Automatic driving system expected function safety test evaluation model construction method

Abstract

The invention provides a method for constructing an expected function safety test evaluation model of an automatic driving system. The method comprises the following steps: constructing a comprehensive test scene case library consisting of a closed site test scene and a simulation test scene; the test vehicle performs automatic driving scene test; constructing a comprehensive evaluation index; performing scene complexity analysis on the test scene in the comprehensive test scene case library, and introducing a scene factor weight according to the scene complexity; and determining a scene factor weight, and establishing an expected function safety evaluation model of the automatic driving system. A comprehensive test scene case library in which a closed site test scene and a simulation test scene are combined is constructed, so that the test scene coverage is improved; more comprehensive scientific quantitative evaluation of the expected function safety of the automatic driving system is realized through subjective and objective combined evaluation indexes; by establishing an evaluation model, the expected function safety performance of the automatic driving system can be comprehensively represented, and transverse comparison between single test scenes in a comprehensive test scene case library is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile autonomous driving testing, and in particular to a method for constructing a safety test evaluation model for the expected functions of an autonomous driving system. Background Art

[0002] At present, the pilot work for the access and road use of intelligent connected vehicles equipped with L3 and above autonomous driving systems has been carried out. Its main goal is to promote the promotion and application of intelligent connected vehicles and guide automobile manufacturers and users to strengthen capacity building. In order to ensure the safety of the access and road use of intelligent connected vehicles, their autonomous driving systems must be comprehensively tested and evaluated. In addition to conventional functional tests, it is also necessary to carry out special expected functional safety tests of autonomous driving systems. Expected functional safety means that there are no unreasonable risks caused by hazards caused by insufficient design or performance limitations. Through expected functional safety tests, it is possible to evaluate the potential hazards that may exist in the autonomous driving system due to performance limitations and reasonably foreseeable human misuse, explore the system performance boundaries, ensure that the autonomous driving system can operate safely and reliably in various driving scenarios, and at the same time increase public trust in autonomous driving.

[0003] In terms of intended functional safety, relevant standards and regulations have been established at home and abroad. ISO 21448, intended functional safety, first proposed the concept of intended functional safety and standardized the overall process of intended functional safety of autonomous vehicles; "Intended Functional Safety of Road Vehicles" (GB / T43267-2023) provides a general demonstration framework and measures to ensure the safety of intended functions. However, these existing standards and regulations mainly regulate from the perspective of product development, and do not regulate specific test procedures and test methods. At the same time, the expected functional safety test methods in the existing technology have problems such as lack of diversity in test scenarios, single evaluation indicators, and the failure of the evaluation system to achieve horizontal comparison between multiple scenarios. The limited scenario coverage reduces the credibility of the test results, and it is impossible to achieve horizontal comparison between multiple scenario test results, which is not conducive to obtaining the boundary scenario information of the expected functional safety of the autonomous driving system.

[0004] In addition, there are two main types of expected functional safety testing methods: simulation testing and closed-field testing. Among them, the simulation testing method can conduct efficient testing of various scenarios while avoiding the risks of actual testing, but there is a deviation between the test results and the actual situation caused by simulation errors. The closed-field testing method is based on testing the reaction of the test vehicle under actual road conditions, which can provide a more realistic performance evaluation, but the test scenarios and situations are relatively simple, it is difficult to cover the complex situations on the real road, and it is easily restricted by external factors. Summary of the invention

[0005] One of the purposes of the present invention is to provide a method for constructing an evaluation model for the expected functional safety test of an autonomous driving system, so as to solve the problems of low test scenario coverage and single evaluation index in the expected functional safety test method in the prior art.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] A method for constructing an expected functional safety test evaluation model for an autonomous driving system comprises the following steps:

[0008] S1. Build a comprehensive test scenario case library consisting of closed field test scenarios and simulation test scenarios;

[0009] S2. Based on the comprehensive test scenario case library, perform an autonomous driving scenario test on the test vehicle, and obtain and save scenario test data;

[0010] S3. Constructing comprehensive evaluation indicators of the expected functional safety of the autonomous driving system based on the scenario test data, including objective evaluation indicators and subjective evaluation indicators;

[0011] Performing scenario complexity analysis on the test scenarios in the comprehensive test scenario case library, introducing scenario factor weights according to the scenario complexity, and determining the scenario factor weights according to the scenario test data;

[0012] S4. Establish an evaluation model for the expected functional safety of the autonomous driving system based on the scenario factor weights.

