recognition on the autonomous driving literature data according to preset rules to obtain an entity recognition data set, wherein the entity recognition data set includes a vehicle initial speed, a vehicle deceleration, and a driver reaction time;

[0078] A plurality of target passing moments are determined according to the entity recognition data set, and a vehicle simulation test scenario data set is generated based on the plurality of target passing moments.

[0079] Optionally, the processing module 302 is specifically used for:

[0080] Merging the vehicle simulation test scenario data set and the vehicle actual road test full data set to obtain a merged scenario data set;

[0081] Calling a preset data screening model to screen the merged scene data set to obtain an initial test scene data set;

[0082] The initial test scene data set is subjected to scene annotation processing by a preset data annotation model to obtain a target test scene annotation data set, and the target test scene annotation data set is stored in a preset scene database.

[0083] Optionally, the testing module 303 further includes:

[0084] A screening unit 3031 is used to screen scenes according to the target test scene annotation data set to obtain a screened scene data set, where the screened scene data set is used to indicate a real scene that interacts with the target;

[0085] A segmentation unit 3032 is used to segment the scene according to each interaction annotation start time and each interaction annotation end time in the filtered scene data set to obtain a plurality of simulation scene segment information;

[0086] A conversion unit 3033, configured to convert the plurality of simulation scene fragment information into a plurality of target simulation scenes through a preset simulation tool;

[0087] The testing unit 3034 is used to perform autonomous driving simulation tests on the multiple target simulation scenarios to obtain autonomous driving simulation test results.

[0088] Optionally, the testing unit 3034 is specifically used for:

[0089] Performing scenario simulation evaluation on the multiple target simulation scenarios according to a preset target prediction model to obtain multiple scenario safety scores, multiple scenario comfort scores, and multiple driver position difference values;

[0090] The autonomous driving simulation test result is determined according to the multiple scenario safety scores, the multiple scenario comfort scores and the multiple driver position difference values.

[0091] Optionally, the autonomous driving simulation test device further includes:

[0092] A generating module 304, configured to generate an autonomous driving simulation test report according to the autonomous driving simulation test result and a preset test report template;

[0093] The adjustment module 305 is used to readjust the target test scene annotation data set according to the autonomous driving simulation test report to obtain an updated test scene annotation data set, and retest based on the updated test scene annotation data set.

[0094] In an embodiment of the present invention, a target test scene labeled data set is obtained by preprocessing and labeling a vehicle simulation test scenario data set and a full data set of vehicle actual road tests, and an autonomous driving simulation test is performed based on the target test scene labeled data set to obtain an autonomous driving simulation test result, thereby improving the safety and stability of the autonomous driving scenario test and reducing the cost of the autonomous driving test.

[0095] above Figure 3 and Figure 4 The autonomous driving simulation test device in the embodiment of the present invention is described in detail from a modular perspective. The autonomous driving simulation test device in the embodiment of the present invention is described in detail from a hardware processing perspective.

[0096] Figure 5: is a structural schematic diagram of an automatic driving simulation test device provided by an embodiment of the present invention. The automatic driving simulation test device 500 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 510 (for example, one or more processors) and a memory 520, and one or more storage media 530 (for example, one or more mass storage devices) storing application programs 533 or data 532. Among them, the memory 520 and the storage medium 530 can be short-term storage or permanent storage. The program stored in the storage medium 530 may include one or more modules (not shown in the figure), and each module may include a series of computer program operations in the automatic driving simulation test device 500. Furthermore, the processor 510 can be configured to communicate with the storage medium 530 to execute a series of computer program operations in the storage medium 530 on the automatic driving simulation test device 500.

[0097] The autonomous driving simulation test device 500 may also include one or more power supplies 540, one or more wired or wireless network interfaces 550, one or more input and output interfaces 560, and / or one or more operating systems 531, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, etc. It will be appreciated by those skilled in the art that Figure 5 The structure of the autonomous driving simulation test equipment shown does not constitute a limitation of the autonomous driving simulation test equipment, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0098] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the autonomous driving simulation test method.

