Integrated fault injection automatic configuration system and method for automatic driving

Abstract

The invention discloses an integrated fault injection automatic configuration system and method for automatic driving, and belongs to the technical field of automatic driving function safety testing. The system comprises a graphical user interface module for receiving multi-dimensional fault parameters and supporting AUTOSAR signal tree navigation; the template generation module is used for generating a structured template file embedded with safe upper and lower text fields according to the fault parameters; the XML analysis module is used for performing grammar and semantic double-layer analysis, calling a whole vehicle communication matrix to identify a signal dependency relationship, and generating a fault propagation path diagram and an injection matrix data structure; the multi-platform adaptation engine module is used for automatically outputting scripts of platforms such as ETAS INCA, dSPACE Control Desk, Vector CANoe, NI VeriStand and the like; and the script previewing and debugging module renders and executes grammar check in real time, and dynamically adjusts the fault weight and the sorting priority through online learning. According to the system, automatic configuration and multi-platform adaptation of the fault injection test are realized, and the efficiency and accuracy of the safety test of the automatic driving function are remarkably improved.

Description

Technical Field

[0001] This application relates to the field of autonomous driving test equipment, specifically to an integrated fault injection automatic configuration system and method for autonomous driving. Background Technology

[0002] With the rapid development of autonomous driving technology, functional safety testing has become a crucial step in ensuring the reliability of autonomous driving systems. Fault injection testing, as an important means of functional safety verification, verifies the safety response capability of autonomous driving systems under abnormal conditions by artificially introducing various fault modes into the system. However, existing fault injection testing techniques face many challenges in practical applications.

[0003] At present, fault injection testing mainly relies on manual configuration and script writing. Chinese patent application CN116466687A discloses a fault injection method for autonomous driving testing. This method realizes the judgment and execution of fault injection through fault injection configuration file and behavior tree mechanism. Chinese patent application CN121028744A discloses an autonomous driving hardware-in-the-loop (HIL) test fault injection system, which realizes a precise triggering mechanism through dynamic configuration module and fault injection module [2]. Chinese patent application CN113535547B proposes a functional safety-based testing method, which generates fault injection test scripts by establishing a chip security mechanism library and a functional safety test case library. In addition, Chinese patent application CN119269120A discloses a functional testing method, which realizes automatic testing of autonomous driving functions by injecting fault data into the target test environment. Chinese patent application CN113467416A describes a real vehicle functional safety fault injection test method, which realizes fault injection by changing and integrating CAN messages.

[0004] However, existing technologies have the following shortcomings: First, configuration is complex and error-prone; manually editing XML or scripts is time-consuming and laborious, and complex configuration parameters are prone to human error, severely affecting the accuracy and repeatability of testing. Second, there is a lack of multi-platform compatibility; different testing devices such as ETAS, dSPACE, CANoe, and VeriStand use different script formats, resulting in high maintenance costs and difficulty in supporting cross-platform testing needs. Third, the level of automation is low; the lack of a unified parameterized configuration interface and automatic script generation tools leads to long test preparation cycles, severely impacting development efficiency. Finally, injection coverage and effectiveness verification are insufficient; there is a lack of automated detection mechanisms for whether fault injection is effective, and real-time feedback on injection coverage is impossible, affecting the accurate assessment of test quality. These problems restrict the efficiency and quality of autonomous driving functional safety testing, urgently requiring a fault injection matrix configuration and script generation system that can achieve multi-platform compatibility, automated configuration, and intelligent optimization. Summary of the Invention

[0005] To address the technical challenges of complex and error-prone fault injection configuration, lack of multi-platform compatibility, low automation level, and insufficient injection coverage and validity verification in existing autonomous driving functional safety testing, this paper proposes an integrated automatic fault injection configuration system for autonomous driving. This system aims to significantly simplify the fault injection configuration process, improve test configuration efficiency and accuracy, support collaborative work in heterogeneous test environments, and meet the needs of large-scale autonomous driving safety testing.

[0006] The technical problem this application aims to solve is to address the shortcomings of existing fault injection configurations in autonomous driving functional safety testing: configuration is complex and error-prone, manual editing of XML or scripts is time-consuming and laborious, configuration parameters are complex and prone to errors, affecting test accuracy and repeatability; lack of multi-platform compatibility, different script formats for different test devices, high maintenance costs, and difficulty in supporting cross-platform testing needs; low level of automation, lack of a unified parameterized configuration interface and automatic script generation tools, resulting in long test preparation cycles and affecting development efficiency; insufficient injection coverage and validity verification, lack of automated detection of whether injection is effective, inability to provide real-time feedback on injection coverage, affecting test quality assessment.

