A scene-based unmanned system credibility evaluation method and device

Abstract

The application relates to the technical field of unmanned driving and discloses a scene-based credibility evaluation method and device for an unmanned driving system. According to a specific function to be tested of the unmanned driving system, scene elements and function test requirements are randomly combined to generate a function test scene library. According to the function test requirements, evaluation indexes of the unmanned driving system and experience values of the evaluation indexes are determined. Simulation testing is performed based on the function test scene library and the experience values to obtain a failure rate of the unmanned driving system under the evaluation indexes. A dynamic Bayesian network credibility evaluation model is constructed by using the failure rate, and a state probability of the unmanned driving system is obtained by using the dynamic Bayesian network credibility evaluation model. Finally, the credibility of the unmanned driving system and the importance of components are calculated according to the state probability, and the credibility of the unmanned driving system can be evaluated.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, specifically to a scenario-based reliability evaluation method and apparatus for autonomous driving systems. Background Technology

[0002] The Chinese Association for Artificial Intelligence's white paper on artificial intelligence, "Intelligent Driving (2017)," points out that driverless vehicles are an emerging technology under the backdrop of a new round of technological revolution. The proposals of strategic plans such as "Made in China 2025," "Internet Plus," and the intelligent manufacturing development plan also indicate that the development of driverless vehicles will be of paramount importance in the future. Currently, the actual deployment of driverless systems is affected by many factors, including a lack of technological means, limited market and consumer understanding and acceptance, and ethical and legal dilemmas. These problems indefinitely prolong the process of large-scale popularization of driverless cars.

[0003] A key step in developing and deploying autonomous vehicles is comprehensively testing and evaluating the trustworthiness of autonomous driving systems to meet the urgent needs of high-level intelligent vehicles that prioritize safety, understandability, predictability, and traceability. Research into the domestic and international technological landscape reveals that trustworthiness evaluation technology for autonomous driving systems is currently lacking. Summary of the Invention

[0004] This application provides a scenario-based reliability evaluation method for autonomous driving systems, addressing the problem that existing technologies for reliability evaluation of autonomous driving systems are still lacking, making it difficult to meet the urgent needs of high-level intelligent vehicles that prioritize safety, understandability, predictability, and traceability.

[0005] Accordingly, embodiments of this application also provide a scenario-based autonomous driving system reliability evaluation device, an electronic device, and a computer-readable storage medium to ensure the implementation and application of the above methods.

[0006] To address the aforementioned technical problems, this application discloses a scenario-based reliability evaluation method for autonomous driving systems, the method comprising:

[0007] Based on the specific functions to be tested in the autonomous driving system, scene elements and functional testing requirements are randomly combined to generate a functional test scenario library.

[0008] Based on the functional testing requirements, determine the evaluation indicators and empirical values ​​for the autonomous driving system.

[0009] Simulation tests are conducted based on a functional test scenario library and empirical values ​​of indicators to obtain the failure rate of the autonomous driving system under the evaluation indicators.

[0010] A dynamic Bayesian network credibility evaluation model is constructed using the failure rate, and the state probability of the autonomous driving system is obtained by using the output of the dynamic Bayesian network credibility evaluation model.

[0011] The reliability of the autonomous driving system and the importance of its components are calculated based on state probabilities.

[0012] This application also discloses a scenario-based reliability evaluation device for autonomous driving systems, the device comprising:

[0013] The functional test scenario confirmation module is used to randomly combine scenario elements and functional test requirements based on the specific functions to be tested in the autonomous driving system, and generate a functional test scenario library.

[0014] The evaluation index generation module is used to determine the evaluation indexes of the autonomous driving system and the empirical values ​​of the evaluation indexes based on the functional testing requirements.

[0015] The system failure rate calculation module is used to conduct simulation tests based on the functional test scenario library and index experience values ​​to obtain the failure rate of the autonomous driving system under the evaluation index.

[0016] The model building and training module is used to build a dynamic Bayesian network credibility evaluation model using the failure rate, and to obtain the state probability of the autonomous driving system using the output of the dynamic Bayesian network credibility evaluation model.

[0017] The system reliability calculation module is used to calculate the reliability of the autonomous driving system and the importance of its components based on state probabilities.

[0018] This application also discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements one or more of the methods described in this application.

[0019] This application also discloses a computer-readable storage medium storing a computer program that, when executed by a processor, implements one or more of the methods described in this application.

[0020] In this embodiment, based on the specific functions to be tested in the autonomous driving system, scene elements and functional testing requirements are randomly combined to generate a functional testing scene library. Based on the functional testing requirements, evaluation indicators and empirical values ​​for these indicators are determined. Simulation tests are conducted based on the functional testing scene library and the empirical values ​​to obtain the failure rate of the autonomous driving system under the evaluation indicators. A dynamic Bayesian network credibility evaluation model is constructed using the failure rate, and the output of this model is used to obtain the state probabilities of the autonomous driving system. Finally, the credibility and component importance of the autonomous driving system are calculated based on the state probabilities, thus enabling a credibility evaluation of the autonomous driving system.