[0013] According to the above-mentioned technical means, by constructing a comprehensive test scenario case library that combines closed field test scenarios and simulation test scenarios, the diversity of test scenarios is increased, the test scenario coverage is improved, and the expected functional safety of the autonomous driving system can be tested more comprehensively; through a combination of subjective and objective evaluation indicators, a more comprehensive scientific quantitative evaluation of the expected functional safety of the autonomous driving system is achieved; by introducing scenario factor weights based on scenario complexity to establish the evaluation model, it is possible to comprehensively characterize the expected functional safety performance of the autonomous driving system, realize horizontal comparison between individual test scenarios in the comprehensive test scenario case library, and help to obtain boundary scenario information from low-scoring test result scenarios.

[0014] Furthermore, the objective evaluation index in step S3 includes:

[0015] (1) Collision avoidance rate I1, calculated as: Where C1 is the number of potential collision situations that exist during the test vehicle test, and C2 is the number of times the test vehicle successfully avoids collisions;

[0016] (2) Collision speed correction degree I2, calculated as: Where V0 is the vehicle speed before executing the avoidance or collision mitigation operation, and V1 is the vehicle speed at the time of collision;

[0017] (3) Collision severity I3, calculated as follows: I3 = [0, 1, 2, 3], where I3 = 0 means no casualties or minor injuries, I3 = 1 means moderate injuries, I3 = 2 means serious injuries but not life-threatening, and I3 = 3 means serious injuries and life-threatening;

[0018] (4) Driving path accuracy I4, calculated as: Where x is the longitudinal distance of the actual driving trajectory of the test vehicle, x∈[0,L], L is the maximum cumulative longitudinal distance of the test vehicle in the automatic driving mode; F(x) is the absolute value of the lateral distance deviation between the planned trajectory and the actual trajectory;

[0019] (5) Driving task failure rate I5, calculated as: Wherein, H is the number of tasks that the test vehicle attempts to perform in the automatic driving mode in the test scenario of the comprehensive test scenario case library, and h is the number of tasks that the test vehicle fails to successfully perform in the automatic driving mode in the test scenario of the comprehensive test scenario case library;

[0020] (6) Compliance with driving traffic rules I6, calculated as follows: I6 = [0, 1], where I6 = 0 means that there is no violation of traffic rules during the test, and I6 = 1 means that there is a violation of traffic rules during the test.

[0021] According to the above-mentioned technical means, multiple objective evaluation indicators are constructed, and the test results are quantified based on the test data of the test vehicle, which can make the test results more intuitively understood and analyzed, while ensuring the accuracy and comparability of the test results.

[0022] Further, when the number C1 of potential collision situations existing during the test vehicle test is equal to the number C2 of collisions successfully avoided by the test vehicle, the collision speed correction degree I2=1, and the collision severity I3=0.

[0023] Furthermore, in the step S3, the subjective evaluation index includes:

[0024] The driving trust I7 is calculated as follows: I7=[0,1,2], where I7=0 means that 30% or less of the driving behavior of the autonomous driving system is consistent with user expectations; I7=1 means that 30%-80% of the driving behavior of the autonomous driving system is consistent with user expectations; I7=2 means that 80% or more of the driving behavior of the autonomous driving system is consistent with user expectations.

[0025] Based on the above-mentioned technical means, subjective evaluation indicators are constructed to supplement the test results that cannot be expressed by the objective evaluation indicators, and provide rich and comprehensive information for the safety test of the expected functions of the autonomous driving system, so as to comprehensively evaluate the safety performance of the expected functions of the autonomous driving system.