[0007] The technical solution adopted by this application to solve its technical problem is: In a first aspect, this application provides an integrated fault injection automatic configuration system for autonomous driving, comprising: A graphical user interface module is used to receive multi-dimensional fault parameters, including at least fault type, injection point path, injection mode, parameter range and step size, and supports signal tree navigation and automatic protocol type identification; The template generation module is used to generate a structured fault injection template file that conforms to a predefined schema based on multi-dimensional fault parameters. The template embeds security context fields, associated security target IDs, corresponding security mechanism state transition paths, and upstream driving signals. The XML parsing module is used to perform two-layer parsing of the structured fault injection template file: the syntax layer verifies the schema compliance; the semantic layer calls the vehicle communication matrix to identify the upstream and downstream signal dependencies of the injection point, extracts multi-dimensional fault parameters, generates a fault propagation path diagram, and thus forms the injection matrix data structure. The multi-platform adaptation engine module is used to call the corresponding template generation function based on the injection matrix data structure and the selected target test platform, and automatically output the control script format of the corresponding platform. The script preview and debugging module is used to render the generated script content in real time and perform static syntax checks.

[0008] As a further improvement to this application, the automatic output of the control script format corresponding to the platform includes automatically generating ETAS INCA macro scripts, dSPACE ControlDesk experimental configurations, Vector CANoe CAPL code snippets, and NIVeriStand configurations; wherein: Insert the WaitForSignal() instruction into the INCA script to match the timing constraints of the upstream signal; Enable cascading injection logic in the CANoe script. When the main injection point is triggered, the synchronous injection of downstream related signals will be automatically activated. A virtual ECU consistency comparison mechanism is introduced in ControlDesk and VeriStand. Before exporting, it is loaded into the simulation environment and run to compare the fault effective time deviation ≤5ms, safe state entry delay error ≤10ms, and key signal transition edge jitter <3 sampling cycles under each platform. If the conditions are not met, a correction prompt is returned. All scripts come with a traceable JSON-formatted metadata package.

[0009] As a further improvement to this application, the security context field includes a security target ID, a corresponding FMEA failure mode number, a security status code to be entered, and a minimum response time threshold required by the security mechanism.

[0010] As a further improvement to this application, the script preview and debugging module also learns from historical injection results online and dynamically adjusts the weight and sorting priority of similar faults in subsequent matrices.

[0011] As a further improvement to this application, the online learning of historical injection results and the dynamic adjustment of the weights and sorting priorities of similar faults in subsequent matrices include: The online learning model takes injection point ID, platform type, test priority, whether the injection point field is triggered, response latency, and number of resets as input features, and outputs effective fault weight coefficients to guide the sorting priority of fault combinations in the injection matrix.

[0012] As a further improvement to this application, the template file adopts a unified naming structure, which includes: metadata, fault category, injection point information, injection mode, and parameter definition list.

[0013] As a further improvement to this application, the step of extracting multi-dimensional fault parameters, generating a fault propagation path graph, and then forming an injection matrix data structure includes: During the analysis process, the vehicle communication matrix is ​​invoked to identify the direct upstream drive signals and downstream dependent signals of the injection point, and to construct a fault propagation path diagram. Traverse the parameter space, generate a complete injection matrix data structure based on the fault propagation path graph, and optimize the injection order according to the test priority.

[0014] As a further improvement to this application, the step of extracting multi-dimensional fault parameters, generating a fault propagation path graph, and then forming an injection matrix data structure specifically includes: Read the fault injection points from the XML template, call the propagation path graph G, and automatically identify all associated nodes, including upstream driving signals and downstream dependent signals; determine the path weights and dimensions of the injection matrix based on the associated nodes; The path weights are mapped to parameter constraint rules, and combined constraints are applied to each row of the matrix based on the dimension of the injection matrix. Based on the propagation path graph G, the injection points are divided into levels according to their levels, and a matrix row structure is generated based on the combination constraints. After the matrix row structure undergoes consistency verification, the injected matrix data structure is output.