[0021] Additional aspects and advantages of the embodiments of this application will be set forth in the following description, and will become apparent from the description or may be learned by practice of this application. Attached Figure Description

[0022] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0023] Figure 1 A flowchart illustrating the scenario-based reliability evaluation method for autonomous driving systems provided in this application embodiment;

[0024] Figure 2 This is a schematic diagram of the structure of the human-vehicle-road-environment system provided in the embodiments of this application;

[0025] Figure 3 A flowchart for selecting evaluation indicators provided in the embodiments of this application;

[0026] Figure 4 This is a structural diagram of the disassembled autonomous driving system provided in an embodiment of this application;

[0027] Figure 5 A schematic diagram of the dynamic Bayesian network graphical structure provided in the embodiments of this application;

[0028] Figure 6 A schematic diagram of the dynamic Bayesian network graphical structure provided in the examples of embodiments of this application;

[0029] Figure 7 A schematic diagram of the structure of the scenario-based autonomous driving system reliability evaluation device provided in the embodiments of this application;

[0030] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0031] The embodiments of this application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0032] Those skilled in the art will understand that, unless explicitly stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this application means the presence of features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or combinations thereof. It should be understood that when we say an element is “connected” or “coupled” to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein can include wireless connections or wireless coupling. The term “and / or” as used herein includes all or any units and all combinations of one or more associated listed items.

[0033] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0034] The solutions provided in this application can be executed by any electronic device, such as a terminal device or a server. The server can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. The terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, etc., but is not limited to these. The terminal and server can be directly or indirectly connected via wired or wireless communication, which is not limited herein. Regarding the technical problems existing in the prior art, the scenario-based autonomous driving system credibility evaluation method, apparatus, and electronic device provided in this application aim to solve at least one of the technical problems of the prior art.

[0035] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0036] This application provides a possible implementation method, such as... Figure 1 The diagram shows a flowchart of a scenario-based reliability evaluation method for autonomous driving systems. This method can be executed by any electronic device, and optionally, it can be executed on a server or a terminal device.

[0037] like Figure 1 As shown, the method may include the following steps:

[0038] Step 101: Based on the specific functions to be tested in the autonomous driving system, select scene elements and functional test requirements and randomly combine them to generate a functional test scenario library.

[0039] The autonomous driving system in this application embodiment can be a "human-vehicle-road-environment system". Figure 2 A schematic diagram of the human-vehicle-road-environment system is shown. (For example...) Figure 2 As shown, the scene elements in the human-vehicle-road-environment system may include, but are not limited to, traffic participants, environmental physical elements, and weather conditions; functional test requirements include, but are not limited to, traffic sign recognition and response, traffic light recognition and response, pedestrian avoidance, roundabout passage, parking, and overtaking. In this embodiment, based on the functional test requirements of the autonomous driving system in selected typical scenarios, a functional test scenario library for typical scenarios can be constructed by freely / randomly combining the above-mentioned various scene elements and various functional test requirements.

[0040] Step 102: Based on the functional testing requirements, determine the evaluation indicators of the autonomous driving system and the empirical values ​​of the evaluation indicators.

[0041] Optionally, determining the evaluation indicators and empirical values ​​of the evaluation indicators for the autonomous driving system based on functional testing requirements includes:

[0042] Based on the functional testing requirements, determine the functional modules in the autonomous driving system;

[0043] Identify the components included in the functional module, and determine the evaluation indicators and empirical values ​​of the evaluation indicators based on the components.

[0044] Figure 3 A flowchart for selecting evaluation indicators is shown. For example... Figure 3As shown, in this embodiment, the tasks to be completed by the autonomous driving system in typical scenarios (i.e., the specific functions to be tested by the autonomous driving system) are first defined, and then the functional modules used by the autonomous driving system to complete the tasks are identified. The functional modules are then broken down to identify the components (including hardware components and related algorithms) included in each module. Based on the decomposed components, relevant indicators for evaluating task completion are selected as evaluation indicators. Then, combined with pre-set standards (such as existing policies or generally recognized industry standards), an evaluation indicator system is formed based on all selected evaluation indicators, and empirical values ​​for all evaluation indicators within the evaluation indicator system are obtained.

[0045] An indicator value is a quantitative description that defines the standard for achieving an indicator. The selected evaluation indicators, based on empirical values, are derived from the pre-defined standards for the reliability evaluation of scenario-based autonomous driving systems, resulting in existing, widely accepted indicator values ​​within the industry.

[0046] Step 103: Conduct simulation tests based on the functional test scenario library and indicator experience values ​​to obtain the failure rate of the autonomous driving system under the evaluation indicators.

[0047] Based on a functional test scenario library and empirical indicator values, simulation tests are conducted on a simulation platform. Through repeated and extensive testing, the indicator values ​​can be verified and optimized, ultimately yielding the failure rate of the autonomous driving system under various indicators. The simulation platform used for testing can be selected based on the specific application requirements.

[0048] Step 104: Construct a dynamic Bayesian network credibility evaluation model using the failure rate, and use the output of the dynamic Bayesian network credibility evaluation model to obtain the state probability of the autonomous driving system.

[0049] Bayesian Networks (BNs) are uncertain knowledge reasoning models based on Bayesian theory. A Bayesian Network model consists of a graph structure of nodes and a conditional probability table (CPT), which can intuitively describe the logical relationships between nodes. Dynamic Bayesian Networks (DBNs) are an extension of Bayesian Networks, capable of reflecting the time-varying effects of variables through multiple discrete time slices. In a DBN structure, each time slice corresponds to a static Bayesian Network. Therefore, DBNs can achieve dynamic prediction and real-time data updates, and can express and reason about dynamic stochastic processes, offering advantages in applications such as dynamic system fault diagnosis and reliability assessment. In this embodiment, a DBN is used as the architecture for constructing a Bayesian Network reliability evaluation model, and then the output of the Bayesian Network reliability evaluation model is used to obtain the state probabilities of the autonomous driving system.

[0050] Step 105: Calculate the reliability and component importance of the autonomous driving system based on the state probability.

