Comprehensive evaluation method for autonomous driving test scenario and related device

Abstract

Provided in the present invention are a comprehensive evaluation method for an autonomous driving test scenario and a related device. The method comprises: acquiring accident data from historical traffic accidents, extracting pre-collision trajectories of accident participants, and determining test scenario elements on the basis of the pre-collision trajectories; performing weighting calculation on each test scenario element to obtain the weighting values of each test scenario element under a scenario complexity degree and a scenario risk degree; by means of the weighting values of each test scenario element under the scenario complexity degree and scenario risk degree, quantifying a risk index and a complexity degree index of each test scenario; and finally, using the complexity degree index and the risk index to perform comprehensive evaluation on each test scenario, thereby improving the effectiveness and accuracy of evaluation.

Description

A comprehensive evaluation method and related equipment for autonomous driving test scenarios Technical Field

[0001] The present invention relates to the technical field of autonomous driving performance testing, and in particular to a comprehensive evaluation method for autonomous driving test scenarios and related equipment. Background Art

[0002] With the development of software, hardware and communication technologies, the intelligence and networking of automobiles have received more and more attention. It is also a major trend of future development. Therefore, it is crucial to ensure the safe implementation of corresponding functions before autonomous driving vehicles are on the road.

[0003] Autonomous vehicles typically use various onboard sensors to perceive the road environment and plan their routes. Testing of autonomous driving performance is essential to ensure accuracy and safety. Testing and validation are crucial components of the development of intelligent connected vehicles, and developing autonomous driving test scenarios is particularly crucial. Identifying key scenarios is crucial for validating intelligent connected vehicles. Current methods for identifying key scenarios include analyzing autonomous driving scenarios using road potential risk analysis and using safety and traffic quality as evaluation indicators. Currently, the risk level of autonomous driving scenarios is primarily assessed based on three factors: road risk exposure, accident risk, and accident severity. These factors fail to accurately assess the risk level of autonomous driving scenarios. For example, on May 7, 2018, a small car traveling straight collided head-on with a turning truck, resulting in the driver's death. At the time of the accident, the car was being controlled by an autonomous driving system equipped with eight cameras, one millimeter-wave radar, and twelve ultrasonic radars. The investigation concluded that the car's autonomous driving system failed to detect the vehicle ahead and the driver was inattentive, contributing to the accident. When a small car travels straight on a normal road, its road risk exposure, accident risk, and accident severity levels are all relatively low, indicating a relatively low potential road risk in autonomous driving scenarios. The risk level in autonomous driving scenarios is within the normal range, so the small car test scenario can be considered normal and the system will not issue an alert.

[0004] A Chinese patent discloses a method for testing autonomous driving performance, comprising: receiving data related to actual traffic scenarios collected by sensors installed on a manually driven vehicle; constructing a test scenario for autonomous driving performance based on the data; and testing the autonomous driving performance in the constructed test scenario. The test scenario is easy to construct and can reflect the actual traffic environment, making the test results of autonomous driving performance more accurate. However, it only considers the road traffic environment and cannot accurately assess the risk level of autonomous driving scenarios. Furthermore, since the testing and evaluation of autonomous vehicles is still in its infancy, there is a lack of systematic theoretical research and support for test scenario analysis and the selection, matching, and construction of scene segments. This results in a weak system for test scenario evaluation and a lack of a reasonable and effective system for evaluating the complexity of various typical autonomous driving test scenarios.

[0005] Summary of the Invention

[0006] The present invention provides a comprehensive evaluation method and related equipment for autonomous driving test scenarios, the purpose of which is to improve the effectiveness and accuracy of the evaluation.

[0007] To achieve the above objectives, the present invention provides a comprehensive evaluation method for autonomous driving test scenarios, comprising:

[0008] Step 1: Based on the testing requirements of autonomous vehicles, historical traffic accidents are deconstructed to obtain accident data.

