Method and system for updating autonomous driving test scenario set based on real data

Abstract

The present invention relates to the field of autonomous driving technology, and provides a method and system for updating an autonomous driving test scenario set based on real data. The method includes: according to a first condition, dividing the real road operation data of autonomous driving into complex scene data segments and non-complex scene data segments; extracting the design operation domain ODD elements and dynamic parameter groups in the complex scene data segments, and generating iterative test scenarios according to the design operation domain ODD elements and dynamic parameter groups in the complex scene data segments; updating the parameter space of the logical scene in the existing test scenario set based on the non-complex scene data segments; the present invention introduces a scenario access rule through the first condition, extracts complex scene data segments from the real road operation data, can quickly locate the data segments with poor processing effect of the autonomous driving model, and updates the test scenario set on this basis, providing a more efficient test scenario foundation for iterative updates of the autonomous driving model.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to a method and system for updating an autonomous driving test scenario set based on real data. Background Art

[0002] Autonomous driving, also known as unmanned driving, computer driving or wheeled mobile robot, is a cutting-edge technology that relies on computers and artificial intelligence technology to complete complete, safe and effective driving without human control.

[0003] However, autonomous driving also faces many challenges, such as bad weather, complex traffic conditions, etc., so its performance in some scenarios may not match that of human drivers. To solve this problem, more testing and iterative updates of the autonomous driving model are needed to improve its processing capabilities in complex scenarios.

[0004] The above complex scenarios have a generally accepted long-tail distribution hypothesis, that is, 80% of the situations are common, and 20% are rare. Even if the autonomous driving system has mastered the common situations, rare situations will continue to appear, making it difficult for the autonomous driving model to improve its capabilities through learning and iteration.

[0005] At the same time, autonomous driving scenarios are infinitely rich, extremely complex, unpredictable, and inexhaustible. The huge number of scenario test sets places high demands on the autonomous driving test cycle, efficiency, and cost, which is not conducive to the rapid iteration and upgrade of autonomous driving software.

[0006] Therefore, how to provide a more effective method and system for updating the autonomous driving test scenario set to solve the problem that the existing autonomous driving models have insufficient learning and iteration capabilities for complex scenarios with long-tail distribution and low efficiency in updating test scenarios has become a technical problem that needs to be urgently solved in the industry. Summary of the invention

[0007] The present invention provides a method and system for updating an autonomous driving test scenario set based on real data, so as to solve the defects in the prior art that the autonomous driving model has insufficient iterative learning ability for complex scenarios with long-tail distribution and low updating efficiency of test scenarios, and realize more efficient updating of autonomous driving test scenarios.

[0008] The present invention provides a method for updating an autonomous driving test scenario set based on real data, comprising:

[0009] According to a first condition, the real road operation data of the autonomous driving is divided into a complex scene data segment and a non-complex scene data segment; the first condition is a complex criterion determined according to the autonomous driving capability boundary and / or the autonomous driving safety boundary;

[0010] Extracting the design operation domain (ODD) elements and the dynamic parameter group in the complex scenario data fragment, and generating an iterative test scenario according to the design operation domain (ODD) elements and the dynamic parameter group in the complex scenario data fragment;

[0011] Based on the non-complex scene data fragment, the parameter space of the logical scene in the existing test scene set is updated; the logical scene includes a static road network and a dynamic target; the parameter space refers to the physical parameter range and / or physical parameter value set in the logical scene.

[0012] According to the method for updating an autonomous driving test scenario set based on real data provided by the present invention, after the step of dividing the real road operation data of the autonomous driving into complex scenario data segments and non-complex scenario data segments according to the first condition, the method further includes:

[0013] Labeling the complex scene data segments corresponding to the data IDs one by one to obtain one or more first labels;

[0014] The first tag includes driving actions, design operation domain (ODD) elements or dynamic parameters in the complex scene data segment.

[0015] According to a method for updating an autonomous driving test scenario set based on real data provided by the present invention, the first label further includes any one or any combination of the following:

[0016] The driving action, design operation domain (ODD) element or dynamic parameter in the complex scene data segment corresponds to the functional unit in the automatic driving system;

[0017] Standard label, which refers to a label that meets the ISO 34504 standard;

[0018] Custom labels.

[0019] According to a method for updating an autonomous driving test scenario set based on real data provided by the present invention, the step of extracting the design operation domain ODD element and the dynamic parameter group in the complex scenario data segment comprises:

[0020] Locate and segment the complex scene data segment in the real road operation data according to the data ID;

[0021] The complex scene data segments obtained by parsing and segmenting are extracted to extract the design operation domain ODD elements and dynamic parameter groups.

[0022] According to the method for updating an autonomous driving test scenario set based on real data provided by the present invention, after the step of extracting the design operation domain ODD elements and dynamic parameter groups in the complex scenario data fragments, and generating iterative test scenarios according to the design operation domain ODD elements and dynamic parameter groups in the complex scenario data fragments, the method further includes:

[0023] Labeling the iterative test scenario to obtain one or more second labels;

[0024] The second tag includes a driving action, a design operation domain (ODD) element or a dynamic parameter in the iterative test scenario determined according to the first tag.

[0025] According to a method for updating an autonomous driving test scenario set based on real data provided by the present invention, the second tag further includes any one or any combination of the following:

[0026] The driving actions, design operation domain (ODD) elements or dynamic parameters in the iterative test scenario correspond to functional units in the autonomous driving system;

[0027] Standard label, which refers to a label that meets the ISO 34504 standard;

[0028] Custom labels;

[0029] The version number of the automated driving system and / or test scenario generation software;

[0030] Test scenario types, including real vehicle tests, simulation tests, problem scenarios, regulatory scenarios, danger scenarios, accident scenarios, and perception scenarios.

[0031] According to a method for updating an autonomous driving test scenario set based on real data provided by the present invention, the step of updating the parameter space of the logical scene in the existing test scenario set based on the non-complex scene data fragment comprises:

[0032] Label the non-complex scene data segments corresponding to the data IDs one by one to obtain one or more third tags; the third tags include driving actions, design operation domain (ODD) elements or dynamic parameters in the non-complex scene data segments;

[0033] Clustering the non-complex scene data segments according to the third label to obtain one or more clustering results;

[0034] Based on the clustering results, the scenario distribution probability, the design operation domain ODD element distribution and the dynamic parameter space are calculated to obtain statistical results;

[0035] The parameter space of the logical scenarios in the existing test scenario set is updated according to the statistical results, and the parameter space is filtered through a scenario generation tool to generate a specific scenario; the specific scenario refers to a logical scenario including specific parameter values.

[0036] According to a method for updating an autonomous driving test scenario set based on real data provided by the present invention, the third tag further includes any one or any combination of the following:

[0037] The driving action, design operation domain (ODD) element or dynamic parameter in the non-complex scene data segment corresponds to a functional unit in the automatic driving system;

[0038] Standard label, which refers to a label that meets the ISO 34504 standard;

[0039] Custom labels.