[0026] Furthermore, in the step S3, introducing the scene factor weight according to the scene complexity includes the following sub-steps:

[0027] S31, stratifying each test scenario in the comprehensive test scenario case library, and extracting scene elements of each layer of the test scenario, wherein the scene elements include: road type, number of lanes, degree of structuring, number of static facilities, number of dynamic facilities, type of traffic participants, and ground visibility;

[0028] S32, defining basic element attributes of the scene elements and complexity scores corresponding to the basic element attributes i0 ;

[0029] Define the sub-element attributes of each scene element and the complexity score f corresponding to each sub-element attribute i ;

[0030] S33, according to the complexity score f corresponding to each of the basic element attributes i0 Calculate the basic scene complexity value F0, the calculation formula is:

[0031] According to the complexity score f corresponding to each of the sub-element attributes i Calculate the complexity value F of the test scenario described in each layer i , the calculation formula is:

[0032] S34, determining the scenario factor weight λ of each test scenario in the comprehensive test scenario case library according to the complexity value of the basic scenario and the complexity value of the test scenario at each layer i , the calculation formula is: Where i∈[1,n], n is the number of test scenarios.

[0033] According to the above-mentioned technical means, each test scenario is layered and the scene elements in each layer of the test scenario are quantified, which can unify the complex scene elements into comparable values, and is conducive to objectively evaluating the quality and effect of the expected functional safety test of the autonomous driving system.

[0034] Furthermore, the S4 step includes the following sub-steps:

[0035] S41. Establish the test result matrix X of a single test scenario i= [I1, I2, I3, I4, I5, I6, I7], and based on the test result matrix X of a single test scenario i Create a test result scoring matrix Where, X i is the test result under the i-th test scenario, i∈[1,n], n is the number of test scenarios, I ij is the jth comprehensive evaluation index I of the test results in the i-th test scenario j , j∈[1,7];

[0036] S42, for the j-th comprehensive evaluation index I of the test result in the i-th test scenario. j Conduct standardized operations;

[0037] S43, the jth comprehensive evaluation index I based on the standardized test results in the i-th test scenario j , determine the indicator weights of each comprehensive evaluation indicator in combination with information entropy;

[0038] S44. Combine the scenario factor weights and the indicator weights to establish evaluation models for single test scenarios and multiple test scenarios for the expected functional safety of the autonomous driving system.

[0039] Furthermore, the comprehensive evaluation index is divided into positive index and negative index, the positive index includes collision avoidance rate I1, collision speed correction degree I2 and driving confidence I7, the negative index includes collision severity I3, driving path accuracy I4, driving task failure rate I5 and compliance with driving traffic rules I6; the jth comprehensive evaluation index I of the test result under the i-th test scenario in the step S42 is j The standardization operation includes positive indicator standardization and negative indicator standardization;

[0040] The standardized formula of the positive indicator is:

[0041] The standardized formula of the negative indicator is:

[0042] In the formula, x ij1 and x ij2 is the jth indicator I of the test results in the i-th test scenario j The normalized value.

[0043] According to the above technical means, by evaluating the jth comprehensive evaluation index I of the test results under the i-th test scenario j Standardization operations enable direct comparison and calculation between different indicators, improving the comparability and calculability of test data.

[0044] Furthermore, the indicator weight ω of each comprehensive evaluation indicator in the step S43 is j The calculation formula is: In the formula, e j is information entropy; Where P ij is the indicator information ratio; Where i∈[1,n], j∈[1,7], and n is the number of test scenarios.

[0045] According to the above technical means, the indicator weights of each of the comprehensive evaluation indicators are determined in combination with information entropy. The indicator weights of each of the comprehensive evaluation indicators are calculated based on the discreteness and uncertainty of the data itself, making the calculation of the indicator weights of each of the comprehensive evaluation indicators more objective, fair and reliable.

[0046] Furthermore, in the step S44:

[0047] The calculation formula for the expected functional safety test results of the autonomous driving system in a single test scenario is: In the formula, c i is the test result value of the i-th test scenario, λ i is the scenario factor weight of the i-th test scenario, I j ×ω j is the product of the jth indicator and the corresponding weight, i∈[1,n], j∈[1,7], n is the number of test scenarios;

[0048] The calculation formula for the expected functional safety test results of the autonomous driving system in multiple test scenarios is: Where C is the expected functional safety test result value of the autonomous driving system in multiple test scenarios, i∈[1,n], and n is the number of test scenarios.