[0009] Step 2: reconstructing a test scenario based on the accident data, extracting pre-crash trajectories of the accident participants, and determining multiple test scenarios and test scenario elements under each test scenario based on the pre-crash trajectories, where the test scenario elements include multiple test scenarios;

[0010] Step 3: Calculate the weight of each test scenario element to obtain the weight value of each test scenario element under the scenario complexity and the weight value of each test scenario element under the scenario risk;

[0011] Step 4: for each test scenario element in each test scenario, combine the test scenario element with the attribute classification category corresponding to the test scenario element to obtain an initial scenario element matrix for each test scenario;

[0012] Step 5: Classify the categories according to the attributes and quantize each element in the initial scene element matrix to obtain the scene element matrix of each test scene;

[0013] Step 6: Combine the scenario element matrix of each test scenario with the weight value of the test scenario element under the scenario complexity to obtain the complexity index of each test scenario;

[0014] Step 7: Combine the scenario element matrix of each test scenario with the weight value of the test scenario element under the scenario risk to obtain the risk index of each test scenario;

[0015] Step 8: Conduct a comprehensive evaluation of each test scenario based on the complexity index and risk index to obtain the evaluation results of each test scenario.

[0016] Furthermore, the testing requirements for autonomous vehicles include vehicle functional safety testing and functional reliability testing that need to be carried out before autonomous vehicles are put on the road.

[0017] Furthermore, test scenario elements include dynamic elements and static elements;

[0018] The dynamic element includes five sub-elements: the speed of the host vehicle before the collision, the dynamic driving task, the relative motion state, the relative position difference, and the behavior of the target vehicle;

[0019] The static elements include 12 sub-elements, namely: vehicle type of the main vehicle, vehicle type of the target vehicle, lighting conditions, weather, road water conditions, number of lanes, road type, road flatness, signal control, number of obstacles, number of target vehicles, and road slope.

[0020] More specifically, step 3 includes:

[0021] Input the test scene elements into the spherical blur network;

[0022] The spherical fuzzy module in the spherical fuzzy network generates a language scale standard and a pre-stored expert scoring table, and scores all test scenario elements and sub-elements based on risk and complexity to obtain a comparison matrix.

[0023] The fuzzy values ​​of the comparison matrix were calculated using the spherical weighted geometric mean;

[0024] The fuzzy value is defuzzified to obtain the weight value of each test scenario element under the scenario complexity and the weight value of each test scenario element under the scenario risk.

[0025] Furthermore, before using the spherical weighted geometric mean to calculate the fuzzy value of the comparison matrix, it also includes:

[0026] Through the consistency check function Eliminate the uncertainty of the comparison matrix. When the value of CR is less than the preset value, it is determined that the comparison matrix meets the uncertainty requirements.

[0027] Furthermore, the expression for calculating the fuzzy value of the comparison matrix using the spherical weighted geometric mean is:

[0028] Among them, μij is the membership degree, v ij is the non-membership degree, π ij is the fuzziness, and n is the number of elements in each row of the comparison matrix.

[0029] Furthermore, the expression for defuzzifying the fuzzy value is:

[0030] Among them, μ s is the membership degree, v s is the non-membership degree, π s is the fuzziness, and s represents a spherical fuzzy set.

[0031] The present invention also provides a comprehensive evaluation device for an autonomous driving test scenario, comprising:

[0032] The deconstruction module is used to deconstruct historical traffic accidents and obtain accident data based on the testing requirements of autonomous vehicles;

[0033] An extraction module is used to reconstruct a test scenario based on the accident data, extract the pre-crash trajectory of the accident participants, and determine multiple test scenarios and test scenario elements under each test scenario based on the pre-crash trajectory, where the test scenario elements include multiple;

[0034] A calculation module is used to calculate the weight of each test scenario element to obtain the weight value of each test scenario element under the scenario complexity and the weight value of each test scenario element under the scenario risk;

[0035] A first combining module is used to combine each test scenario element in each test scenario with the attribute classification category corresponding to the test scenario element to obtain an initial scenario element matrix for each test scenario;

[0036] The quantification module is used to classify the categories according to the attributes and quantify each element in the initial scene element matrix to obtain the scene element matrix of each test scene;

[0037] The second combining module is used to combine the scenario element matrix of each test scenario with the weight value of the test scenario element under the scenario complexity to obtain the complexity index of each test scenario;

[0038] The third combining module is used to combine the scenario element matrix of each test scenario with the weight value of the test scenario element under the scenario risk degree to obtain the risk index of each test scenario;

[0039] The evaluation module is used to conduct a comprehensive evaluation of each test scenario based on complex indicators and risk indicators to obtain the evaluation results of each test scenario.