[0040] According to the method for updating an autonomous driving test scenario set based on real data provided by the present invention, after the step of dividing the real road operation data of the autonomous driving into complex scenario data segments and non-complex scenario data segments according to the first condition, the method further includes:

[0041] According to the second condition, the complex scene data fragments are screened to obtain uncovered fragments and covered fragments, and the uncovered fragments are used as corrected complex scene data fragments;

[0042] The second condition is a criterion for determining whether the complex scene data segment is covered by the existing test scene set based on the existing test scene set; the uncovered segment is a complex scene data segment that is not covered by the existing test scene set; and the covered segment is a complex scene data segment that is covered by the existing test scene set.

[0043] According to a method for updating an autonomous driving test scene set based on real data provided by the present invention, the step of screening the complex scene data segments according to the second condition to obtain uncovered segments and covered segments comprises:

[0044] Obtain an existing information set of an existing test scenario in an existing test scenario set;

[0045] According to the first tag, a judgment is performed on the complex scene data fragment:

[0046] If the first tag of the complex scene data segment does not belong to the existing information set, determining that the complex scene data segment is an uncovered segment;

[0047] If the first tag of the complex scene data segment belongs to an existing information set, the complex scene data segment is determined to be a covered segment.

[0048] According to the present invention, a method for updating an autonomous driving test scenario set based on real data also includes:

[0049] Based on the iterative test scenario, software-in-the-loop testing and / or hardware-in-the-loop testing is performed on the autonomous driving model, and the autonomous driving model is updated according to the results of the software-in-the-loop testing and / or hardware-in-the-loop testing.

[0050] According to a method for updating an autonomous driving test scenario set based on real data provided by the present invention, after the step of performing a software-in-the-loop test and / or a hardware-in-the-loop test on the autonomous driving model based on the iterative test scenario, and updating the autonomous driving model according to the results of the software-in-the-loop test and / or the hardware-in-the-loop test, the method further includes:

[0051] Based on the iterative test scenario, a closed-scene real-vehicle test is performed on the autonomous driving model, and the autonomous driving model is updated according to the closed-scene real-vehicle test results.

[0052] According to a method for updating an autonomous driving test scenario set based on real data provided by the present invention, after the step of performing a closed-scene real-vehicle test on the autonomous driving model based on the iterative test scenario and updating the autonomous driving model according to the closed-scene real-vehicle test result, the method further includes:

[0053] Based on the iterative test scenario, an open scenario reliability test and / or a driver-in-the-loop test is performed on the autonomous driving model, and the autonomous driving model is updated according to the results of the open scenario reliability test and / or the driver-in-the-loop test.

[0054] According to a method for updating an autonomous driving test scenario set based on real data provided by the present invention, after the step of performing a software-in-the-loop test and / or a hardware-in-the-loop test on the autonomous driving model based on the iterative test scenario, and updating the autonomous driving model according to the results of the software-in-the-loop test and / or the hardware-in-the-loop test, the method further includes:

[0055] The first condition and / or the second condition are adjusted according to the software-in-the-loop test and / or the hardware-in-the-loop test results.

[0056] According to a method for updating an autonomous driving test scenario set based on real data provided by the present invention, after the step of performing a closed-scene real-vehicle test on the autonomous driving model based on the iterative test scenario and updating the autonomous driving model according to the closed-scene real-vehicle test result, the method further includes:

[0057] Adjust the first condition and / or the second condition according to the closed scene actual vehicle test results.

[0058] According to a method for updating an autonomous driving test scenario set based on real data provided by the present invention, after the step of performing an open scenario reliability test and / or a driver-in-the-loop test on the autonomous driving model based on the iterative test scenario, and updating the autonomous driving model according to the open scenario reliability test and / or the driver-in-the-loop test results, the method further comprises:

[0059] The first condition and / or the second condition are adjusted according to the open scenario reliability test and / or the driver-in-the-loop test results.

[0060] The present invention also provides an automatic driving test scenario set updating system based on real data, comprising:

[0061] a complex judgment module, configured to divide the real road operation data of the autonomous driving into complex scene data segments and non-complex scene data segments according to a first condition; wherein the first condition is a complex criterion determined according to the autonomous driving capability boundary and / or the autonomous driving safety boundary;

[0062] An iterative test scenario module, used to extract the design operation domain ODD elements and dynamic parameter groups in the complex scenario data fragment, and generate an iterative test scenario according to the design operation domain ODD elements and dynamic parameter groups in the complex scenario data fragment;

[0063] A parameter space updating module is used to update the parameter space of the logical scene in the existing test scene set based on the non-complex scene data fragment; the logical scene includes a static road network and a dynamic target; the parameter space refers to the physical parameter range and / or physical parameter value set in the logical scene.

[0064] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the method for updating an autonomous driving test scenario set based on real data as described in any one of the above-mentioned methods are implemented.

[0065] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method for updating an autonomous driving test scenario set based on real data as described in any one of the above are implemented.

[0066] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of any of the above-mentioned methods for updating an autonomous driving test scenario set based on real data.

[0067] The method and system for updating the autonomous driving test scenario set based on real data provided by the present invention introduce scene access rules through the first condition, extract complex scene data fragments from real road operation data, can quickly locate data fragments that are not well processed by the autonomous driving model, and update the test scenario set on this basis, providing a more efficient test scenario foundation for the iterative update of the autonomous driving model. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0069] Figure 1 It is a flowchart of a method for updating an autonomous driving test scenario set based on real data provided by the present invention;

[0070] Figure 2 is a schematic diagram of a scene admission rule provided by an embodiment of the present invention;

[0071] Figure 3 It is a structural schematic diagram of a system for updating an autonomous driving test scenario set based on real data provided by the present invention;

[0072] Figure 4 is a schematic diagram of the structure of an electronic device provided by the present invention;

[0073] Figure 5 It is a schematic diagram of a link flow provided by an embodiment of the present invention.

[0074] Reference numerals:

[0075] 1: Complex judgment module; 2: Iterative test scenario module; 3: Parameter space update module;

[0076] 410: processor; 420: communication interface; 430: memory;

[0077] 440: Communication bus. DETAILED DESCRIPTION

[0078] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings 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 ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0079] Combine the following Figure 1 , Figure 2 , Figure 5 The present invention describes a method for updating an autonomous driving test scenario set based on real data.