[0049] Based on the above-mentioned technical means, the evaluation model for a single test scenario can more accurately reflect the expected functional safety performance of the autonomous driving system under the test scenario, and is highly targeted; the multi-test scenario evaluation model covers a variety of different test scenarios and can more comprehensively evaluate the expected functional safety performance of the autonomous driving system under different scenarios.

[0050] Furthermore, the simulation test scenarios in step S1 include 60% common scenarios and 40% long-tail scenarios.

[0051] Based on the above-mentioned technical means, common scenarios are designed to verify the expected functional safety performance of the autonomous driving system under high-frequency, typical operating conditions to ensure its reliability and stability; long-tail scenarios focus on extreme or special situations that do not occur frequently but have potentially higher risks, and evaluate the performance of the autonomous driving system under boundary conditions.

[0052] The beneficial effects of the present invention are:

[0053] The present invention increases the diversity of test scenarios and improves the coverage of test scenarios by constructing a comprehensive test scenario case library that combines closed field test scenarios and simulation test scenarios, so as to more comprehensively test the expected functional safety of the autonomous driving system; a more comprehensive scientific quantitative evaluation of the expected functional safety of the autonomous driving system is achieved through a combination of subjective and objective evaluation indicators; an evaluation model for the expected functional safety of the autonomous driving system is established by introducing scenario factor weights based on scenario complexity, which can comprehensively characterize the expected functional safety performance of the autonomous driving system, realize horizontal comparison between single test scenarios in the comprehensive test scenario case library, and help to obtain boundary scenario information from low-scoring test result scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only 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.

[0055] Figure 1 is a flow chart of a method for constructing an expected functional safety test evaluation model for an autonomous driving system of the present invention;

[0056] Figure 2 It is a sub-step flow chart of step S3 in the present invention;

[0057] Figure 3 It is a sub-step flow chart of step S4 in the present invention. DETAILED DESCRIPTION

[0058] The following will describe the embodiments of the present invention with reference to the accompanying drawings and preferred embodiments. The accompanying drawings are only used for exemplary description and cannot be understood as limiting the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention, not for limiting the scope of protection of the present invention.

[0059] like Figures 1 to 3 As shown, this embodiment provides a method for constructing an expected functional safety test evaluation model for an autonomous driving system, comprising the following steps:

[0060] S1. Build a comprehensive test scenario case library consisting of closed field test scenarios and simulation test scenarios;

[0061] S2. Based on the comprehensive test scenario case library, perform autonomous driving scenario tests on the test vehicle, and obtain and save scenario test data;

[0062] S3. Construct comprehensive evaluation indicators for the expected functional safety of the autonomous driving system based on scenario test data, including objective evaluation indicators and subjective evaluation indicators;

[0063] Conduct scenario complexity analysis on the test scenarios in the comprehensive test scenario case library, introduce scenario factor weights based on scenario complexity, and determine scenario factor weights based on scenario test data;

[0064] S4. Establish an evaluation model for the expected functional safety of the autonomous driving system based on the weights of scenario factors.

[0065] The present invention increases the diversity of test scenarios and improves the coverage of test scenarios by constructing a comprehensive test scenario case library that combines closed field test scenarios and simulation test scenarios, so as to more comprehensively test the expected functional safety of the autonomous driving system. By combining subjective and objective evaluation indicators, a more comprehensive scientific quantitative evaluation of the expected functional safety of the autonomous driving system is achieved. By introducing scenario factor weights based on scenario complexity to establish an evaluation model, it is possible to comprehensively characterize the expected functional safety performance of the autonomous driving system, realize horizontal comparison between individual test scenarios in the comprehensive test scenario case library, and help to obtain boundary scenario information from low-scoring test result scenarios.