[0015] As a further improvement to this application, the matrix row structure outputs an injected matrix data structure after consistency verification, including: Load all matrix row structures using a unified virtual ECU model; Under the same operating conditions, after running and comparing, the following conditions must be met: Fault activation time deviation ≤ 5ms; The time error for entering a safe state is ≤ 10ms; The jitter of the critical signal transition edge is less than 3 sampling cycles; If all the above conditions are met, insertion is recommended; otherwise, the row is removed from the matrix, and the propagation graph is re-evaluated. Only rows that pass the validation are included in the final output injection matrix data structure.

[0016] Secondly, this application provides an integrated fault injection automatic configuration method for autonomous driving, including: It can receive multi-dimensional fault parameters, including at least fault type, injection point path, injection mode, parameter range and step size, and supports signal tree navigation and automatic protocol type identification. A structured fault injection template file conforming to a predefined schema is generated based on multidimensional fault parameters. The template embeds security context fields, associated security target IDs, corresponding security mechanism state transition paths, and upstream driving signals. The structured fault injection template file is parsed in two layers: syntax and semantics. The syntax layer verifies the compliance of the schema. The semantic layer calls the vehicle communication matrix to identify the upstream and downstream signal dependencies of the injection point, extracts multi-dimensional fault parameters, generates a fault propagation path diagram, and thus forms the injection matrix data structure. Based on the injection matrix data structure and the selected target test platform, the corresponding template generation function is called to automatically output the control script format for the corresponding platform. Render the generated script content in real time and perform static syntax checks.

[0017] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement an integrated fault injection automatic configuration method for autonomous driving.

[0018] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements an integrated fault injection automatic configuration method for autonomous driving.

[0019] Fifthly, this application provides a computer program product, which includes computer instructions that instruct a computer to execute an integrated fault injection automatic configuration method for autonomous driving.

[0020] The beneficial effects of the technical solution proposed in this application are: This application greatly simplifies the fault injection configuration process by defining a unified XML template and automatically generating multi-platform scripts, avoiding repetitive manual maintenance of multi-platform scripts and improving test configuration efficiency and accuracy. Users can quickly complete complex parameter configurations through an intuitive graphical interface, significantly improving test preparation speed and coverage. The integrated multi-platform script generation function effectively supports collaborative work in heterogeneous test environments and adapts to the diverse needs of autonomous driving functional safety testing. It achieves full-process automation from parameter input to multi-platform script output, significantly improving the efficiency and reliability of functional safety fault injection testing and meeting the needs of large-scale autonomous driving safety testing. The graphical user interface module receives multi-dimensional fault parameters and supports AUTOSAR signal tree navigation and protocol identification, solving the problems of complex and error-prone manual configuration. The template generation module generates structured XML templates with embedded safety context fields, achieving strong binding between configuration data and functional safety requirements (such as safety target ID, FMEA number, safety status code, and response time threshold), improving traceability and compliance. The XML parsing module uses a two-layer parsing mechanism of syntax and semantics, combined with the vehicle communication matrix, to automatically identify upstream and downstream dependencies of injection points and generate fault propagation path diagrams, overcoming the shortcomings of unclear propagation paths and high injection blindness in traditional methods. The multi-platform adaptation engine module automatically generates control scripts that conform to the specifications of each platform based on the injection matrix and target platform, eliminating the problem of repeated development and maintenance of multi-platform scripts. The script preview and debugging module integrates static syntax checking and online learning mechanisms for historical injection results, realizing intelligent sorting and continuous optimization of test cases, thereby significantly improving the configuration efficiency, coverage completeness, cross-platform consistency, and safety verification effectiveness of fault injection testing. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of an integrated fault injection automatic configuration system for autonomous driving provided in this application; Figure 2 A flowchart illustrating the integrated fault injection automatic configuration method for autonomous driving provided in this application; Figure 3 This is a schematic diagram of a visual UI interface provided for an embodiment of this application; Figure 4 A schematic diagram of an electronic device provided in this application. Detailed Implementation

[0022] 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. The step numbers in the following embodiments are set only for ease of explanation, and there is no limitation on the order between the steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0023] In the description of this application, unless otherwise expressly defined, terms such as "setup," "installation," and "connection" should be interpreted broadly, and those skilled in the art can reasonably determine the specific meaning of the above terms in this application in conjunction with the specific content of the technical solution.

[0024] 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. The step numbers in the following embodiments are set only for ease of explanation, and there is no limitation on the order between the steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0025] In the description of this application, unless otherwise expressly defined, terms such as "setup," "installation," and "connection" should be interpreted broadly, and those skilled in the art can reasonably determine the specific meaning of the above terms in this application in conjunction with the specific content of the technical solution.