[0080] like Figure 1 As shown, an embodiment of the present invention provides a method for updating an autonomous driving test scenario set based on real data, comprising:

[0081] Step 102: dividing the real road operation data of the autonomous driving into a complex scene data segment and a non-complex scene data segment according to a first condition; the first condition is a complex criterion determined according to an autonomous driving capability boundary and / or an autonomous driving safety boundary;

[0082] Step 104, extracting the design operation domain ODD elements and dynamic parameter groups in the complex scenario data fragment, and generating an iterative test scenario according to the design operation domain ODD elements and dynamic parameter groups in the complex scenario data fragment;

[0083] Step 106, updating the parameter space of the logical scene in the existing test scene set based on the non-complex scene data fragment; the logical scene includes a static road network and a dynamic target; the parameter space refers to the physical parameter range and / or physical parameter value set in the logical scene.

[0084] The following will explain and illustrate the keywords that may appear in this embodiment and subsequent embodiments.

[0085] 1) ODD: operating design domain, design operation domain;

[0086] 2)AD: autonomous driving;

[0087] 3) MPD XX: Miles Per Disengage XX, which means the average number of times the autonomous driving system needs to be disengaged is once every XX km.

[0088] In this embodiment, the complex scene data fragment is determined according to the first condition, i.e., the scene access rule, such as the scenario in which the driver actively or passively intervenes during the operation of autonomous driving (autonomous driving at all levels from L0 to L5); or the scenario in which the vehicle's motion state (velocity vector, acceleration vector, front direction, vehicle position, etc.) is abnormal during the operation of autonomous driving.

[0089] In a preferred embodiment, the scene admission rule can be set based on the driver's takeover action, that is, all driver takeover scenes are regarded as complex scenes.

[0090] In a preferred embodiment, Figure 2 As shown, the scene access rules include AD capability boundary rules and safety rules, which are determined according to the autonomous driving capability boundary and the autonomous driving safety boundary respectively.

[0091] The AD capability boundary rules include:

[0092] N-type and / or NA-type ODD element (i.e., preset special ODD element) scenarios, such as wind, rain, fog, snow, small animals, debris, motorcycles, bicycles, etc. Among them, the N-type ODD element scenario is an element scenario that the AD system can identify and fallback (exit), that is, the AD system exits and the driver takes over; the NA-type ODD element scenario is an element scenario that the AD system cannot identify. (Takeover level MPD 3 / 5)

[0093] AD system failure scenarios, such as fallback caused by AD system module failure, line control system failure, etc. (takeover level MPD 2)

[0094] AD function defect scenarios, such as incorrect path planning.

[0095] The safety rules include:

[0096] Dangerous scenarios, such as vehicle dynamic parameter overshoot: CC (cruise control) lateral acc (acceleration) greater than 0.5mpss (m / s 2 ), ILC (lever lane change) lateral acc greater than 1mpss, AD longitudinal acc less than -5mpss, TTC (time to collision) less than 5s, etc. (takeover level MPD 0 / 1)

[0097] Accident scenarios, such as parallel overtaking, rear-end collision, etc.

[0098] Special behavior scenarios of the target vehicle, such as reversing, driving in the wrong direction, merging into the ramp at low speed, parking in the overtaking lane, driving at low speed in the overtaking lane, etc.

[0099] Driver violation scenarios, such as the driver not being in the driving position, etc. (MPD 6)

[0100] The takeover level MPD in the scene access rules is defined as follows:

[0101] MPD 0: Unexpected, there is a risk of collision;

[0102] MPD 1: Unexpected, no collision risk;

[0103] MPD 2: Fallback caused by software or hardware problems;

[0104] MPD 3: Fallback takeover caused by unprocessed ODD;

[0105] MPD 4: fallback takeover caused by driver’s violation;

[0106] MPD 5: Out of ODD range, ODD = NA, the system cannot recognize;

[0107] MPD 6: Takeover due to the driver’s own reasons.

[0108] The design operation domain ODD elements in this embodiment, such as straight roads, slopes, curves, daytime, nighttime, wind, rain, fog, highways, etc., also correspond to physical parameter spaces, such as the slope of the slope, the curvature of the curve, the light intensity during the day, the visibility in the fog, etc.

[0109] The beneficial effects of this embodiment are:

[0110] By introducing scene access rules through the first condition and extracting complex scene data fragments from real road operation data, it is possible to quickly locate data fragments that are not well processed by the autonomous driving model, and update the test scene set on this basis, providing a more efficient test scenario foundation for the iterative update of the autonomous driving model.

[0111] According to the above embodiment, in this embodiment:

[0112] The step of extracting the design operation domain ODD element and the dynamic parameter group in the complex scene data segment comprises:

[0113] Locate and segment the complex scene data segment in the real road operation data according to the data ID;

[0114] The complex scene data segments obtained by parsing and segmenting are extracted to extract the design operation domain ODD elements and dynamic parameter groups.

[0115] After the step of dividing the real road operation data of the autonomous driving into complex scene data segments and non-complex scene data segments according to the first condition, the step further includes:

[0116] Labeling the complex scene data segments corresponding to the data IDs one by one to obtain one or more first labels;

[0117] The first tag includes driving actions, design operation domain (ODD) elements or dynamic parameters in the complex scene data segment.

[0118] The purpose of the first label is to provide a label for the reason why the complex scene data segment exceeds the autonomous driving capability boundary and / or the autonomous driving safety boundary. For example, if sudden braking causes the longitudinal negative acceleration to overshoot (i.e., the longitudinal negative acceleration exceeds a given range, which may be determined based on the autonomous driving capability boundary and / or the autonomous driving safety boundary), there are two first labels, namely sudden braking (driving action), longitudinal negative acceleration and its value (dynamic parameter).

[0119] The first tag also includes any one or any combination of the following:

[0120] The driving action, design operation domain (ODD) element or dynamic parameter in the complex scene data segment corresponds to the functional unit in the automatic driving system;

[0121] Standard label, which refers to a label that meets the ISO 34504 standard;

[0122] Custom labels.

[0123] It is worth noting that the first label may be marked after step 102 is executed or during step 102 is executed; the marked subject of the first label may be the execution subject of step 102 and / or the execution subject of step 104 .

[0124] After the step of extracting the design operation domain ODD elements and dynamic parameter groups in the complex scenario data fragment, and generating an iterative test scenario according to the design operation domain ODD elements and dynamic parameter groups in the complex scenario data fragment, the method further includes:

[0125] Labeling the iterative test scenario to obtain one or more second labels;

[0126] The second tag includes a driving action, a design operation domain (ODD) element or a dynamic parameter in the iterative test scenario determined according to the first tag.

[0127] The second tag also includes any one or any combination of the following:

[0128] The driving actions, design operation domain (ODD) elements or dynamic parameters in the iterative test scenario correspond to functional units in the autonomous driving system;

[0129] Standard label, which refers to a label that meets the ISO 34504 standard;

[0130] Custom labels;

[0131] The version number of the automated driving system and / or test scenario generation software;

[0132] Test scenario types, including real vehicle tests, simulation tests, problem scenarios, regulatory scenarios, danger scenarios, accident scenarios, and perception scenarios.