[0051] In this embodiment, based on the specific functions to be tested in the autonomous driving system, scene elements and functional testing requirements are randomly combined to generate a functional testing scene library. Based on the functional testing requirements, evaluation indicators and empirical values ​​for these indicators are determined. Simulation tests are conducted based on the functional testing scene library and the empirical values ​​to obtain the failure rate of the autonomous driving system under the evaluation indicators. A dynamic Bayesian network credibility evaluation model is constructed using the failure rate, and the output of this model is used to obtain the state probabilities of the autonomous driving system. Finally, the credibility and component importance of the autonomous driving system are calculated based on the state probabilities, thus enabling a credibility evaluation of the autonomous driving system.

[0052] In addition, the embodiments of this application can closely link the testing requirements of autonomous driving systems with the testing scenarios, enrich the performance evaluation indicators of each functional module of the autonomous driving system, reduce the development and maintenance costs of the autonomous driving system, and improve the testing efficiency of the autonomous driving system, thereby improving the safety of autonomous driving technology and promoting the deployment and application of high-level autonomous driving on actual roads.

[0053] In an optional embodiment, the step of conducting simulation tests based on a functional test scenario library and empirical index values ​​to obtain the failure rate of the autonomous driving system under the evaluation index includes:

[0054] Determine the empirical value of each evaluation indicator under each functional module; let the empirical value of the selected evaluation indicator be:

[0055]

[0056] Where: n i Let be the number of evaluation indicators in the i-th functional module; This is the empirical value of the nth indicator in the i-th functional module.

[0057] Test and verify the empirical values ​​of the indicators, and obtain the optimized empirical values ​​of the indicators as the evaluation indicators.

[0058] Specifically, through repeated testing and verification on the simulation platform until a preset threshold is reached, the optimized evaluation index value is finally obtained as follows:

[0059]

[0060] Where: n i Let be the number of evaluation indicators in the i-th functional module; This is the optimized value of the nth indicator in the i-th functional module;

[0061] The autonomous driving system is tested based on the evaluation index values ​​to obtain the failure rate of the autonomous driving system under the evaluation index.

[0062] By conducting repeated and extensive tests on the autonomous driving system under test on a simulation platform, combined with optimized evaluation index values, until a preset threshold is reached, the failure rate of the autonomous driving system under the corresponding evaluation index can be obtained:

[0063]

[0064] in: Let f(x) represent the failure rate of the autonomous driving system under the nth metric in the i-th functional module.

[0065] The failure rate needs to be measured using different standards depending on the test scenario or test requirements. The chosen standard for measuring the failure rate of an autonomous driving system can be based on the test mileage (system failure rate per 100 kilometers / human intervention rate), the test duration (system failure rate per 100 hours / human intervention rate), or the frequency of successful or failed system task completion (system failure frequency), etc.

[0066] In an optional embodiment, the step of constructing a dynamic Bayesian network credibility evaluation model using the failure rate, and obtaining the state probability of the autonomous driving system using the output of the dynamic Bayesian network credibility evaluation model, includes:

[0067] The autonomous driving system is functionally and hierarchically divided to obtain multiple functional modules and multiple system levels;

[0068] Based on the functional modules and system hierarchy, the graph structure of the dynamic Bayesian network is determined, and the credibility evaluation model of the dynamic Bayesian network is constructed.

[0069] Based on the failure rate, the conditional probability table parameters of each node in the Bayesian network graph structure are calculated for all possible states under various evaluation indicators; where the conditional probability table parameters include the probability values ​​of each node in all possible states.

[0070] The state probability values ​​are used as input to the dynamic Bayesian network credibility evaluation model, and the state probability of the autonomous driving system is obtained based on the output of the Bayesian network credibility evaluation model.

[0071] This application uses a Data Structured Network (DBN) as the architecture for constructing a trustworthiness evaluation model. Since the trustworthiness of an autonomous driving system is a function of time, the trustworthiness of complex autonomous driving systems changes dynamically over time. When modeling the trustworthiness of complex autonomous driving systems, nodes in the DBN can represent the states of the system and components, while directed edges between nodes can represent the influence relationships between components and the system. Conditional probability tables are used to quantitatively describe the strength of the association between variables, i.e., the conditional probability of each node relative to all possible states of its parent node.

[0072] Based on the above, the reliability evaluation model of dynamic Bayesian network is constructed in the embodiments of this application in the following way:

[0073] Specifically, Figure 4 A schematic diagram of the disassembled structure of an autonomous driving system is shown. Figure 4 As shown, the autonomous driving system is first divided into functional and hierarchical categories.

[0074] In terms of functional division, autonomous driving systems can be divided into three functional modules from both the system itself and the overall vehicle perspective: perception module, decision-making and control module, and vehicle performance module. Each functional module can be further divided into components, and components can be divided into hardware and related algorithms. For example, the perception module can be divided into hardware and perception algorithms. Hardware can be divided into cameras, LiDAR, millimeter-wave radar, inertial measurement units (IMU), global positioning systems (GPS), etc.; perception algorithms can be divided into algorithms for target detection, target tracking (localization), and trajectory prediction.

[0075] The decision-making and control module can be divided into decision algorithms and control algorithms. Decision algorithms can be further categorized into algorithms based on empirical rules, data-driven approaches, utility functions, and those considering uncertainty and interactivity. Control algorithms can be classified into lateral control, longitudinal control, and lateral-longitudinal coordinated control algorithms. Lateral control can be further divided into model-free control methods, control methods based on vehicle kinematics models, and control methods based on vehicle dynamics models. Longitudinal control can be divided into control methods for cruise control, adaptive cruise control, and emergency braking. Lateral-longitudinal coordinated control can be divided into distributed and integrated collaborative methods.

[0076] The overall vehicle performance module needs to consider performance aspects such as stability, comfort, safety, and efficiency.