[0066] In this embodiment, the objective evaluation indicators in step S3 include:

[0067] (1) Collision avoidance rate I1, calculated as: Where C1 is the number of potential collision situations that exist during the test vehicle test, and C2 is the number of times the test vehicle successfully avoids collisions;

[0068] (2) Collision speed correction degree I2, calculated as: Where V0 is the vehicle speed before executing the avoidance or collision mitigation operation, and V1 is the vehicle speed at the time of collision;

[0069] (3) Collision severity I3, calculated as follows: I3 = [0, 1, 2, 3], where I3 = 0 means no casualties or minor injuries, I3 = 1 means moderate injuries, I3 = 2 means serious injuries but not life-threatening, and I3 = 3 means serious injuries and life-threatening;

[0070] (4) Driving path accuracy I4, calculated as: Where x is the longitudinal distance of the actual driving trajectory of the test vehicle, x∈[0,L], L is the maximum cumulative longitudinal distance of the test vehicle in the automatic driving mode; F(x) is the absolute value of the lateral distance deviation between the planned trajectory and the actual trajectory;

[0071] (5) Driving task failure rate I5, calculated as: Where H is the number of tasks that the test vehicle attempts to perform in the autonomous driving mode in the test scenarios of the comprehensive test scenario case library, and h is the number of tasks that the test vehicle fails to successfully perform in the autonomous driving mode in the test scenarios of the comprehensive test scenario case library;

[0072] (6) Compliance with driving traffic rules I6, calculated as follows: I6 = [0, 1], where I6 = 0 means that there is no violation of traffic rules during the test, and I6 = 1 means that there is a violation of traffic rules during the test.

[0073] Constructing multiple objective evaluation indicators and quantifying the test results based on the test data of the test vehicle can make the test results more intuitive to understand and analyze, while ensuring the accuracy and comparability of the test results.

[0074] In this embodiment, when the number C1 of potential collision situations existing during the test vehicle test is equal to the number C2 of collisions successfully avoided by the test vehicle, the collision speed correction degree I2=1, and the collision severity I3=0.

[0075] In this embodiment, in step S3, the subjective evaluation index includes:

[0076] The driving trust I7 is calculated as follows: I7=[0,1,2], where I7=0 means that 30% or less of the driving behavior of the autonomous driving system is consistent with user expectations; I7=1 means that 30%-80% of the driving behavior of the autonomous driving system is consistent with user expectations; I7=2 means that 80% or more of the driving behavior of the autonomous driving system is consistent with user expectations.

[0077] By constructing subjective evaluation indicators, we can supplement the deficiencies that cannot be expressed by objective evaluation indicators, and provide rich and comprehensive information for the expected functional safety test of the autonomous driving system, so as to comprehensively evaluate the expected functional safety performance of the autonomous driving system.

[0078] like Figure 2 As shown, in this embodiment, in step S3, introducing the scene factor weight according to the scene complexity includes the following sub-steps:

[0079] S31, stratifying each test scenario in the comprehensive test scenario case library, and extracting scenario elements of each layer of the test scenario, wherein the scenario elements include: road type, number of lanes, degree of structuring, number of static facilities, number of dynamic facilities, type of traffic participants, and ground visibility;

[0080] S32, defining basic element attributes of scene elements and complexity scores f corresponding to each basic element attribute i0 ;

[0081] Define the sub-element attributes of each scene element and the complexity score f corresponding to each sub-element attribute i ;

[0082] S33, according to the complexity score f corresponding to the attributes of each basic element i0 Calculate the basic scene complexity value F0, the calculation formula is:

[0083] According to the complexity score f corresponding to each sub-element attribute i Calculate the complexity value F of each layer of test scenarios i , the calculation formula is:

[0084] S34, determining the scenario factor weight λ of each test scenario in the comprehensive test scenario case library according to the complexity value of the basic scenario and the complexity value of each layer of the test scenario i , the calculation formula is: Where i∈[1,n], n is the number of test scenarios.

[0085] Layering the test scenarios and quantifying the scenario elements in each layer of the test scenarios can unify complex scenario elements into comparable values, which is conducive to objectively evaluating the quality and effectiveness of the expected functional safety testing of the autonomous driving system.