[0026] In functional safety (FuSa) testing of autonomous driving systems, fault injection is a core method for verifying the effectiveness and robustness of safety mechanisms corresponding to ASIL levels. Currently, test engineers need to manually write XML configuration files or platform-specific scripts for heterogeneous platforms such as ETASINCA, dSPACE ControlDesk, Vector CANoe, and NI VeriStand, involving multiple technical elements such as signal path mapping, timing constraint embedding, safety state transition logic, and upstream / downstream dependency modeling. For example, when verifying the safety objective of "BCU communication timeout triggering L2 level degradation", it is necessary to manually write CAPL scripts in CANoe to simulate CAN message delay, configure WaitForSignal() in INCA to wait for the rising edge of the brake request signal, and set the virtual ECU response threshold and compare the time deviation in VeriStand. The same fault scenario requires the repeated configuration of four sets of logic, and fine-tuning the parameters will cause inconsistencies across platforms. Another example is that when the injection point is located in the Rte layer signal of the AUTOSAR architecture (such as ` / Appl / BrkReq / BrkPedalPos`), different tools have significantly different signal path parsing capabilities. INCA relies on static binding of A2L files, CANoe relies on automatic identification of the DBC database, while ControlDesk requires manual association of Simulink model ports, which leads to injection point positioning errors, misjudgment of upstream drive signals, and breakage of fault propagation paths. Ultimately, the safety mechanism is not effectively activated or is falsely triggered, which seriously weakens the compliance support capability of ISO 26262 Part 6 regarding fault coverage and traceability of safety analysis. The aforementioned problems are mainly manifested in the following ways: fault parameter configuration is highly dependent on human experience, scripts cannot be reused across multiple platforms, signal dependencies lack semantic-level modeling, injection matrix generation lacks safety context guidance, and historical test data fails to feed back into subsequent test strategies. These have become key bottlenecks restricting the large-scale, automated, and auditable testing of autonomous driving functional safety.

[0027] Example 1 like Figure 1 As shown, this application provides an integrated fault injection automatic configuration system for autonomous driving, which includes a graphical user interface module, a template generation module, an XML parsing module, a multi-platform adaptation engine module, and a script preview and debugging module.

[0028] The graphical user interface (GUI) module receives multi-dimensional fault parameters input by the user. These parameters include at least the fault category, injection point path, injection mode, parameter range, and step size. This module supports AUTOSAR signal tree navigation, automatically identifies the protocol type, and provides users with an intuitive parameter configuration interface. Users can select different fault categories through this interface, including hardware faults, software faults, and communication faults, and specify the exact injection point path and corresponding injection mode.

[0029] The template generation module generates a structured fault injection template file conforming to a predefined schema based on the multidimensional fault parameters received from the graphical user interface module. This template file uses a unified naming structure and includes metadata, fault category, injection point information, injection mode, and a parameter definition list. The template embeds a security context field, which includes the security target ID, the corresponding FMEA failure mode number, the expected security state code, and the minimum response time threshold required by the security mechanism. The template also associates the security target ID, the corresponding security mechanism state transition path, and upstream driving signals to ensure effective coordination between the fault injection process and the security mechanism.

[0030] The XML parsing module performs both syntactic and semantic parsing on the structured fault injection template file. At the syntactic level, this module verifies the schema compliance of the template file to ensure correct file format. At the semantic level, this module calls the vehicle communication matrix to identify the upstream and downstream signal dependencies of the injection point and extract multi-dimensional fault parameters. During parsing, the vehicle communication matrix is ​​invoked to identify the direct upstream drive signals and downstream dependent signals of the injection point, constructing a fault propagation path graph. Subsequently, the parameter space is traversed, and a complete injection matrix data structure is generated based on the fault propagation path graph, optimizing the injection order according to test priority.

[0031] The multi-platform adaptation engine module calls the corresponding template generation function based on the injection matrix data structure and the selected target test platform, and automatically outputs the control script format for the corresponding platform.

[0032] As a further improvement, this module supports the automatic generation of ETAS INCA macro scripts, dSPACE ControlDesk experimental configurations, Vector CANoe CAPL code snippets, and NI VeriStand configurations. The `WaitForSignal()` instruction is inserted into the INCA script to match upstream signal timing constraints, ensuring the accuracy of fault injection timing. Cascaded injection logic is enabled in the CANoe script; when the main injection point is triggered, the synchronous injection of downstream related signals is automatically activated, achieving multi-point collaborative fault injection. A virtual ECU consistency comparison mechanism is introduced in ControlDesk and VeriStand. Before exporting, the ECU is loaded into the simulation environment and run, comparing the fault activation time deviation (≤5ms), safe state entry delay error (≤10ms), and critical signal transition edge jitter (<3 sampling periods) across platforms. If these conditions are not met, a correction prompt is returned. All scripts include traceable JSON format metadata packages for easy tracking and analysis of subsequent test results.