[0133] It is worth noting that the second label can be marked after the iterative test scenario is generated, when the iterative test scenario is generated, or before the iterative test scenario is generated (that is, after determining the static design operation domain ODD elements and dynamic parameter groups).

[0134] The step of updating the parameter space of the logical scene in the existing test scene set based on the non-complex scene data fragment comprises:

[0135] Label the non-complex scene data segments corresponding to the data IDs one by one to obtain one or more third tags; the third tags include driving actions, design operation domain (ODD) elements or dynamic parameters in the non-complex scene data segments;

[0136] Clustering the non-complex scene data segments according to the third label to obtain one or more clustering results;

[0137] Based on the clustering results, the scenario distribution probability, the design operation domain ODD element distribution and the dynamic parameter space are calculated to obtain statistical results;

[0138] The parameter space of the logical scenarios in the existing test scenario set is updated according to the statistical results, and the parameter space is filtered through a scenario generation tool to generate a specific scenario; the specific scenario refers to a logical scenario including specific parameter values.

[0139] The third tag also includes any one or any combination of the following:

[0140] The driving action, design operation domain (ODD) element or dynamic parameter in the non-complex scene data segment corresponds to a functional unit in the automatic driving system;

[0141] Standard label, which refers to a label that meets the ISO 34504 standard;

[0142] Custom labels.

[0143] Similar to the first label, one of the purposes of the second label and the third label is to provide a label for the reason why the iterative test scenario and the non-complex scenario data segment exceed the autonomous driving capability boundary and / or the autonomous driving safety boundary.

[0144] In addition, the second tag can also be used as a keyword to retrieve the test scene during the test process; the third tag can also be used for clustering.

[0145] In a preferred embodiment, the first tag, the second tag and the third tag all meet the ISO34504 standard. This preferred embodiment can improve the versatility of the test scenario set obtained in this embodiment.

[0146] This embodiment introduces a first label for complex scene data fragments and a second label for iterative test scenarios; the introduction of the first label can, on the one hand, effectively improve the execution efficiency of step 104 (i.e., screening for coverage / non-coverage through the first label), and on the other hand, can provide a labeling basis for the second label of the subsequently generated iterative test scenario; further, by introducing the second label, the search efficiency of the test scene during the testing process of the autonomous driving model can be improved, thereby improving the efficiency and pertinence of the autonomous driving test.

[0147] According to any of the above embodiments, in this embodiment:

[0148] After the step of dividing the real road operation data of the autonomous driving into complex scene data segments and non-complex scene data segments according to the first condition, the step further includes:

[0149] According to the second condition, the complex scene data fragments are screened to obtain uncovered fragments and covered fragments, and the uncovered fragments are used as corrected complex scene data fragments;

[0150] The second condition is a criterion for determining whether the complex scene data segment is covered by the existing test scene set based on the existing test scene set; the uncovered segment is a complex scene data segment that is not covered by the existing test scene set; and the covered segment is a complex scene data segment that is covered by the existing test scene set.

[0151] The step of screening the complex scene data segments according to the second condition to obtain uncovered segments and covered segments comprises:

[0152] Obtain an existing information set of an existing test scenario in an existing test scenario set;

[0153] According to the first tag, a judgment is performed on the complex scene data fragment:

[0154] If the first tag of the complex scene data segment does not belong to the existing information set, determining that the complex scene data segment is an uncovered segment;

[0155] If the first tag of the complex scene data segment belongs to an existing information set, the complex scene data segment is determined to be a covered segment.

[0156] Based on the above embodiment, this embodiment defines a second execution link different from the generated iterative test scenario (referred to as the first execution link). The first execution link and the second execution link will be described in detail below.

[0157] For the real road operation data of the autonomous driving, the data is divided into a complex scene data segment and a non-complex scene data segment based on the first condition, and the non-complex scene data segment is used as the input of the second execution link;

[0158] For the complex scene data fragment, further filter it into uncovered fragments and covered fragments based on the second condition, and use the uncovered fragments (i.e., the updated complex scene data fragments) as inputs of the first execution link, and the covered fragments as inputs of the second execution link;

[0159] In this embodiment, every fragment of the real road operation data of autonomous driving can be effectively utilized.

[0160] In a preferred embodiment, Figure 5 As shown in the figure, the real road operation data of autonomous driving is processed to obtain the operation moment data as the basic input, and then the scene recognition is performed through the scene access definition (i.e. the first condition):

[0161] For data segments that meet the definition of complex scenes, i.e., complex scene data segments, complex scene annotation is performed, and then based on the evaluation rules, a secondary judgment is performed through the tool chain data analysis tool to screen out non-complex scenes missed in scene identification. After the secondary judgment, scene extraction is performed through the scene generation tool (i.e., step 104), and iterative test scenes are generated and added to the scene library for testing by the test platform;

[0162] For data segments that do not meet the definition of complex scenarios and data segments that are excluded (passed) in the secondary judgment, that is, non-complex data segments, based on tag clustering, scene statistics, and scene screening, combined with the test results of the test platform, the parameter space of the logical scenes in the scenario library is updated through the scenario generation tool.

[0163] In a preferred embodiment, the existing test scenario set includes three levels of test scenarios, namely functional scenarios, logical scenarios, and specific scenarios; the functional scenarios, such as the static road network is a highway, a two-lane road, a curve, etc., and the dynamic targets include the main vehicle (left lane), the target vehicle, the following vehicle, etc., including at least one logical scenario; the logical scenarios, such as the left lane width is 2.5-3.75m, the right lane width is 2.5-3.75m, the curve curvature is 300-900m, the main vehicle position coordinates are 0-100m, the front vehicle position coordinates are 0-110m, and the front vehicle position coordinates are greater than the main vehicle position coordinates, include at least one specific scenario (such as the left lane width is 3m, the right lane width is 3m, the curve curvature is 500m, the main vehicle position coordinates are 80m, the front vehicle position coordinates are 100m, and the front vehicle position coordinates are greater than the main vehicle position coordinates).

[0164] The purpose of the first execution link is to generate a new logical scenario (i.e., iterative test scenario) that is different from the logical scenarios existing in the existing test scenario set through the input of the uncovered fragment, thereby introducing a new test scenario found in the real road operation data of autonomous driving.

[0165] The purpose of the second execution link is to update the logical scenarios existing in the existing test scenario set through the input of the non-complex scenario data fragment and / or the coverage fragment, thereby obtaining new specific scenarios and further enriching the parameter values ​​of the existing logical scenarios.

[0166] In other words, the first execution link and the second execution link can respectively update and improve the existing test scenario sets, thereby making subsequent autonomous driving model tests more targeted and achieving better autonomous driving model iteration efficiency.

[0167] According to any of the above embodiments, the method of this embodiment further includes:

[0168] Based on the iterative test scenario, software-in-the-loop testing and / or hardware-in-the-loop testing is performed on the autonomous driving model, and the autonomous driving model is updated according to the results of the software-in-the-loop testing and / or hardware-in-the-loop testing.