[0040] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, a comprehensive evaluation method for autonomous driving test scenarios is implemented.

[0041] The present invention also provides a terminal device, including a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the computer program, a comprehensive evaluation method for autonomous driving test scenarios is implemented.

[0042] The above solution of the present invention has the following beneficial effects:

[0043] The present invention deconstructs historical traffic accidents according to test requirements, reconstructs test scenarios after obtaining accident data, extracts pre-collision trajectories of accident participants, and determines multiple test scenarios and test scenario elements under each test scenario based on the pre-collision trajectories, and performs weight calculation on each test scenario element to obtain the weight value of each test scenario element under the scenario complexity and the scenario risk; for each test scenario element under each test scenario, the test scenario element is combined with the attribute classification category corresponding to the test scenario element to obtain the initial scenario element matrix of each test scenario; according to the attribute classification category, each element in the initial scenario element matrix is ​​quantified to obtain the scenario element matrix of each test scenario; the scenario element matrix of each test scenario is combined with the weight value of the test scenario element under the scenario complexity to obtain the complexity of each test scenario. Complex indicators; combining the scenario element matrix of each test scenario with the weight value of the test scenario element under the scenario risk to obtain the risk indicator of each test scenario; comprehensively evaluating each test scenario according to the complex indicators and risk indicators to obtain the evaluation result of each test scenario; compared with the prior art, the present invention obtains accident data from historical traffic accidents and extracts the pre-collision trajectory of the accident participants, and then determines the test scenario and test scenario elements according to the pre-collision trajectory and performs weight calculation on each test scenario element to obtain the weight value of each test scenario element under the scenario complexity and scenario risk, quantifies the risk indicator and complex indicator of each test scenario by the weight value of each test scenario element under the scenario complexity and scenario risk, and finally uses the complex indicators and risk indicators to comprehensively evaluate each test scenario, thereby improving the effectiveness and accuracy of the evaluation.

[0044] Other beneficial effects of the present invention will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] FIG1 is a schematic diagram of a process flow of an embodiment of the present invention;

[0046] FIG2 is a schematic diagram of a partial structure of an ANP network;

[0047] FIG3 is a schematic diagram of a partial structure of an ANP network;

[0048] Figure 4 is a schematic diagram of the structure of a spherical fuzzy set. DETAILED DESCRIPTION

[0049] To make the technical problems, technical solutions, and advantages to be solved by the present invention more clear, the following is a detailed description with reference to the accompanying drawings and specific embodiments. It is obvious that the embodiments described are only some of the embodiments of the present invention, not all of them. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0050] In the description of the present invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate and simplify the description of the present invention. They are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation, and therefore should not be construed as limitations on the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0051] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood broadly. For example, they may refer to a locking connection, a detachable connection, or an integral connection; they may refer to a mechanical connection or an electrical connection; they may refer to a direct connection or an indirect connection through an intermediate medium; and they may refer to internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0052] In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0053] In response to existing problems, the present invention provides a comprehensive evaluation method and related equipment for autonomous driving test scenarios.

[0054] As shown in FIG1 , an embodiment of the present invention provides a comprehensive evaluation method for an autonomous driving test scenario, including:

[0055] Step 1: Based on the testing requirements of autonomous vehicles, historical traffic accidents are deconstructed to obtain accident data.

[0056] Step 2: reconstructing a test scenario based on the accident data, extracting pre-crash trajectories of the accident participants, and determining multiple test scenarios and test scenario elements under each test scenario based on the pre-crash trajectories, where the test scenario elements include multiple test scenarios;

[0057] Step 3: Calculate the weight of each test scenario element to obtain the weight value of each test scenario element under the scenario complexity and the weight value of each test scenario element under the scenario risk;

[0058] Step 4: for each test scenario element in each test scenario, combine the test scenario element with the attribute classification category corresponding to the test scenario element to obtain an initial scenario element matrix for each test scenario;

[0059] Step 5: Classify the categories according to the attributes and quantize each element in the initial scene element matrix to obtain the scene element matrix for each test scene;

[0060] Step 6: Combine the scenario element matrix of each test scenario with the weight value of the test scenario element under the scenario complexity to obtain the complexity index of each test scenario;

[0061] Step 7: Combine the scenario element matrix of each test scenario with the weight value of the test scenario element under the scenario risk to obtain the risk index of each test scenario;

[0062] Step 8: Conduct a comprehensive evaluation of each test scenario based on the complexity index and risk index to obtain the evaluation results of each test scenario.