[0086] Preferably, each test scenario layer includes a road structure layer, a road facility layer, a traffic participant layer, and a weather environment layer; wherein the scene elements of the road structure layer include: number of lanes, road type, and degree of structuring; the scene elements of the road facility layer include: number of static facilities and number of dynamic facilities; the scene elements of the traffic participant layer include: type of traffic participants; and the scene elements of the weather environment layer include: ground visibility. Sub-element attributes of each scene element and the complexity score corresponding to each sub-element attribute f i As shown in Table 1:

[0087] Table 1

[0088]

[0089] The properties of each basic element are:

[0090] Straight road, number of lanes m is 1, structured degree J is 1, no road facility layer elements, no traffic participant layer elements, ground visibility l is greater than 2 kilometers, the corresponding complexity score f i0 As shown in Table 1 above.

[0091] like Figure 3 As shown, in this embodiment, step S4 includes the following sub-steps:

[0092] S41. Establish the test result matrix X of a single test scenario i = [I1, I2, I3, I4, I5, I6, I7], and based on the test result matrix X of a single test scenario i Create a test result scoring matrix Where, X i is the test result under the i-th test scenario, i∈[1,n], n is the number of test scenarios, I ij is the jth comprehensive evaluation index I of the test results in the i-th test scenario j , j∈[1,7]; where I1, I2, I3, I4, I5, I6, and I7 represent collision avoidance rate, collision speed correction degree, collision severity, driving path accuracy, driving task failure rate, compliance with driving traffic rules, and driving trust, respectively;

[0093] S42, the jth comprehensive evaluation index I of the test results in the i-th test scenario j Conduct standardized operations;

[0094] S43, the jth comprehensive evaluation index I based on the standardized test results in the i-th test scenario j , combined with information entropy, determine the indicator weights of each comprehensive evaluation indicator;

[0095] S44. Combine the scenario factor weights and indicator weights to establish evaluation models for single test scenarios and multiple test scenarios for the expected functional safety of the autonomous driving system.

[0096] In this embodiment, the comprehensive evaluation index is divided into positive index and negative index. The positive index includes collision avoidance rate I1, collision speed correction degree I2 and driving confidence I7. The negative index includes collision severity I3, driving path accuracy I4, driving task failure rate I5 and compliance with driving traffic rules I6. In step S42, the jth comprehensive evaluation index I of the test result under the i-th test scenario is calculated. j The standardization operation includes positive indicator standardization and negative indicator standardization;

[0097] The standardized formula for the positive indicator is:

[0098] The standardized formula for the negative indicator is:

[0099] In the formula, x ij1 and x ij2 is the jth indicator I of the test results in the i-th test scenario j The normalized value.

[0100] By using the jth comprehensive evaluation index I of the test results under the i-th test scenario j Standardization operations enable direct comparison and calculation between different indicators, improving the comparability and calculability of data.

[0101] In this embodiment, the indicator weight ω of each comprehensive evaluation indicator in step S43 is j The calculation formula is: In the formula, e j is information entropy; Where P ij is the indicator information ratio; Where i∈[1,n], j∈[1,7], and n is the number of test scenarios.

[0102] Combining information entropy to determine the indicator weights of each comprehensive evaluation indicator is to calculate the indicator weights of each comprehensive evaluation indicator based on the discreteness and uncertainty of the data itself, making the indicator weight calculation of each comprehensive evaluation indicator more objective, fair and reliable.

[0103] In this embodiment, in step S44:

[0104] The calculation formula for the expected functional safety test results of the autonomous driving system in a single test scenario is: In the formula, c i is the test result value of the i-th test scenario, λ i is the scenario factor weight of the i-th test scenario, I j ×ω j is the product of the jth indicator and the corresponding weight, i∈[1,n], j∈[1,7], n is the number of test scenarios;

[0105] The calculation formula for the expected functional safety test results of the autonomous driving system in multiple test scenarios is: Where C is the expected functional safety test result value of the autonomous driving system in multiple test scenarios, i∈[1,n], and n is the number of test scenarios.