[0033] The script preview and debugging module renders and generates script content in real time and performs static syntax checks, providing users with instant script preview functionality. This module learns online from historical injection results, dynamically adjusting the weights and ranking priorities of similar faults in subsequent matrices. The online learning model takes injection point ID, platform type, test priority, whether the injection point field is triggered, response latency, and number of resets as input features, and outputs effective fault weight coefficients to guide the ranking priority of fault combinations in the injection matrix. Through machine learning algorithms, the fault injection strategy is continuously optimized, improving testing efficiency and fault detection capabilities.

[0034] The system works as follows: First, the user inputs fault injection parameters through the graphical user interface module, and the system generates a standardized XML template file based on these parameters. Then, the XML parsing module parses the template file, constructs a fault propagation path diagram using the vehicle communication matrix, and generates the injection matrix data structure. Next, the multi-platform adaptation engine module converts the injection matrix into control scripts for the target platform selected by the user. Finally, the script preview and debugging module provides script preview and debugging functions and continuously optimizes the fault injection strategy through online learning.

[0035] Through this design, the system can automatically generate fault injection scripts across platforms, greatly improving the efficiency and standardization of autonomous driving functional safety testing. At the same time, through an intelligent learning mechanism, it continuously optimizes testing strategies, providing strong support for the safety verification of autonomous driving systems.

[0036] This embodiment provides an integrated and scalable automated fault injection configuration system. It achieves end-to-end transformation from security requirements to executable test scripts by constructing a closed-loop technical chain encompassing graphical interaction, structured modeling, semantic parsing, multi-platform mapping, and intelligent feedback. Its core lies in using standardized XML templates as an intermediary to decouple and organically unify abstract security goals, specific signal paths, dynamic parameter spaces, and heterogeneous platform interfaces. This frees the fault injection process from strong reliance on engineers' platform experience, instead relying on formal rules and data-driven logic to complete configuration generation and continuous optimization.

[0037] This application relates to automated configuration and multi-platform adaptation technology for fault injection in functional safety (FuSa) testing of autonomous driving. Specifically, it automatically generates a fault injection matrix based on a standardized fault injection template and supports the generation of control scripts for multiple mainstream test platforms such as ETAS INCA, dSPACE ControlDesk, Vector CANoe, and NI VeriStand.

[0038] Based on the above embodiments, the present application's solution extracts multi-dimensional fault parameters, generates a fault propagation path diagram, and then forms an injection matrix data structure; specifically, it includes: calling the vehicle communication matrix during the parsing process, identifying the direct upstream drive signal and downstream dependent signal of the injection point, and constructing a fault propagation path diagram; traversing the parameter space, generating a complete injection matrix data structure based on the fault propagation path diagram, and optimizing the injection order according to the test priority.

[0039] Furthermore, this embodiment provides the extraction of multi-dimensional fault parameters, generation of a fault propagation path graph, and subsequent formation of an injection matrix data structure, specifically including: Step 1: Path Graph Parsing and Signal Dimension Extraction: The system reads the fault injection point from the XML template, calls the propagation path graph G, and automatically identifies all related nodes. Upstream drive signal: VSC_ECU::BrakePressure (corresponding edge weight 0.92); Downstream dependent signals: ESP_ECU::BrakeTorque (weight 0.95), HVAC_ECU::FanSpeed ​​(weight 0.62), TBOX::EventLog (weight 0.38).

[0040] Therefore, the number of columns in the injection matrix is ​​determined to be 4 (BrakeCmd, BrakePressure, BrakeTorque, FanSpeed), meaning that the matrix dimension is dynamically determined by G, rather than manually set.

[0041] Step 2: Mapping path weights to parameter constraint rules: Apply combined constraints to each row of the matrix based on the SafetyWeight and TimingWeight of each edge: If the weight of a certain side is ≥0.85 (such as from BrakeCmd to BrakeTorque), then the two columns must be strongly coupled and injected (that is, when BrakeCmd=0, BrakeTorque must be set to 0 or ≤0.1N·m simultaneously). If the weight ∈ [0.6, 0.85), it is defined as weak coupling injection (e.g., when BrakeCmd=0, FanSpeed ​​randomly perturbs within ±20%). If the weight is less than 0.6, it will not be included in the matrix and will only be used for propagation coverage statistics.