[0063] Specifically, the testing requirements for autonomous vehicles include vehicle functional safety testing and functional reliability testing that need to be carried out before autonomous vehicles are put on the road.

[0064] In an embodiment of the present invention, since autonomous driving vehicles face many challenges when driving on the road, including bad weather, cyber attacks, and complex traffic scenarios, vehicle functional safety and functional reliability tests are required before going on the road.

[0065] In an embodiment of the present invention, a test scenario is reconstructed based on accident data to extract the pre-collision trajectory of the accident participants, that is, the vehicle trajectory information within 5 seconds before the accident is reconstructed through historical traffic accident information in the China In-depth Study on Travel Safety CIMSS-TA database system through PC-crash.

[0066] PC-crash is a traffic accident reconstruction software, which is mainly used for accident analysis. The accident analysis process is mainly based on the collection, recording, investigation and analysis of the accident scene. The vehicles involved in the accident are reversed from their end position after the collision to the collision process, and then reversed to the operating state before the collision to analyze the cause of the accident.

[0067] The embodiment of the present invention determines the speed of the main vehicle before the collision, dynamic driving tasks, relative motion state, relative position difference, and behavior of the target vehicle by analyzing the pre-collision trajectory of the accident; and determines the vehicle type of the main vehicle, the vehicle type of the target vehicle, lighting conditions, weather, road waterlogging, number of lanes, road type, road flatness, signal control, number of obstacles, number of target vehicles, and road slope through information in the accident database.

[0068] Specifically, test scenario elements are necessary elements required to carry out simulation test scenarios, which are divided into dynamic elements and static elements;

[0069] The dynamic element includes the status of the accident participants immediately before the collision, which includes five sub-elements: the speed of the main vehicle before the collision, the dynamic driving task, the relative motion state, the relative position difference, and the behavior of the target vehicle;

[0070] The static elements include 12 sub-elements, namely: vehicle type of the main vehicle, vehicle type of the target vehicle, lighting conditions, weather, road water conditions, number of lanes, road type, road flatness, signal control, number of obstacles, number of target vehicles, and road slope.

[0071] Specifically, step 3 includes:

[0072] Input the test scene elements into the Spherical Fuzzy-Analysis Network (SFAN);

[0073] The Spherical Fuzzy-Analysis Network (SFAN) is a model that uses spherical fuzzy sets (SFS) in conjunction with ANP. Spherical fuzzy sets are added to the network structure constructed by ANP during the comparison matrix construction stage to reduce the limitations of human subjective evaluation. The network structure of ANP is shown in Figures 2 and 3, and the structure of spherical fuzzy sets is shown in Figure 4.

[0074] Generate language scale standards and pre-stored expert scoring tables through the spherical fuzzy module in the spherical fuzzy network;

[0075] In the embodiment of the present invention, the spherical fuzzy module places the membership, non-membership and fuzzy values ​​on the set sphere to calculate the principle method of obtaining the spherical fuzzy set. The specific method is as follows: s = [ <x,(μs (x),υ s (x),π s (x))|x∈X>]

[0076]

[0077] The language scale standards generated by this method are shown in Table 1 below:

[0078] Table 1

[0079] Each language scale is (μ s (x), v s (x),π s (x)), where μ s (x) represents the degree of membership, v s (x) represents the non-membership degree, π s (x) represents the fuzziness. All test scenario elements and sub-elements are scored under the risk and complexity to obtain a comparison matrix, as shown in Table 2 below:

[0080] Table 2

[0081] In the embodiment of the present invention, before using the spherical weighted geometric mean to calculate the fuzzy value of the comparison matrix, the consistency check function Eliminate the uncertainty of the comparison matrix. When the value of CR is less than the preset value, it is determined that the comparison matrix meets the uncertainty requirements.