[0106] The evaluation model for a single test scenario can more accurately reflect the expected functional safety performance of the autonomous driving system under that test scenario and is highly targeted; the multi-test scenario evaluation model covers a variety of different test scenarios and can more comprehensively evaluate the expected functional safety performance of the autonomous driving system under different scenarios.

[0107] In this embodiment, the simulation test scenarios in step S1 include 60% common scenarios and 40% long-tail scenarios. Common scenarios are intended to verify the expected functional safety performance of the autonomous driving system under high-frequency, typical operating conditions to ensure its reliability and stability; long-tail scenarios focus on extreme or special situations that are not common but have higher potential risks, and evaluate the performance of the autonomous driving system under boundary conditions.

[0108] Obviously, the above embodiments of the present invention are only examples for clearly illustrating the present invention, and are not intended to limit the implementation methods of the present invention. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A method for constructing a safety test evaluation model for the expected function of an autonomous driving system, characterized in that: The following steps are involved: S1. Build a comprehensive test scenario case library consisting of closed field test scenarios and simulation test scenarios; S2. Based on the comprehensive test scenario case library, perform an autonomous driving scenario test on the test vehicle, and obtain and save scenario test data; S3. Constructing comprehensive evaluation indicators of the expected functional safety of the autonomous driving system based on the scenario test data, including objective evaluation indicators and subjective evaluation indicators; Performing scenario complexity analysis on the test scenarios in the comprehensive test scenario case library, introducing scenario factor weights according to the scenario complexity, and determining the scenario factor weights according to the scenario test data; S4. Establish an evaluation model for the expected functional safety of the autonomous driving system based on the weights of the scenario factors.

2. The method for constructing a safety test evaluation model for the expected function of an autonomous driving system according to claim 1, characterized in that: The objective evaluation indicators in step S3 include: (1) Collision avoidance rate I1, calculated as: Where C1 is the number of potential collision situations that exist during the test vehicle test, and C2 is the number of times the test vehicle successfully avoids collisions; (2) Collision speed correction degree I2, calculated as: Where V0 is the vehicle speed before executing the avoidance or collision mitigation operation, and V1 is the vehicle speed at the time of collision; (3) Collision severity I3, calculated as follows: I3 = [0, 1, 2, 3], where I3 = 0 means no casualties or minor injuries, I3 = 1 means moderate injuries, I3 = 2 means serious injuries but not life-threatening, and I3 = 3 means serious injuries and life-threatening; (4) Driving path accuracy I4, calculated as: Where x is the longitudinal distance of the actual driving trajectory of the test vehicle, x∈[0,L], L is the maximum cumulative longitudinal distance of the test vehicle in the automatic driving mode; F(x) is the absolute value of the lateral distance deviation between the planned trajectory and the actual trajectory; (5) Driving task failure rate I5, calculated as: Wherein, H is the number of tasks that the test vehicle attempts to perform in the automatic driving mode in the test scenario of the comprehensive test scenario case library, and h is the number of tasks that the test vehicle fails to successfully perform in the automatic driving mode in the test scenario of the comprehensive test scenario case library; (6) Compliance with driving traffic rules I6, calculated as follows: I6 = [0, 1], I6 = 0 means there is no violation of traffic rules during the test, and I6 = 1 means there is a violation of traffic rules during the test.

3. The method for constructing a safety test evaluation model for the expected function of an autonomous driving system according to claim 2, characterized in that: When the number C1 of potential collision situations existing during the test vehicle test is equal to the number C2 of collisions successfully avoided by the test vehicle, the collision speed correction level I2=1, and the collision severity level I3=0.

4. The method for constructing a safety test evaluation model for the expected function of an autonomous driving system according to claim 2 or 3, characterized in that: In the step S3, the subjective evaluation index includes: The driving trust level I7 is calculated as follows: I7=[0, 1, 2], where I7=0 means that 30% or less of the driving behavior of the autonomous driving system is consistent with user expectations; I7=1 means that 30%-80% of the driving behavior of the autonomous driving system is consistent with user expectations; I7=2 means that 80% or more of the driving behavior of the autonomous driving system is consistent with user expectations.