[0077] In terms of hierarchical division, it can be divided into system level (autonomous driving system), subsystem level (perception module, decision control module, vehicle performance module), and component level (hardware and related algorithms included in the system).

[0078] According to such Figure 4 The functional modules and system hierarchy shown can be used to set a dynamic Bayesian network graphical structure:

[0079] Figure 5 A schematic diagram of the dynamic Bayesian network graphical structure is shown, such as... Figure 5 As shown, the Bayesian network graphical structure follows... Figure 4 The functions of a mid-range autonomous driving system are divided into three parts: perception, decision-making, and vehicle performance; according to Figure 4 The hierarchical classification of autonomous driving systems is divided into three levels: system level, subsystem level, and component level.

[0080] The nodes in the dynamic Bayesian network graph structure, from top to bottom, are composed as follows:

[0081] Table 1. Node composition of the dynamic Bayesian network graph structure

[0082]

[0083] It should be noted that in Table 1 above, "system" refers to the autonomous driving system; "module" refers to the functional module; and "model" refers to the dynamic Bayesian network credibility evaluation model.

[0084] in, The loop arrow on the parent node indicates that the data of the dynamic Bayesian network credibility evaluation model is updated in real time, with an update interval of "unit 1", i.e., Δt = 1; the prediction time interval set by the dynamic Bayesian network credibility evaluation model is T0; therefore, the time setting of the entire dynamic Bayesian network credibility evaluation model satisfies: t = {0, 1, ... T0}.

[0085] Subsequently, based on the constructed dynamic Bayesian network graph structure, the conditional probability of each node in the dynamic Bayesian network graph structure is calculated.

[0086] It is important to note that this application uses a Bayesian Network (DBN) as the architecture for the reliability evaluation model. A DBN is a Bayesian network that incorporates temporal information. A DBN can be viewed as multiple Bayesian networks whose time slices are unidirectionally connected. The structure of the Bayesian network in each time slice and the relationships between its nodes are completely identical. The time slice network in frame t is only related to the time slice networks in frames t-1 and t+1, and is independent of other network slices. Therefore, regardless of the value of time t, the conditional probability table of each node remains the same and unchanged.

[0087] Optionally, the conditional probability table parameters for calculating all possible states of each node in the dynamic Bayesian network graph structure based on the failure rate include:

[0088] The failure rate is input into the dynamic Bayesian network credibility evaluation model;

[0089] The failure rate of the autonomous driving system obtained in step 103 under the corresponding evaluation index is used as the training input for the entire Bayesian network credibility evaluation model:

[0090]

[0091] in, Let f(x) represent the failure rate of the autonomous driving system under the nth evaluation metric in the i-th functional module.

[0092] Calculate the working probability of the autonomous driving system under each of the aforementioned evaluation indicators;

[0093]

[0094]

[0095] in, Let be the working probability of the autonomous driving system under the nth evaluation index in the i-th functional module;

[0096] Based on the aforementioned working probabilities, calculate and output the conditional probability table parameters for all possible states of each node in the Bayesian network graph structure under various evaluation indicators:

[0097]

[0098] Where: P r {Ω} represents the state probability values ​​of the Ω node set, where Ω = {X1(t), X2(t), ..., X} n (t)};

[0099] n is the number of nodes contained in the Ω node set;

[0100] T0 is the set time interval;

[0101] For X i (t) conditional probability of node;

[0102] For X i (t) represents all parent nodes of node (t).

[0103] The process involves using the state probability value as input to a dynamic Bayesian network credibility evaluation model, and obtaining the state probability of the autonomous driving system based on the output of the dynamic Bayesian network credibility evaluation model. Specifically:

[0104] The state probability values ​​of the system at time t under various evaluation indicators are used as the input to the dynamic Bayesian network credibility evaluation model:

[0105]

[0106] Where: n i Let be the number of evaluation indicators in the i-th functional module;

[0107] Let n be the state probability value of the autonomous driving system under the nth evaluation index of the i-th functional module;

[0108] The output of the dynamic Bayesian network credibility evaluation model is the predicted state probability S of the autonomous driving system at time t. W (t).

[0109] Based on the output of the constructed dynamic Bayesian network credibility evaluation model (the state probability S of the autonomous driving system) W (t)) is used to calculate the reliability of the autonomous driving system and the importance of its components.

[0110] To comprehensively elucidate the relationship between component reliability and the trustworthiness of autonomous driving systems, and to quantify the impact of component state changes on the trustworthiness of autonomous driving systems, a metric is needed. This application introduces Birnbaum Importance Measures (BIMs) as a standard to measure the impact of component reliability on the trustworthiness of autonomous driving systems.

[0111] The calculation process is as follows:

[0112] Reliability calculation of autonomous driving system:

[0113]

[0114] Where: R S (t) represents the reliability of the entire system at time t;

[0115] P r {S W (t)=1} is S W The probability that a node is in a "1" state at time t (when the entire system is in a working state);

[0116] For S W All parent nodes of (t);

[0117] Component importance calculation:

[0118]

[0119] in, Indicates the k-th functional module of the i-th function. i One component;

[0120] express The importance value of the component at time t;

[0121] To indicate When node S is in state "1" at time t, W The probability that a node is in a "1" state at time t (when The probability that the autonomous driving system is in a working state when the component is in a working state; this value can... As known conditions, they are obtained from the output of the dynamic Bayesian network credibility evaluation model;

[0122] Indicates when C i When node S is in state "0" at time t, W The probability that a node is in a "1" state at time t (when The probability that the system is in a working state when a component is in a failed state; this value can... As known conditions, they are obtained from the output of the dynamic Bayesian network credibility evaluation model.