[0082] in, λ represents the largest eigenvector, n represents the dimension of the matrix, and in the embodiment of the present invention, the preset value is 0.1;

[0083] The spherical weighted geometric mean is used to calculate the fuzzy value of the comparison matrix, and the expression is:

[0084] Among them, μ ij is the membership degree, v ij is the non-membership degree, π ij is the fuzziness, n is the number of elements in each row of the comparison matrix, and what do i and j represent respectively;

[0085] Defuzzify the fuzzy value, the expression is:

[0086] Among them, μ s is the membership degree, v s is the non-membership degree, π s is the fuzziness, s represents the spherical fuzzy set;

[0087] Calculate the local weight determined by each comparison matrix, the expression is as follows:

[0088] The weight values ​​of each test scenario element under the scenario complexity and the weight values ​​of each test scenario element under the scenario risk are shown in Table 3 below:

[0089] Table 3

[0090] In the embodiment of the present invention, different comparison matrices are obtained based on the relative importance of elements determined by the expert scoring table under the complexity and risk. Different weight values ​​are calculated.

[0091] Specifically, step 4 includes:

[0092] For each test scenario element, the test scenario element is combined with the attribute classification category corresponding to the test scenario element, and the expression of the initial scenario element matrix of each test scenario is obtained as follows:

[0093] Specifically, step 5 includes:

[0094] Classify the test scene elements according to fixed categories to obtain the attribute classification categories of each element in each accident scene. The attribute classification categories are divided into five equal parts according to the range of values ​​of each test scene element;

[0095] According to the attribute classification, a hierarchical quantitative index P = [0.2, 0.4, 0.6, 0.8, 1.0] is established. Each element in the initial scene element matrix is ​​quantified to obtain the scene element matrix B of each test scene. The expression is:

[0096] in, represents the weight value of complexity and risk, and U represents the initial scenario element matrix of each of the above test scenarios;

[0097] The embodiment of the present invention runs simulations for all the above test scenarios based on APOLLO. The vehicle driving results are divided into collision scenarios and collision avoidance scenarios. The collision scenario uses collision speed as an indicator and performs correlation analysis with the complexity index and risk index of the test scenario proposed in the embodiment of the present invention. The analysis results are shown in Table 4 below:

[0098] Table 4

[0099] Based on the calculation results of the correlation coefficients in Table 4 above, it is found that the operating indicators are weakly positively correlated with the scenario complexity and strongly positively correlated with the risk level.

[0100] Specifically, the collision avoidance scenario uses the minimum GTTC as an indicator and performs a correlation analysis with the complexity index and risk index of the test scenario proposed in the embodiment of the present invention. The analysis results are shown in Table 5 below:

[0101] Table 5

[0102] Based on the calculation results of the correlation coefficients in Table 5 above, it is found that the operating indicators are weakly negatively correlated with the scenario complexity and moderately negatively correlated with the risk level.

[0103] An embodiment of the present invention develops a comprehensive evaluation method for autonomous driving test scenarios based on a spherical fuzzy network. The method adopts a combination of factor weights and factor quantification to obtain the risk and complexity of each scenario, thereby realizing the calibration of the scenario attributes. The complexity value and the risk value are used to select the scenario according to the function to be tested to realize the function test of the autonomous driving vehicle.

[0104] In summary, the embodiment of the present invention deconstructs historical traffic accidents according to test requirements, reconstructs test scenarios after obtaining accident data, extracts pre-collision trajectories of accident participants, and determines test scenario elements of the autonomous driving vehicle based on the pre-collision trajectories and performs weight calculation on each test scenario element to obtain the weight value of each test scenario element under the scenario complexity and the scenario risk; for each test scenario element under each test scenario, the test scenario element is combined with the attribute classification category corresponding to the test scenario element to obtain the initial scene element matrix of each test scenario; according to the attribute classification category, each element in the initial scene element matrix is ​​quantified to obtain the scene element matrix of each test scenario; the scene element matrix of each test scenario is combined with the weight value of the test scenario element under the scenario complexity to obtain the Complexity index; combining the scenario element matrix of each test scenario with the weight value of the test scenario element under the scenario risk to obtain the risk index of each test scenario; comprehensively evaluating each test scenario based on the complexity index and the risk index to obtain the evaluation result of each test scenario; compared with the prior art, the present invention obtains accident data from historical traffic accidents and extracts the pre-collision trajectory of the accident participants, and then determines the test scenario elements according to the pre-collision trajectory and performs weight calculation on each test scenario element to obtain the weight value of each test scenario element under the scenario complexity and the scenario risk, quantifies the risk index and complexity index of each test scenario by the weight value of each test scenario element under the scenario complexity and the scenario risk, and finally uses the complexity index and the risk index to comprehensively evaluate each test scenario, thereby improving the effectiveness and accuracy of the evaluation.