[0169] After the step of performing software-in-the-loop testing and / or hardware-in-the-loop testing on the autonomous driving model based on the iterative test scenario, and updating the autonomous driving model according to the results of the software-in-the-loop testing and / or hardware-in-the-loop testing, the method further includes:

[0170] Based on the iterative test scenario, a closed-scene real-vehicle test is performed on the autonomous driving model, and the autonomous driving model is updated according to the closed-scene real-vehicle test results.

[0171] After the step of performing a closed-scene real-vehicle test on the autonomous driving model based on the iterative test scenario, and updating the autonomous driving model according to the closed-scene real-vehicle test result, the method further includes:

[0172] Based on the iterative test scenario, an open scenario reliability test and / or a driver-in-the-loop test is performed on the autonomous driving model, and the autonomous driving model is updated according to the results of the open scenario reliability test and / or the driver-in-the-loop test.

[0173] After the step of performing software-in-the-loop testing and / or hardware-in-the-loop testing on the autonomous driving model based on the iterative test scenario, and updating the autonomous driving model according to the results of the software-in-the-loop testing and / or hardware-in-the-loop testing, the method further includes:

[0174] The first condition and / or the second condition are adjusted according to the software-in-the-loop test and / or the hardware-in-the-loop test results.

[0175] After the step of performing a closed-scene real-vehicle test on the autonomous driving model based on the iterative test scenario, and updating the autonomous driving model according to the closed-scene real-vehicle test result, the method further includes:

[0176] Adjust the first condition and / or the second condition according to the closed scene actual vehicle test results.

[0177] After the step of performing an open scenario reliability test and / or a driver-in-the-loop test on the autonomous driving model based on the iterative test scenario, and updating the autonomous driving model according to the open scenario reliability test and / or the driver-in-the-loop test results, the method further includes:

[0178] The first condition and / or the second condition are adjusted according to the open scenario reliability test and / or the driver-in-the-loop test results.

[0179] This embodiment sets up a test process and gradually iterates the autonomous driving model, which can make the risks of personnel, equipment, and software in the defect scenario testing process more controllable, thereby achieving a safer test-iteration-update process.

[0180] Furthermore, this embodiment revises and adjusts the first condition and / or the second condition through the test results (such as the test pass rate) during the iteration process of the autonomous driving model version to form a feedback adjustment mechanism. As the autonomous driving model version is iterated and updated, this mechanism can bring higher scene set update efficiency.

[0181] In a preferred embodiment, for the iterative test scenario of a specific second label, continuous attention should be paid to the test pass rate of this scenario in several versions after the iterative update of the autonomous driving model (in a preferred scheme, the number and proportion of tests for this scenario can also be increased as appropriate) to ensure that the updated iterative autonomous driving model can maintain the ability to resolve such defects.

[0182] According to any of the above embodiments, in this embodiment:

[0183] This embodiment further solves the following problems based on the solution of scene generation based on real data (such as generating logical scenes after performing feature extraction, data annotation, scene clustering and statistical analysis on real data, and generating specific scenes on a large scale by performing different permutations and combinations and traversal values ​​through static element association relationships and dynamic parameter distribution; for example, based on massive road sampling data, constructing a logical scene parameter space statistically, analyzing the correlation of scene parameters, and then evaluating the scene KPI to generate specific scenes through random sampling):

[0184] 1. Scenario coverage: Currently, there are two main inputs for scenario library development: the autonomous driving system function definition document (FD) and the autonomous driving system operation area (ODD) definition. By combing and analyzing the functional design documents, analyzing the complexity of static and dynamic scenarios, and reasonably expanding the ODD element combination and design parameter range, a real vehicle and simulation scenario test set with high coverage is generated. However, the specific scenarios after theoretical expansion cannot fully map / cover the complex scenarios encountered on real roads. Therefore, through real data input, extreme scenarios, dangerous scenarios, AD capability boundary scenarios, AD system failures, AD functional logic defects and other complex scenarios in the data are excavated, and the scenarios are counted and clustered through the test management platform. Then, through reasonable parameters and ODD element combinations, more reasonable complex scenarios are reconstructed to improve the coverage of the scenarios.

[0185] 2. Reasonable and efficient scenarios: Define reasonable scenario screening rules, and use scenario generation tools to automatically screen out scenario test sets that are effective in combination, controllable in quantity, and oriented to different test requirements. For example, module algorithm iteration scenario test sets, perception system performance scenario test sets, phased vehicle performance verification scenario test sets, key problem scenario test sets, etc. Therefore, through real data analysis and calculation, the probability distribution of typical scenarios, parameter probability distribution, and ODD element combination probability distribution can be obtained to correct scenario screening rules. At the same time, complex scenarios are labeled for scenario clustering and convenient screening of different test scenario sets.

[0186] Specifically, the method provided in this embodiment includes 5 stages:

[0187] Phase 1: Define scenario access rules, that is, which road test scenarios we need to focus on, from the perspectives of AD capability boundaries, security, and user experience.

[0188] Phase 2: Organize the scene types that need attention into a requirements document and mark them during open road testing. The marking information needs to concisely describe the specific scene and label the scene with an applicable label. At the same time, classify complex scenes in the test report and indicate the complex scene data ID number.

[0189] Phase 3: Use the test report scene data ID to perform data analysis on the test data playback platform. First, determine whether the marked scene is consistent with the scene access definition, whether the scene label is correct, and whether the scene data is correctly linked to the complex scene database. If the above operations are incorrect, they need to be corrected. Secondly, determine whether the complex scene data is covered by the current scene library and real vehicle simulation test set.

[0190] Phase 4: Extract static elements / dynamic parameters of complex scenarios that are not covered by the scenario library or scenario test set, input them into the scenario generation tool, and generate specific scenario sets for different testing requirements based on the scenario screening rules and tags defined in the scenario generation tool.

[0191] The fifth stage: The generated scenario test set is subjected to real vehicle and simulation tests, and the scenario access rules and scenario screening rules are further revised through the test result feedback, so that the scenario library construction forms a virtuous circle.

[0192] More specifically, the following example will be used to illustrate a complex scenario of data extraction for an autonomous driving trip of 1,700 km from Shanghai to Jinan.

[0193] Phase 1: Scenario access definition.

[0194] First, we define the key scenes in the open road. The scene type definitions that can be included in the scene library can be referred to Figure 2 The content shown.

[0195] Phase 2: Obtaining complex scene data fragments.

[0196] Based on the above scenario access definition and labeling instructions, data labeling was carried out on the test platform during the AD road test in Shanghai and Jinan. During the entire test process, 156 complex scenarios were labeled, and 31 scenario categories were formed through classification.

[0197] Phase 3: Analyze scenario data.