[0123] In an optional embodiment, the method further includes:

[0124] The reliability of the autonomous driving system is updated based on the collected observation data.

[0125] Updating the credibility of an autonomous driving system refers to updating the credibility assessment results of the system using collected observation data. Specifically, when new observation data about the autonomous driving system's state is collected, the probability S of the new operating state of the system is inferred using a dynamic Bayesian network credibility evaluation model. W (t).

[0126] In an optional embodiment, updating the reliability of the autonomous driving system based on the collected observation data includes:

[0127] Accept observation data, input the observation data into the dynamic Bayesian network credibility evaluation model, and output the corresponding state probability;

[0128] The reliability of the autonomous driving system is recalculated based on the state probabilities.

[0129] Let the collected observation data be E. The observation data can be divided into single-level observation data and multi-level observation data.

[0130] Optionally, for reliability updates of single-level observation data:

[0131]

[0132] Where: P r {S W The result of E(t1) can be obtained by taking E(t1) as a known condition and outputting the reliability evaluation model of dynamic Bayesian network.

[0133] E(t1) is the single-level observation value at time t1;

[0134] S W (t) represents the predicted state probability of the autonomous driving system at time t, where t>t1;

[0135] For reliability updates of multi-level observation data:

[0136]

[0137] Where: P r {S W (t)|E(t1)} can be obtained by dividing E1(t1),E2(t1),…E m (t1) is a known condition, obtained from the output of the dynamic Bayesian network credibility evaluation model;

[0138] E1(t1),E2(t1),…E m (t1) represents the multi-level observation value at time t1;

[0139] S W(t) represents the predicted state probability of the autonomous driving system at time t, where t>t1.

[0140] As an example:

[0141] (1) In conjunction with step 101 above, this embodiment of the application takes three typical scenarios—urban intersections, roundabouts, and highway entrances / exits—as examples, based on... Figure 2 The structure of the human-vehicle-road-environment system is shown in Table 2. Table 2 lists the main functions to be tested in the selected typical scenarios for the autonomous driving system.

[0142] Table 2. Key Functional Test Points and Typical Scenario Components of Autonomous Driving Systems in Typical Urban Scenarios

[0143]

[0144]

[0145] (2) Combining the above step 102, based on the clearly defined functional testing requirements, determine the evaluation indicators and empirical values ​​of the evaluation indicators for the autonomous driving system.

[0146] Specifically, taking urban intersections as an example, the corresponding evaluation indicators for autonomous driving systems are shown in Table 3:

[0147] Table 3. Selection of Evaluation Indicators for Intersection Test Scenarios

[0148]

[0149]

[0150] (3) Combining step 103 above, conduct simulation tests based on the functional test scenario library and empirical index values ​​to obtain the failure rate of the autonomous driving system under the evaluation index. Specifically, as follows:

[0151] The simulation platform selected for virtual testing is SCANeR. TM studio.

[0152] SCANeR TM Studio is a modular autonomous driving simulation platform that provides all the tools and models needed to build hyper-realistic virtual worlds: road environments, vehicle dynamics, traffic, sensors, real / virtual drivers, headlights, weather conditions, and scene scripts. It is highly versatile and extremely powerful.

[0153] The overall simulation test process is as follows:

[0154] ①The empirical values ​​of the corresponding evaluation indicators selected in step (2) above are known to be:

[0155]

[0156] ② Through repeated testing and verification on the simulation platform, the optimized evaluation index values ​​were finally obtained as follows:

[0157]

[0158] ③ By conducting repeated and extensive tests on the autonomous driving system under test on a simulation platform, combined with the optimized evaluation index values ​​obtained in ②, the failure rate of the autonomous driving system under the corresponding evaluation index values ​​can be obtained:

[0159]

[0160] (4) Combining the above step 104, construct a dynamic Bayesian network credibility evaluation model using the failure rate, and use the output of the dynamic Bayesian network credibility evaluation model to obtain the state probability of the autonomous driving system.

[0161] The construction process of the dynamic Bayesian network credibility evaluation model is as follows:

[0162] ① Divide the autonomous driving system into functional and hierarchical categories:

[0163] In terms of functional division, the autonomous driving system is divided into three functional modules: perception module, decision control module, and vehicle performance module. Each functional module can be further divided into components, and each component can be divided into hardware and related algorithms.

[0164] In terms of hierarchical division, it can be divided into system level (autonomous driving system), subsystem level (perception module, decision control module, vehicle performance module), and component level (hardware and related algorithms included in the system).

[0165] ②Based on the functional modules and system hierarchy of the autonomous driving system, define the dynamic Bayesian network graph structure:

[0166] Figure 6 A schematic diagram of the dynamic Bayesian network graphical structure in the example is shown. Figure 6 As shown, the dynamic Bayesian network graphical structure is divided into three parts according to the functional modules of the above system decomposition: perception module, decision control module, and vehicle performance module; and into three levels according to the system hierarchy: system level, subsystem level, and component level.

[0167] In the dynamic Bayesian network graph structure, the leases of nodes, from top to bottom, are as follows:

[0168] Table 4. Node composition of the DBN graphical structure of the autonomous driving system in the intersection test scenario.

[0169]

[0170] The update interval is set to "unit 1", i.e., Δt = 1. The prediction time interval set for the dynamic Bayesian network credibility evaluation model is T0. Therefore, the time setting of the entire dynamic Bayesian network credibility evaluation model satisfies: t = {0, 1, ... T0}.