[0105] The present invention also provides a comprehensive evaluation device for an autonomous driving test scenario, comprising:

[0106] The deconstruction module is used to deconstruct historical traffic accidents and obtain accident data based on the testing requirements of autonomous vehicles;

[0107] An extraction module, configured to reconstruct a test scenario based on the accident data, extract the pre-crash trajectories of the accident participants, and determine a test scenario element for the autonomous driving vehicle based on the pre-crash trajectories. The test scenario elements may include multiple elements.

[0108] A calculation module is used to calculate the weight of each test scenario element to obtain the weight value of each test scenario element under the scenario complexity and the weight value of each test scenario element under the scenario risk;

[0109] A first combining module is used to combine each test scenario element in each test scenario with the attribute classification category corresponding to the test scenario element to obtain an initial scenario element matrix for each test scenario;

[0110] The quantification module is used to classify the categories according to the attributes and quantify each element in the initial scene element matrix to obtain the scene element matrix of each test scene;

[0111] The second combining module is used to combine the scenario element matrix of each test scenario with the weight value of the test scenario element under the scenario complexity to obtain the complexity index of each test scenario;

[0112] The third combining module is used to combine the scenario element matrix of each test scenario with the weight value of the test scenario element under the scenario risk degree to obtain the risk index of each test scenario;

[0113] The evaluation module is used to conduct a comprehensive evaluation of each test scenario based on complex indicators and risk indicators to obtain the evaluation results of each test scenario.

[0114] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the method embodiment of the embodiment of the present invention. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0115] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the embodiments of the present invention. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0116] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, a comprehensive evaluation method for autonomous driving test scenarios is implemented.

[0117] If the integrated module 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 present invention implements all or part of the process in the above-mentioned method embodiment, which can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by the processor, it can implement the steps of the above-mentioned various method embodiments. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may at least include: any entity or device capable of carrying the computer program code to a construction device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electric carrier signal, a telecommunication signal and a software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk.

[0118] The present invention also provides a terminal device, including a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the computer program, a comprehensive evaluation method for autonomous driving test scenarios is implemented.

[0119] The terminal device may be a computing device such as a desktop computer, a notebook computer, a PDA, a server, a server cluster, a cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0120] The processor may be a central processing unit (CPU), other general-purpose processors, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0121] In some embodiments, the memory may be an internal storage unit of the terminal device, such as a hard disk or memory of the terminal device. In other embodiments, the memory may also be an external storage device of the terminal device, such as a plug-in hard disk equipped on the terminal device, a smart memory card (SMC, SmartMedia Card), a secure digital (SD, Secure Digital) card, a flash card, etc. Furthermore, the memory may include both an internal storage unit of the terminal device and an external storage device. The memory is used to store an operating system, an application program, a boot loader, data, and other programs, such as the program code of the computer program. The memory may also be used to temporarily store data that has been output or is about to be output.

[0122] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the method embodiment of the embodiment of the present invention. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0123] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the embodiments of the present invention. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0124] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A comprehensive evaluation method for autonomous driving test scenarios, characterized in that: include: Step 1: Based on the testing requirements of autonomous vehicles, historical traffic accidents are deconstructed to obtain accident data. Step 2: reconstructing a test scenario based on the accident data, extracting pre-crash trajectories of the accident participants, and obtaining multiple test scenarios and test scenario elements under each test scenario based on the pre-crash trajectories, wherein the test scenario elements include multiple test scenarios; Step 3: Calculate the weight of each test scenario element to obtain the weight value of each test scenario element under the scenario complexity and the weight value of each test scenario element under the scenario risk; Step 4: for each of the test scenario elements in each of the test scenarios, combine the test scenario element with the attribute classification category corresponding to the test scenario element to obtain an initial scenario element matrix for each of the test scenarios; Step 5, classifying the categories according to the attributes, quantizing each element in the initial scene element matrix, and obtaining a scene element matrix for each of the test scenes; Step 6: combining the scene element matrix of each test scene with the weight value of the test scene element under the scene complexity to obtain the complexity index of each test scene; Step 7: Combining the scenario element matrix of each test scenario with the weight value of the test scenario element under the scenario risk to obtain the risk index of each test scenario; Step 8: Perform a comprehensive evaluation on each of the test scenarios based on the complexity index and the risk index to obtain an evaluation result for each of the test scenarios.