[0198] We mapped 31 types of complex road test scenarios with the current scenario library and real vehicle and simulation scenario test sets, and found that 29 types of complex scenarios have been covered in the test, while the remaining 12 types are not covered by the current scenario set and need to be extracted.

[0199] Stage 4: scene extraction and generation.

[0200] Through the data playback platform, extract which function the scene belongs to, the static ODD elements of the scene, the initial position of the main vehicle and the target vehicle in the scene, speed parameters, action type, action duration and other parameters, and fill all the elements of the scene generation into the scene generation tool in the format. In addition, in order to facilitate scene classification and screening of different test sets, it is necessary to label the scene in the scene generation tool.

[0201] Phase 5: Scenario application.

[0202] The specific scenarios generated can be converted into different formats and tested on different test platforms. For example, they can be applied to SIL (software in the loop) testing, HIL (hardware in the loop) testing, closed-scene real-car testing, and DIL (driver in the loop) testing. Filtering out test sets through scenario screening rules can meet different testing requirements. For example, the problem scenarios in the current version can first be applied to the SIL / HIL simulation test of the next software version. After passing, the problem scenarios are added to the closed-scene test set for closed-scene real-car testing, and then released to open roads for reliability testing after passing. This test process is usually completed within 1-2 days, which greatly accelerates the iteration and upgrade of autonomous driving software, while also ensuring the safety of open road testing to a great extent.

[0203] The test result pass rate and the scene distribution probability calculated by the test platform will be used to modify the scene access rules and scene screening rules, which will not be elaborated here.

[0204] The beneficial effects of this embodiment are:

[0205] Based on the input of real data from the development roads of autonomous driving, by defining a set of scene access rules, combined with test platform data recording, automatic labeling, and data visualization playback functions, complex scene data fragments can be quickly located in the massive autonomous driving data. After analysis and review, the scene elements are extracted and filled into the scene generation tool to quickly generate a scene set that can be used for real vehicle / simulation testing. In addition, the test platform can automatically calculate the distribution of typical scenes for correcting the scene screening rules based on theory. The scene library obtained in this embodiment can effectively cover the complex scenes of real roads, and the generated specific scenes are screened out through a set of scene screening rules. The combination is reasonable, the number is controllable, and the scene test set that can support different testing requirements of autonomous driving is selected. It further supports the needs of rapid iteration of software development, and the construction and application of the scene library form a positive cycle.

[0206] That is to say, in this embodiment:

[0207] 1. Improve the coverage of the scenario library generated based on theory, so that the construction and application of the scenario library form a positive cycle;

[0208] 2. It can quickly transform complex scenarios encountered on actual roads, generate scenario test sets that meet different test platforms and different test requirements, and accelerate rapid software iteration.

[0209] The following describes an apparatus for updating an autonomous driving test scenario set based on real data provided by the present invention. The apparatus for updating an autonomous driving test scenario set based on real data described below and the method for updating an autonomous driving test scenario set based on real data described above can refer to each other.

[0210] like Figure 3 As shown, an embodiment of the present invention further provides an automatic driving test scenario set updating system based on real data, comprising:

[0211] A complex judgment module 1, used for dividing the real road operation data of the autonomous driving into a complex scene data segment and a non-complex scene data segment according to a first condition; the first condition is a complex criterion determined according to the autonomous driving capability boundary and / or the autonomous driving safety boundary;

[0212] Iterative test scenario module 2, used for extracting the design operation domain ODD elements and dynamic parameter groups in the complex scenario data fragment, and generating an iterative test scenario according to the design operation domain ODD elements and dynamic parameter groups in the complex scenario data fragment;

[0213] The parameter space updating module 3 is used to update the parameter space of the logical scene in the existing test scene set based on the non-complex scene data fragment; the logical scene includes a static road network and a dynamic target; the parameter space refers to the physical parameter range and / or physical parameter value set in the logical scene.

[0214] Furthermore, the automatic driving test scenario set updating system based on real data also includes:

[0215] A first labeling module, used to label the complex scene data segments corresponding to the data IDs one by one, to obtain one or more first labels;

[0216] The first tag includes driving actions, design operation domain (ODD) elements or dynamic parameters in the complex scene data segment.

[0217] The first tag also includes any one or any combination of the following:

[0218] The driving action, design operation domain (ODD) element or dynamic parameter in the complex scene data segment corresponds to the functional unit in the automatic driving system;

[0219] Standard label, which refers to a label that meets the ISO 34504 standard;

[0220] Custom labels.

[0221] A second labeling module, used to label the iterative test scenario to obtain one or more second labels;

[0222] The second tag includes a driving action, a design operation domain (ODD) element or a dynamic parameter in the iterative test scenario determined according to the first tag.

[0223] The second tag also includes any one or any combination of the following:

[0224] The driving actions, design operation domain (ODD) elements or dynamic parameters in the iterative test scenario correspond to functional units in the autonomous driving system;

[0225] Standard label, which refers to a label that meets the ISO 34504 standard;

[0226] Custom labels;

[0227] The version number of the automated driving system and / or test scenario generation software;

[0228] Test scenario types, including real vehicle tests, simulation tests, problem scenarios, regulatory scenarios, danger scenarios, accident scenarios, and perception scenarios.

[0229] A second condition screening module is used to screen the complex scene data fragments according to the second condition to obtain uncovered fragments and covered fragments, and use the uncovered fragments as corrected complex scene data fragments;

[0230] The second condition is a criterion for determining whether the complex scene data segment is covered by the existing test scene set based on the existing test scene set; the uncovered segment is a complex scene data segment that is not covered by the existing test scene set; and the covered segment is a complex scene data segment that is covered by the existing test scene set.

[0231] The first testing module is used to perform software-in-the-loop testing and / or hardware-in-the-loop testing on the autonomous driving model based on the iterative testing scenario, and update the autonomous driving model according to the results of the software-in-the-loop testing and / or hardware-in-the-loop testing.

[0232] The second testing module is used to perform a closed-scene real-vehicle test on the autonomous driving model based on the iterative test scenario, and update the autonomous driving model according to the closed-scene real-vehicle test results.

[0233] The third testing module is used to perform an open scenario reliability test and / or a driver-in-the-loop test on the autonomous driving model based on the iterative test scenario, and update the autonomous driving model according to the results of the open scenario reliability test and / or the driver-in-the-loop test.

[0234] The first adjustment module is used to adjust the first condition and / or the second condition according to the software-in-the-loop test and / or the hardware-in-the-loop test results.

[0235] The second adjustment module is used to adjust the first condition and / or the second condition according to the closed scene real vehicle test results.

[0236] A third adjustment module is used to adjust the first condition and / or the second condition according to the open scenario reliability test and / or the driver-in-the-loop test results.