[0171] ③ Based on the constructed dynamic Bayesian network graph structure, calculate the conditional probability table parameters of all possible states of each node in the dynamic Bayesian network graph structure under various evaluation indicators:

[0172] The calculation process for the conditional probability table parameters of each node is as follows:

[0173] The failure rate of the autonomous driving system obtained in step (3) above under the corresponding evaluation index value is used as the training input for the dynamic Bayesian network credibility evaluation model:

[0174]

[0175] The operational probability of an autonomous driving system can be calculated under various evaluation metrics:

[0176]

[0177]

[0178] in: —The probability of the autonomous driving system operating under the nth evaluation metric in the i-th functional module;

[0179] —The failure rate of the autonomous driving system under the nth evaluation index in the i-th functional module;

[0180] Based on the obtained working probabilities, the conditional probability table parameters for all possible states of each node under various evaluation indicators can be calculated:

[0181]

[0182] Where: P r {Ω} represents the state probability values ​​of the Ω node set, where Ω = {X1(t), X2(t), ..., X} n (t)};

[0183] n is the number of nodes contained in the Ω node set;

[0184] T0 is the set time interval;

[0185] For X i (t) conditional probability of node;

[0186] For X i (t) represents all parent nodes of node (t).

[0187] ④ The input to the dynamic Bayesian network credibility evaluation model is: the state probability value of the system at time t under the measurement of various evaluation indicators:

[0188]

[0189] ⑤ The output of the dynamic Bayesian network credibility evaluation model is: the predicted state probability S of the system at time t. W (t).

[0190] (5) Based on the output of the constructed dynamic Bayesian network credibility evaluation model (the state probability S of the autonomous driving system) W (t)) is used to calculate the reliability of the autonomous driving system and the importance of its components.

[0191] The calculation process is as follows:

[0192] ① Reliability calculation of autonomous driving system:

[0193]

[0194] ② Calculation of component-level importance in autonomous driving systems:

[0195] by Taking the component (the camera component of the perception module) as an example, calculation Importance:

[0196]

[0197] in: Indicates when When node S is in state "1" at time t, W The probability that a node is in a "1" state at time t (the probability that the system is in an active state when the camera is active); this value can be used to... As known conditions, they are obtained from the output of the dynamic Bayesian network credibility evaluation model;

[0198] Indicates when When node S is in state "0" at time t, W The probability that a node is in a "1" state at time t (the probability that the system is in a working state when the camera is in a disabled state); this value can be used to... As known conditions, they are obtained from the output of the dynamic Bayesian network credibility evaluation model;

[0199] (6) Based on the collected observation data, update the system's reliability.

[0200] Suppose the collected observation data are: The meaning of the observation data is: time t1 The component (camera) is in working order. The component is in a failed state (vehicle stability does not meet the standard), which belongs to multi-level observation data:

[0201] The credibility update process is as follows:

[0202]

[0203] in: Indicates will As a known condition, it can be obtained from the output of the dynamic Bayesian network credibility evaluation model;

[0204] S W (t) — the predicted state probability of the autonomous driving system at time t, where t>t1.

[0205] Based on the same principles as the methods provided in the embodiments of this application, the embodiments of this application also provide a scenario-based reliability evaluation device for autonomous driving systems, such as... Figure 7 As shown, the device includes:

[0206] The functional test scenario confirmation module 701 is used to randomly combine scenario elements and functional test requirements according to the specific functions to be tested of the autonomous driving system, and generate a functional test scenario library.

[0207] The evaluation index generation module 702 is used to determine the evaluation index of the autonomous driving system and the empirical values ​​of the evaluation index based on the functional testing requirements.

[0208] The system failure rate calculation module 703 is used to perform simulation tests based on the functional test scenario library and index experience values ​​to obtain the failure rate of the autonomous driving system under the evaluation index.

[0209] The model building and training module 704 is used to build a dynamic Bayesian network credibility evaluation model using the failure rate, and to obtain the state probability of the autonomous driving system by using the output of the dynamic Bayesian network credibility evaluation model.

[0210] The system reliability calculation module 705 is used to calculate the reliability of the autonomous driving system and the importance of its components based on the state probability.

[0211] In this embodiment, the functional test scenario confirmation module selects scenario elements and functional test requirements for random combination based on the specific functions to be tested in the autonomous driving system, generating a functional test scenario library; the evaluation index generation module determines the evaluation index of the autonomous driving system and the empirical values ​​of the evaluation index based on the functional test requirements; the system failure rate calculation module performs simulation tests based on the functional test scenario library and the empirical values ​​of the index to obtain the failure rate of the autonomous driving system under the evaluation index; the model building and training module uses the failure rate to build a dynamic Bayesian network credibility evaluation model, and uses the output of the dynamic Bayesian network credibility evaluation model to obtain the state probability of the autonomous driving system; the system credibility calculation module calculates the credibility of the autonomous driving system and the component importance based on the state probability, thereby realizing the credibility evaluation of the autonomous driving system.

[0212] In addition, the embodiments of this application can closely link the testing requirements of autonomous driving systems with the testing scenarios, enrich the performance evaluation indicators of each functional module of the autonomous driving system, reduce the development and maintenance costs of the autonomous driving system, and improve the testing efficiency of the autonomous driving system, thereby improving the safety of autonomous driving technology and promoting the deployment and application of high-level autonomous driving on actual roads.

[0213] In an optional embodiment, the evaluation index generation module 702 includes:

[0214] The first evaluation index generation submodule is used to determine the functional modules in the autonomous driving system based on functional testing requirements.

[0215] The second evaluation index generation submodule is used to determine the components included in the functional module, and to determine the evaluation index and the empirical value of the evaluation index based on the components.