2. The comprehensive evaluation method for autonomous driving test scenarios according to claim 1, characterized in that: The testing requirements for autonomous vehicles include vehicle functional safety testing and functional reliability testing that need to be performed before the autonomous vehicle is put on the road.

3. The comprehensive evaluation method for autonomous driving test scenarios according to claim 2, characterized in that: The test scene elements include dynamic elements and static elements; The dynamic element includes five sub-elements: the speed of the main vehicle before the collision, the dynamic driving task service, relative motion state, relative position difference, and target vehicle behavior; The static elements include 12 sub-elements, namely: vehicle type of the main vehicle, vehicle type of the target vehicle, lighting conditions, weather, road water conditions, number of lanes, road type, road flatness, signal control, number of obstacles, number of target vehicles, and road slope.

4. The comprehensive evaluation method for autonomous driving test scenarios according to claim 2, characterized in that: The step 3 comprises: Inputting the test scene elements into a spherical fuzzy network; Generating a language scale standard and a pre-stored expert scoring table through a spherical fuzzy module in the spherical fuzzy network, scoring all the test scene elements and sub-elements based on risk and complexity to obtain a comparison matrix; Calculating the fuzzy value of the comparison matrix using a spherical weighted geometric mean; The fuzzy value is defuzzified to obtain the weight value of each of the test scene elements under the scene complexity and the weight value of each of the test scene elements under the scene risk.

5. The comprehensive evaluation method for autonomous driving test scenarios according to claim 4, characterized in that: Before calculating the fuzzy value of the comparison matrix using the spherical weighted geometric mean, the method further includes: Through the consistency check function The uncertainty of the comparison matrix is ​​eliminated. When the value of CR is less than a preset value, it is determined that the comparison matrix meets the uncertainty requirement.

6. The comprehensive evaluation method for autonomous driving test scenarios according to claim 5, characterized in that: The expression for calculating the fuzzy value of the comparison matrix using the spherical weighted geometric mean is: Among them, μ ij is the membership degree, v ij is the non-membership degree, π ij is the fuzziness, and n is the number of elements in each row of the comparison matrix.

7. The comprehensive evaluation method for autonomous driving test scenarios according to claim 6, characterized in that: The expression for defuzzifying the fuzzy value is: Among them, μ s is the membership degree, v s is the non-membership degree, π s is the fuzziness, and s represents a spherical fuzzy set.

8. A comprehensive evaluation device for autonomous driving test scenarios, characterized in that: include: The deconstruction module is used to deconstruct historical traffic accidents and obtain accident data based on the testing requirements of autonomous vehicles; an extraction module, configured to reconstruct a test scenario based on the accident data, extract pre-crash trajectories of the accident participants, and determine a plurality of test scenarios and test scenario elements under each of the test scenarios based on the pre-crash trajectories, wherein the test scenario elements include a plurality of test scenario elements; A calculation module, configured to perform weight calculation on each of the test scenario elements to obtain a weight value of each of the test scenario elements under scenario complexity and a weight value of each of the test scenario elements under scenario risk; A first combining module is configured to combine, for each of the test scenario elements in each of the test scenarios, the test scenario element with the attribute classification category corresponding to the test scenario element to obtain an initial scenario element matrix for each of the test scenarios; a quantization module, configured to quantify each element in the initial scene element matrix according to the attribute classification, to obtain a scene element matrix for each of the test scenes; A second combining module is used to combine the scene element matrix of each test scene with the weight value of the test scene element under the scene complexity to obtain the complexity index of each test scene; A third combining module is used to combine the scenario element matrix of each test scenario with the weight value of the test scenario element under the scenario risk to obtain the risk index of each test scenario; An evaluation module is used to perform a comprehensive evaluation on each of the test scenarios based on the complexity index and the risk index to obtain an evaluation result for each of the test scenarios.

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 comprehensive evaluation method for the autonomous driving test scenario as described in any one of claims 1 to 7 is implemented.

10. A terminal 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, it implements the comprehensive evaluation method of the autonomous driving test scenario according to any one of claims 1 to 7.

Images