[0099] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0100] 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 is essentially 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, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a 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, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0101] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An automatic driving simulation test method, characterized in that: The automatic driving simulation test method comprises: Obtain vehicle simulation test scenario datasets and vehicle actual road test full datasets; Preprocessing and labeling the vehicle simulation test scenario dataset and the vehicle actual road test full dataset to obtain a target test scenario labeling dataset, wherein the target test scenario labeling dataset is used to label test scenarios related to the target; Performing an autonomous driving simulation test according to the target test scenario labeled data set to obtain an autonomous driving simulation test result; The method of obtaining a vehicle simulation test scenario data set includes: reading autonomous driving literature data from a preset literature database, wherein the autonomous driving literature data is used to indicate descriptive information related to a target in an autonomous driving scenario; performing named entity recognition on the autonomous driving literature data according to preset rules to obtain an entity recognition data set, wherein the entity recognition data set includes an initial vehicle speed, a vehicle deceleration, and a driver reaction time; and generating a vehicle simulation test scenario data set according to the entity recognition data set.

2. The automatic driving simulation test method according to claim 1, characterized in that: The step of obtaining a complete data set of actual vehicle road tests includes: The preset data mining tasks are used to extract, clean and cluster the preset data sources in turn to obtain the full data set of the actual road test of the vehicle. The preset data sources include vehicle log information, vehicle sensor information and vehicle positioning information.

3. The automatic driving simulation test method according to claim 1, characterized in that: The preprocessing and labeling of the vehicle simulation test scenario dataset and the vehicle actual road test full dataset to obtain the target test scenario labeling dataset includes: Merging the vehicle simulation test scenario data set and the vehicle actual road test full data set to obtain a merged scenario data set; Calling a preset data screening model to screen the merged scene data set to obtain an initial test scene data set; The initial test scene data set is subjected to scene annotation processing by a preset data annotation model to obtain a target test scene annotation data set, and the target test scene annotation data set is stored in a preset scene database.

4. The automatic driving simulation test method according to claim 1, characterized in that: The step of performing an autonomous driving simulation test according to the target test scenario labeled data set to obtain an autonomous driving simulation test result includes: Perform scene screening according to the target test scene annotation data set to obtain a screened scene data set, wherein the screened scene data set is used to indicate a real scene for interacting with the target; Perform scene segmentation according to each interaction annotation start time and each interaction annotation end time in the filtered scene data set to obtain multiple simulation scene segment information; Converting the plurality of simulation scene fragment information into a plurality of target simulation scenes through a preset simulation tool; An autonomous driving simulation test is performed on the multiple target simulation scenarios to obtain an autonomous driving simulation test result.

5. The automatic driving simulation test method according to claim 4, characterized in that: The performing of the autonomous driving simulation test on the multiple target simulation scenarios to obtain the autonomous driving simulation test results includes: Performing scenario simulation evaluation on the multiple target simulation scenarios according to a preset target prediction model to obtain multiple scenario safety scores, multiple scenario comfort scores, and multiple driver position difference values; The autonomous driving simulation test result is determined according to the multiple scenario safety scores, the multiple scenario comfort scores and the multiple driver position difference values.

6. The automatic driving simulation test method according to any one of claims 1 to 5, characterized in that: After performing the autonomous driving simulation test according to the target test scenario labeled data set and obtaining the autonomous driving simulation test result, the autonomous driving simulation test method further includes: Generate an autonomous driving simulation test report according to the autonomous driving simulation test results and the preset test report template; The target test scene annotated dataset is readjusted according to the autonomous driving simulation test report to obtain an updated test scene annotated dataset, and retesting is performed based on the updated test scene annotated dataset.

7. An automatic driving simulation test device, characterized in that: The automatic driving simulation test device comprises: An acquisition module is used to acquire a vehicle simulation test scenario data set and a vehicle actual road test full data set; A processing module, used for preprocessing and labeling the vehicle simulation test scenario data set and the vehicle actual road test full data set to obtain a target test scenario labeling data set, wherein the target test scenario labeling data set is used to label test scenarios related to the target; A testing module, used to perform an autonomous driving simulation test according to the target test scenario labeled data set to obtain an autonomous driving simulation test result; The acquisition module is specifically used to: read autonomous driving literature data from a preset literature database, wherein the autonomous driving literature data is used to indicate descriptive information related to a target in an autonomous driving scenario; perform named entity recognition on the autonomous driving literature data according to preset rules to obtain an entity recognition data set, wherein the entity recognition data set includes a vehicle initial speed, a vehicle deceleration, and a driver reaction time; and generate a vehicle simulation test scenario data set based on the entity recognition data set.

8. An automatic driving simulation test device, characterized in that: The autonomous driving simulation test device comprises: a memory and at least one processor, wherein the memory stores a computer program; The at least one processor calls the computer program in the memory so that the autonomous driving simulation test device executes the autonomous driving simulation test method as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the autonomous driving simulation test method as described in any one of claims 1 to 6 is implemented.

Images