[0237] The second condition screening module includes:

[0238] An existing information submodule is used to obtain an existing information set of an existing test scenario in an existing test scenario set;

[0239] The first label judgment submodule is used to judge the complex scene data segment according to the first label:

[0240] If the first tag of the complex scene data segment does not belong to the existing information set, determining that the complex scene data segment is an uncovered segment;

[0241] If the first tag of the complex scene data segment belongs to an existing information set, the complex scene data segment is determined to be a covered segment.

[0242] The iterative test scenario module 2 includes:

[0243] A positioning and segmentation submodule, used for locating and segmenting the complex scene data fragment in the real road operation data according to the data ID;

[0244] The parsing and extraction submodule is used to parse the complex scene data fragments obtained by segmentation and extract the design operation domain ODD elements and dynamic parameter groups.

[0245] The parameter space updating module 3 comprises:

[0246] A third label submodule is used to label the non-complex scene data segments corresponding to the data IDs one by one, and obtain one or more third labels; the third labels include driving actions, design operation domain ODD elements or dynamic parameters in the non-complex scene data segments;

[0247] A clustering submodule, configured to cluster the non-complex scene data segments according to the third label to obtain one or more clustering results;

[0248] The corresponding submodule of statistical screening is used to calculate the scene distribution probability, the design operation domain ODD element distribution and the dynamic parameter space based on the clustering results to obtain statistical results;

[0249] The parameter updating submodule is used to update the parameter space of the logical scenarios in the existing test scenario set according to the statistical results, and to filter the parameter space through the scenario generation tool to generate a specific scenario; the specific scenario refers to a logical scenario including specific parameter values.

[0250] The third tag also includes any one or any combination of the following:

[0251] The driving action, design operation domain (ODD) element or dynamic parameter in the non-complex scene data segment corresponds to a functional unit in the automatic driving system;

[0252] Standard label, which refers to a label that meets the ISO 34504 standard;

[0253] Custom labels.

[0254] Figure 4 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4 As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430 and a communication bus 440, wherein the processor 410, the communication interface 420 and the memory 430 communicate with each other through the communication bus 440. The processor 410 may call the logic instructions in the memory 430 to execute the method for updating the automatic driving test scenario set based on real data, the method comprising: dividing the real road operation data of the automatic driving into complex scenario data segments and non-complex scenario data segments according to the first condition; the first condition is a complex criterion determined according to the automatic driving capability boundary and / or the automatic driving safety boundary; extracting the design operation domain ODD element and the dynamic parameter group in the complex scenario data segment, and generating an iterative test scenario according to the design operation domain ODD element and the dynamic parameter group in the complex scenario data segment; updating the parameter space of the logical scenario in the existing test scenario set based on the non-complex scenario data segment; the logical scenario includes a static road network and a dynamic target; the parameter space refers to the physical parameter range and / or the physical parameter value set in the logical scenario.

[0255] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for 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, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0256] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the automatic driving test scenario set updating method based on real data provided by the above methods, the method including: according to a first condition, dividing the real road operation data of the automatic driving into complex scene data segments and non-complex scene data segments; the first condition is a complex criterion determined according to the automatic driving capability boundary and / or the automatic driving safety boundary; extracting the design operation domain ODD elements and dynamic parameter groups in the complex scene data segments, and generating iterative test scenarios according to the design operation domain ODD elements and dynamic parameter groups in the complex scene data segments; updating the parameter space of the logical scene in the existing test scenario set based on the non-complex scene data segments; the logical scene includes a static road network and a dynamic target; the parameter space refers to the physical parameter range and / or the physical parameter value set in the logical scene.

[0257] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the method for updating an autonomous driving test scenario set based on real data provided by the above-mentioned methods, the method comprising: according to a first condition, dividing the real road operation data of autonomous driving into complex scene data segments and non-complex scene data segments; the first condition is a complex criterion determined according to the autonomous driving capability boundary and / or the autonomous driving safety boundary; extracting the design operation domain ODD elements and dynamic parameter groups in the complex scene data segments, and generating iterative test scenarios based on the design operation domain ODD elements and dynamic parameter groups in the complex scene data segments; updating the parameter space of the logical scenes in the existing test scenario set based on the non-complex scene data segments; the logical scenes include static road networks and dynamic targets; the parameter space refers to the physical parameter range and / or physical parameter value set in the logical scene.

[0258] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0259] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0260] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. 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. A method for updating an autonomous driving test scenario set based on real data, characterized in that: include: According to the first condition, the real road operation data of the autonomous driving is divided into complex scene data segments and non-complex scene data segments; The first condition is a complex criterion determined according to the autonomous driving capability boundary and / or the autonomous driving safety boundary; Extracting the design operation domain (ODD) elements and the dynamic parameter group in the complex scenario data fragment, and generating an iterative test scenario according to the design operation domain (ODD) elements and the dynamic parameter group in the complex scenario data fragment; Based on the non-complex scene data fragment, the parameter space of the logical scene in the existing test scene set is updated; the logical scene includes a static road network and a dynamic target; the parameter space refers to the physical parameter range and / or the physical parameter value set in the logical scene; The first condition is a scene access rule, and the scene access rule is set based on the driver's takeover action; The scene access rules include autonomous driving capability boundary rules and safety rules. The autonomous driving capability boundary rules are determined according to the autonomous driving capability boundary, and the safety rules are determined according to the autonomous driving safety boundary.

2. The method for updating the automatic driving test scenario set based on real data according to claim 1, characterized in that: After the step of dividing the real road operation data of the autonomous driving into complex scene data segments and non-complex scene data segments according to the first condition, the step further includes: Labeling the complex scene data segments corresponding to the data IDs one by one to obtain one or more first labels; The first tag includes driving actions, design operation domain (ODD) elements or dynamic parameters in the complex scene data segment.

3. The method for updating the automatic driving test scenario set based on real data according to claim 2, characterized in that: The first tag also includes any one or any combination of the following: The driving action, design operation domain (ODD) element or dynamic parameter in the complex scene data segment corresponds to the functional unit in the automatic driving system; Standard label, which refers to a label that meets the ISO 34504 standard; Custom labels.

4. The method for updating the automatic driving test scenario set based on real data according to claim 2 or 3, characterized in that: The step of extracting the design operation domain ODD element and the dynamic parameter group in the complex scene data segment comprises: Locate and segment the complex scene data segment in the real road operation data according to the data ID; The complex scene data segments obtained by parsing and segmenting are extracted to extract the design operation domain ODD elements and dynamic parameter groups.

5. The method for updating the automatic driving test scenario set based on real data according to claim 2 or 3, characterized in that: After the step of extracting the design operation domain ODD elements and dynamic parameter groups in the complex scenario data fragment, and generating an iterative test scenario according to the design operation domain ODD elements and dynamic parameter groups in the complex scenario data fragment, the method further includes: Labeling the iterative test scenario to obtain one or more second labels; The second tag includes a driving action, a design operation domain (ODD) element or a dynamic parameter in the iterative test scenario determined according to the first tag.