5. The method for constructing a safety test evaluation model for the expected function of an autonomous driving system according to claim 4, characterized in that: In the step S3, introducing the scene factor weight according to the scene complexity includes the following sub-steps: S31, stratifying each test scenario in the comprehensive test scenario case library, and extracting scene elements of each layer of the test scenario, wherein the scene elements include: road type, number of lanes, degree of structuring, number of static facilities, number of dynamic facilities, type of traffic participants, and ground visibility; S32, defining basic element attributes of the scene elements and complexity scores corresponding to the basic element attributes i0 ; Define the sub-element attributes of each of the scene elements and the complexity score f corresponding to each of the sub-element attributes i ; S33, according to the complexity score f corresponding to each of the basic element attributes i0 Calculate the basic scene complexity value F0, the calculation formula is: According to the complexity score f corresponding to each of the sub-element attributes i Calculate the complexity value F of the test scenario described in each layer i , the calculation formula is: S34, determining the scenario factor weight λ of each test scenario in the comprehensive test scenario case library according to the complexity value of the basic scenario and the complexity value of the test scenario at each layer i , the calculation formula is: Where i∈[1,n], n is the number of test scenarios.

6. The method for constructing a safety test evaluation model for the expected function of an autonomous driving system according to claim 5, characterized in that: The S4 step includes the following sub-steps: S41. Establish the test result matrix X of a single test scenario i = [I1, I2, I3, I4, I5, I6, I7], and based on the test result matrix X of a single test scenario i Establishing the test result scoring matrix Where, X i is the test result under the i-th test scenario, i∈[1,n], n is the number of test scenarios, I ij is the jth comprehensive evaluation index I of the test results in the i-th test scenario j , j∈[1,7]; S42, for the j-th comprehensive evaluation index I of the test result in the i-th test scenario. j Conduct standardized operations; S43, the jth comprehensive evaluation index I based on the standardized test results in the i-th test scenario j , determine the indicator weights of each comprehensive evaluation indicator in combination with information entropy; S44. Combine the scenario factor weights and the indicator weights to establish evaluation models for single test scenarios and multiple test scenarios for the expected functional safety of the autonomous driving system.

7. The method for constructing a safety test evaluation model for the expected function of an autonomous driving system according to claim 6, characterized in that: The comprehensive evaluation index is divided into positive index and negative index, the positive index includes collision avoidance rate I1, collision speed correction degree I2 and driving confidence I7, the negative index includes collision severity I3, driving path accuracy I4, driving task failure rate I5 and compliance with driving traffic rules I6; the jth comprehensive evaluation index I of the test result under the i-th test scenario in the step S42 is calculated. j The standardization operation includes positive indicator standardization and negative indicator standardization; The standardized formula of the positive indicator is: The standardized formula of the negative indicator is: In the formula, x ij1 and x ij2 is the jth indicator I of the test results in the i-th test scenario j The normalized value.

8. The method for constructing a safety test evaluation model for the expected function of an autonomous driving system according to claim 6, characterized in that: The indicator weight ω of each comprehensive evaluation indicator in the step S43 j The calculation formula is: In the formula, e j is the information entropy, Where P ij is the indicator information ratio, Where i∈[1,n], j∈[1,7], and n is the number of test scenarios.

9. The method for constructing a safety test evaluation model for the expected function of an autonomous driving system according to claim 6, characterized in that: In the step S44: The calculation formula for the expected functional safety test results of the autonomous driving system in a single test scenario is: In the formula, c i is the test result value of the i-th test scenario, λ i is the scenario factor weight of the i-th test scenario, I j ×ω j is the product of the jth indicator and the corresponding weight, i∈[1,n], j∈[1,7], n is the number of test scenarios; The calculation formula for the expected functional safety test results of the autonomous driving system in multiple test scenarios is: Where C is the expected functional safety test result value of the autonomous driving system in multiple test scenarios, i∈[1,n], and n is the number of test scenarios.

10. The method for constructing a safety test evaluation model for the expected function of an autonomous driving system according to claim 1, characterized in that: The simulation test scenarios in step S1 include 60% common scenarios and 40% long-tail scenarios.

Images