[0216] In an optional embodiment, the system failure rate calculation module 703 includes:

[0217] The first system failure rate calculation submodule is used to determine the empirical value of each evaluation indicator under each functional module;

[0218] The second system failure rate calculation submodule is used to test and verify the empirical values ​​of the indicators, and obtain the optimized empirical values ​​of the indicators as evaluation indicator values.

[0219] The third system failure rate calculation submodule is used to test the autonomous driving system based on the evaluation index value and obtain the failure rate of the autonomous driving system under the evaluation index.

[0220] In an optional embodiment, the model building and training module 704 includes:

[0221] The first model construction training sub-module is used to divide the autonomous driving system into functions and levels, and obtain multiple functional modules and multiple system levels.

[0222] The second model construction training submodule is used to determine the dynamic Bayesian network graph structure based on the functional modules and system hierarchy, and to complete the construction of the dynamic Bayesian network credibility evaluation model.

[0223] The third model construction training submodule is used to calculate the conditional probability table parameters of each node in the dynamic Bayesian network graph structure under various evaluation indicators based on the failure rate; the conditional probability table parameters include the probability values ​​of each node in all possible states.

[0224] The fourth model construction training submodule is used to take the state probability value as the input of the winter Bayesian network credibility evaluation model and obtain the state probability of the autonomous driving system based on the output of the dynamic Bayesian network credibility evaluation model.

[0225] In an optional embodiment, the third model construction training submodule includes:

[0226] The first model builds a training unit, which is used to input the failure rate into the dynamic Bayesian network credibility evaluation model.

[0227] The second model builds a training unit to calculate the working probability of the autonomous driving system under various evaluation indicators;

[0228] The third model builds a training unit, which is used to calculate and output the conditional probability table parameters of each node in the dynamic Bayesian network graph structure for all possible states under various evaluation indicators, based on the working probability.

[0229] In an optional embodiment, the apparatus further includes:

[0230] The credibility update module is used to update the credibility of the autonomous driving system based on the collected observation data.

[0231] In an optional embodiment, the trustworthiness update module includes:

[0232] The first credibility update submodule is used to receive observation data, input the observation data into the dynamic Bayesian network credibility evaluation model, and output the corresponding state probability.

[0233] The second credibility update submodule is used to recalculate the credibility of the autonomous driving system based on the state probability.

[0234] This application provides a scenario-based reliability evaluation device for autonomous driving systems, which can achieve... Figures 1 to 6 The various processes implemented in the method embodiments are not described in detail here to avoid repetition.

[0235] The scenario-based autonomous driving system credibility evaluation device of this application embodiment can execute the scenario-based autonomous driving system credibility evaluation method provided in this application embodiment. The implementation principle is similar. The actions performed by each module and unit in the scenario-based autonomous driving system credibility evaluation device in each embodiment of this application correspond to the steps in the scenario-based autonomous driving system credibility evaluation method in each embodiment of this application. For detailed functional descriptions of each module of the scenario-based autonomous driving system credibility evaluation device, please refer to the descriptions in the corresponding scenario-based autonomous driving system credibility evaluation methods shown above, which will not be repeated here.

[0236] Based on the same principles as the methods shown in the embodiments of this application, this application also provides an electronic device, which may include, but is not limited to: a processor and a memory; the memory for storing computer programs; and the processor for executing the scenario-based autonomous driving system credibility evaluation method shown in any optional embodiment of this application by calling the computer program. Compared with the prior art, the scenario-based autonomous driving system credibility evaluation method provided in this application selects scenario elements and functional test requirements for random combination according to the specific test functions of the autonomous driving system to generate a functional test scenario library; and determines the evaluation indicators of the autonomous driving system and the empirical values ​​of the evaluation indicators according to the functional test requirements; performs simulation tests based on the functional test scenario library and the empirical values ​​of the indicators to obtain the failure rate of the autonomous driving system under the evaluation indicators; constructs a dynamic Bayesian network credibility evaluation model using the failure rate, and obtains the state probability of the autonomous driving system by using the output of the dynamic Bayesian network credibility evaluation model; finally, calculates the credibility and component importance of the autonomous driving system based on the state probabilities, thereby realizing the credibility evaluation of the autonomous driving system.

[0237] In an alternative embodiment, an electronic device, such as Figure 8 As shown, Figure 8 The illustrated electronic device 800 can be a server, including a processor 801 and a memory 803. The processor 801 and the memory 803 are connected, for example, via a bus 802. Optionally, the electronic device 800 may also include a transceiver 804. It should be noted that in practical applications, the transceiver 804 is not limited to one type, and the structure of this electronic device 800 does not constitute a limitation on the embodiments of this application.

[0238] Processor 801 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 801 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0239] Bus 802 may include a pathway for transmitting information between the aforementioned components. Bus 802 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 802 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 8 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0240] The memory 803 may be a ROM (Read Only Memory) or other type of static storage device capable of storing static information and instructions, RAM (Random Access Memory) or other type of dynamic storage device capable of storing information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0241] The memory 803 stores application code that executes the scheme of this application, and its execution is controlled by the processor 801. The processor 801 executes the application code stored in the memory 803 to implement the content shown in the foregoing method embodiments.

[0242] Among them, electronic devices include, but are not limited to: mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and in-vehicle terminals (such as in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 8 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0243] The server provided in this application can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, etc., but is not limited to these. The terminal and server can be directly or indirectly connected via wired or wireless communication, which is not limited herein.

[0244] This application provides a computer-readable storage medium storing a computer program that, when run on a computer, enables the computer to execute the corresponding content in the aforementioned method embodiments.