6. The method for updating the automatic driving test scenario set based on real data according to claim 5, characterized in that: The second tag also includes any one or any combination of the following: The driving actions, design operation domain (ODD) elements or dynamic parameters in the iterative test scenario correspond to functional units in the autonomous driving system; Standard label, which refers to a label that meets the ISO 34504 standard; Custom labels; The version number of the automated driving system and / or test scenario generation software; Test scenario types, including real vehicle tests, simulation tests, problem scenarios, regulatory scenarios, danger scenarios, accident scenarios, and perception scenarios.

7. The method for updating the automatic driving test scenario set based on real data according to claim 1, characterized in that: The step of updating the parameter space of the logical scene in the existing test scene set based on the non-complex scene data fragment comprises: Label the non-complex scene data segments corresponding to the data IDs one by one to obtain one or more third tags; the third tags include driving actions, design operation domain (ODD) elements or dynamic parameters in the non-complex scene data segments; Clustering the non-complex scene data segments according to the third label to obtain one or more clustering results; Based on the clustering results, the scenario distribution probability, the design operation domain ODD element distribution and the dynamic parameter space are calculated to obtain statistical results; The parameter space of the logical scenarios in the existing test scenario set is updated according to the statistical results, and the parameter space is filtered through a scenario generation tool to generate a specific scenario; the specific scenario refers to a logical scenario including specific parameter values.

8. The method for updating the automatic driving test scenario set based on real data according to claim 7, characterized in that: The third tag also includes any one or any combination of the following: The driving action, design operation domain (ODD) element or dynamic parameter in the non-complex scene data segment corresponds to a functional unit in the automatic driving system; Standard label, which refers to a label that meets the ISO 34504 standard; Custom labels.

9. The method for updating the automatic driving test scenario set based on real data according to claim 2, characterized in that: After the step of dividing the real road operation data of the autonomous driving into complex scene data segments and non-complex scene data segments according to the first condition, the step further includes: According to the second condition, the complex scene data fragments are screened to obtain uncovered fragments and covered fragments, and the uncovered fragments are used as corrected complex scene data fragments; The second condition is a criterion for determining whether the complex scene data segment is covered by the existing test scene set based on the existing test scene set; the uncovered segment is a complex scene data segment that is not covered by the existing test scene set; and the covered segment is a complex scene data segment that is covered by the existing test scene set.

10. The method for updating the automatic driving test scenario set based on real data according to claim 9, characterized in that: The step of screening the complex scene data segments according to the second condition to obtain uncovered segments and covered segments comprises: Obtain an existing information set of an existing test scenario in an existing test scenario set; According to the first tag, a judgment is performed on the complex scene data segment: If the first tag of the complex scene data segment does not belong to the existing information set, determining that the complex scene data segment is an uncovered segment; If the first tag of the complex scene data segment belongs to an existing information set, the complex scene data segment is determined to be a covered segment.

11. The method for updating the automatic driving test scenario set based on real data according to claim 1 or 10, characterized in that: Also includes: Based on the iterative test scenario, software-in-the-loop testing and / or hardware-in-the-loop testing is performed on the autonomous driving model, and the autonomous driving model is updated according to the results of the software-in-the-loop testing and / or hardware-in-the-loop testing.

12. The method for updating the automatic driving test scenario set based on real data according to claim 11, characterized in that: After the step of performing software-in-the-loop testing and / or hardware-in-the-loop testing on the autonomous driving model based on the iterative test scenario, and updating the autonomous driving model according to the results of the software-in-the-loop testing and / or hardware-in-the-loop testing, the method further includes: Based on the iterative test scenario, a closed-scene real-vehicle test is performed on the autonomous driving model, and the autonomous driving model is updated according to the closed-scene real-vehicle test results.

13. The method for updating the automatic driving test scenario set based on real data according to claim 12, characterized in that: After the step of performing a closed-scene real-vehicle test on the autonomous driving model based on the iterative test scenario, and updating the autonomous driving model according to the closed-scene real-vehicle test result, the method further includes: Based on the iterative test scenario, an open scenario reliability test and / or a driver-in-the-loop test is performed on the autonomous driving model, and the autonomous driving model is updated according to the results of the open scenario reliability test and / or the driver-in-the-loop test.

14. The method for updating the automatic driving test scenario set based on real data according to claim 11, characterized in that: After the step of performing software-in-the-loop testing and / or hardware-in-the-loop testing on the autonomous driving model based on the iterative test scenario, and updating the autonomous driving model according to the results of the software-in-the-loop testing and / or hardware-in-the-loop testing, the method further includes: The first condition and / or the second condition are adjusted according to the software-in-the-loop test and / or the hardware-in-the-loop test results.

15. The method for updating the automatic driving test scenario set based on real data according to claim 12, characterized in that: After the step of performing a closed-scene real-vehicle test on the autonomous driving model based on the iterative test scenario, and updating the autonomous driving model according to the closed-scene real-vehicle test result, the method further includes: Adjust the first condition and / or the second condition according to the closed scene actual vehicle test results.

16. The method for updating the automatic driving test scenario set based on real data according to claim 13, characterized in that: After the step of performing an open scenario reliability test and / or a driver-in-the-loop test on the autonomous driving model based on the iterative test scenario, and updating the autonomous driving model according to the open scenario reliability test and / or the driver-in-the-loop test results, the method further includes: The first condition and / or the second condition are adjusted according to the open scenario reliability test and / or the driver-in-the-loop test results.

17. A system for updating an autonomous driving test scenario set based on real data, characterized in that: include: A complex judgment module, used for dividing the real road operation data of the autonomous driving into complex scene data segments and non-complex scene data segments according to the first condition; The first condition is a complex criterion determined according to the autonomous driving capability boundary and / or the autonomous driving safety boundary; An iterative test scenario module, used to extract the design operation domain ODD elements and dynamic parameter groups in the complex scenario data fragment, and generate an iterative test scenario according to the design operation domain ODD elements and dynamic parameter groups in the complex scenario data fragment; A parameter space updating module, used to update the parameter space of the logical scene in the existing test scene set based on the non-complex scene data fragment; the logical scene includes a static road network and a dynamic target; the parameter space refers to the physical parameter range and / or physical parameter value set in the logical scene; The first condition is a scene access rule, and the scene access rule is set based on the driver's takeover action; The scene access rules include autonomous driving capability boundary rules and safety rules. The autonomous driving capability boundary rules are determined according to the autonomous driving capability boundary, and the safety rules are determined according to the autonomous driving safety boundary.

18. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method for updating the autonomous driving test scenario set based on real data as described in any one of claims 1 to 16 are implemented.

19. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for updating an autonomous driving test scenario set based on real data as described in any one of claims 1 to 16 are implemented.

20. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method for updating an autonomous driving test scenario set based on real data as described in any one of claims 1 to 16 are implemented.

Images