[0245] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0246] It should be noted that the computer-readable storage medium described above in this application can also be a computer-readable signal medium or a combination of computer-readable storage media and computer-readable storage media. Computer-readable storage media can be, for example,—but not limited to—electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0247] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0248] The aforementioned computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the methods shown in the above embodiments.

[0249] According to one aspect of this application, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform a scenario-based autonomous driving system reliability evaluation method and apparatus provided in the various optional implementations described above.

[0250] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0251] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0252] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the module itself; for example, the functional test scenario confirmation module can also be described as "a functional test scenario confirmation module used to randomly combine scenario elements and functional test requirements based on the specific functions to be tested in the autonomous driving system, and generate a functional test scenario library."

[0253] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A scenario-based reliability evaluation method for autonomous driving systems, characterized in that, The method includes: Based on the specific functions to be tested in the autonomous driving system, scene elements and functional testing requirements are randomly combined to generate a functional test scenario library. Based on the functional testing requirements, the evaluation indicators of the autonomous driving system and the empirical values ​​of the evaluation indicators are determined. Simulation tests are conducted based on the functional test scenario library and the empirical values ​​of the indicators to obtain the failure rate of the autonomous driving system under the evaluation indicators. A dynamic Bayesian network reliability evaluation model is constructed using the failure rate, and the state probabilities of the autonomous driving system are obtained using the output of the dynamic Bayesian network reliability evaluation model. This includes: functionally and hierarchically dividing the autonomous driving system to obtain multiple functional modules and multiple system levels; determining the dynamic Bayesian network graph structure based on the functional modules and system levels to complete the construction of the dynamic Bayesian network reliability evaluation model; calculating the conditional probability table parameters of all possible states of each node in the dynamic Bayesian network graph structure under various evaluation indicators based on the failure rate, including: inputting the failure rate into the dynamic Bayesian network reliability evaluation model; calculating the operating probability of the autonomous driving system under each of the evaluation indicators; calculating and outputting the conditional probability table parameters of all possible states of each node in the dynamic Bayesian network graph structure under various evaluation indicators based on the operating probability; the conditional probability table parameters include the probability values ​​of each node in all possible states; using the probability values ​​as input to the dynamic Bayesian network reliability evaluation model, and obtaining the state probabilities of the autonomous driving system based on the output of the dynamic Bayesian network reliability evaluation model. The reliability and component importance of the autonomous driving system are calculated based on the state probabilities.

2. The method according to claim 1, characterized in that, The scenario-based reliability evaluation method for autonomous driving systems is characterized in that determining the evaluation indicators of the autonomous driving system and the empirical values ​​of the evaluation indicators based on the functional testing requirements includes: Based on the aforementioned functional testing requirements, the functional modules of the autonomous driving system are determined; The components included in the functional module are determined, and evaluation indicators and empirical values ​​of the evaluation indicators are determined based on the components.

3. The scenario-based reliability evaluation method for autonomous driving systems according to claim 2, characterized in that, The simulation test based on the functional test scenario library and the empirical values ​​of the indicators, to obtain the failure rate of the autonomous driving system under the evaluation indicators, includes: Determine the empirical values ​​for each evaluation indicator under each functional module; The empirical values ​​of the indicators are tested and verified to obtain optimized empirical values ​​as evaluation indicator values. The autonomous driving system is tested based on the evaluation index value to obtain the failure rate of the autonomous driving system under the evaluation index.

4. The scenario-based reliability evaluation method for autonomous driving systems according to claim 1, characterized in that, The method further includes: The reliability of the autonomous driving system is updated based on the collected observation data.

5. The scenario-based reliability evaluation method for autonomous driving systems according to claim 4, characterized in that, The process of updating the reliability of the autonomous driving system based on the collected observation data includes: The observation data is received and input into the dynamic Bayesian network credibility evaluation model, and the corresponding state probability is output. The reliability of the autonomous driving system is recalculated based on the state probability.

6. A scenario-based reliability evaluation device for autonomous driving systems, characterized in that, The device includes: The functional test scenario confirmation module is used to randomly combine scenario elements and functional test requirements based on the specific functions to be tested in the autonomous driving system, and generate a functional test scenario library. The evaluation index generation module is used to determine the evaluation index of the autonomous driving system and the empirical values ​​of the evaluation index based on the functional testing requirements. The system failure rate calculation module is used to perform simulation tests based on the functional test scenario library and the index experience values ​​to obtain the failure rate of the autonomous driving system under the evaluation index. The model construction and training module is used to construct a dynamic Bayesian network credibility evaluation model using the failure rate, and to obtain the state probabilities of the autonomous driving system using the output of the dynamic Bayesian network credibility evaluation model. This includes: functionally and hierarchically dividing the autonomous driving system to obtain multiple functional modules and multiple system levels; determining the dynamic Bayesian network graph structure based on the functional modules and system levels to complete the construction of the dynamic Bayesian network credibility evaluation model; and calculating all possible states of each node in the dynamic Bayesian network graph structure under various evaluation indicators based on the failure rate. The conditional probability table parameters include: inputting the failure rate into the dynamic Bayesian network credibility evaluation model; calculating the working probability of the autonomous driving system under each of the evaluation indicators; based on the working probability, calculating and outputting the conditional probability table parameters of all possible states of each node in the dynamic Bayesian network graph structure under each evaluation indicator; the conditional probability table parameters include the probability value of each node in all possible states; using the probability value as the input of the dynamic Bayesian network credibility evaluation model, and obtaining the state probability of the autonomous driving system based on the output of the dynamic Bayesian network credibility evaluation model; The system reliability calculation module is used to calculate the reliability of the autonomous driving system and the importance of its components based on the state probabilities.

7. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the method of any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 5.

Images