[0042] Step 3: Generate row vectors by stratified sampling according to ASIL level; based on the propagation path graph G, divide the injection points into levels according to ASIL level to generate a matrix row structure, as shown in Table 1: Table 1

[0043] The total number of rows in the injection matrix = Σ (weight of each path × ASIL weight coefficient). In actual testing, the matrix size of a certain model was reduced from 128 rows to 27 rows, improving efficiency by 79%, and the defect detection rate increased by 15%.

[0044] Step 4: Virtual ECU Consistency Verification Embedded Matrix Generation Process: After each row of injection combination is generated, it is immediately loaded into the Simulink Real-Time Virtual ECU for pre-execution verification: If BrakeCmd=0 but the injection point field _0x1A is not embedded in the 200ms, then the row is marked as "needs correction" and insertion is recommended; If BrakeTorque does not respond, the row is removed from the matrix, and the propagation graph is re-evaluated; only rows that pass the validation are included in the final output matrix.

[0045] Specifically, all scripts are loaded using the unified virtual ECU model (Simulink Real-Time); Input the same operating conditions (e.g., ACC on, vehicle speed 60km / h), and compare the results after running: Fault activation time deviation ≤ 5ms; The time error for entering a safe state is ≤ 10ms; The jitter of the critical signal transition edge is less than 3 sampling cycles.

[0046] This application proposes to perform closed-loop verification based on a virtual ECU; the output result is a consistency report, and when it is below 95%, manual review is prompted, which is a technical means of quality control in the testing process.

[0047] The vehicle communication matrix provides a system-level signal topology benchmark, ensuring the physical authenticity of the propagation path diagram. Path diagram modeling elevates isolated signal faults to cross-ECU functional link faults, preventing parameter traversal from remaining at single-point disturbances. Parameter space pruning and priority sorting based on the path diagram converges massive mathematical combinations into a minimal, complete test set with engineering significance, ensuring the injection matrix reflects both the real system coupling relationships and meets the dual constraints of efficiency and coverage in functional safety testing. Through these technical solutions, this application achieves semantic dimensionality enhancement from configuration parameters to system-level fault impacts during the XML parsing stage. By calling the vehicle communication matrix to identify upstream and downstream signal dependencies, it solves the problems of experience-based judgment of fault propagation path dependencies and the potential for overlooking implicit dependencies leading to blind spots in test coverage caused by traditional manual configuration. By generating the injection matrix based on the constructed fault propagation path diagram through the parameter space, it avoids the waste of test resources caused by indiscriminate permutations, ensuring each injection instance corresponds to a verifiable fault propagation link. By optimizing the injection order according to test priority, high-risk paths are executed first, significantly improving defect discovery rate and test return on investment. Ultimately, this approach aims to systematically improve the structure, comprehensiveness, and cost-effectiveness of fault injection testing for autonomous driving functions without increasing hardware costs.

[0048] Example 2 like Figure 1 As shown, this application provides an integrated automatic fault injection configuration system, including: a graphical user interface module, a template generation module, an XML parsing module, a multi-platform adaptation engine module, and a script preview and debugging module.

[0049] The parameter input module, based on a graphical user interface, enables the input and management of fault template ID, description, injection category, injection point, injection mode, and multi-dimensional parameter range; the specific interface can be found in [link to interface]. Figure 3 As shown.

[0050] The graphical user interface is used to input multidimensional parameters of the fault injection template (including injection type, injection point, parameter range, and injection method, etc.); by inputting fault parameters through a user-friendly GUI, the configuration threshold is greatly reduced and the configuration efficiency and accuracy are improved.

[0051] The XML template generation module automatically generates standardized fault injection template files that conform to a predefined XML Schema based on the parameter input; it automatically generates fault injection template files that conform to a unified XML standard to ensure that the configuration data is standardized and structured; the unified standardized fault injection XML template design uses a custom XML Schema to define fault injection parameters, realizing parameter structuring and standardization, which is convenient for management and automatic parsing.

[0052] Furthermore, the XML template files use a unified namespace and have the following structure: 1) Meta-information (TemplateID, description, etc.); 2) Fault category (hardware, software, communication); 3) Injection point information (ID bound to system architecture); 4) Injection modes (power outage, signal drift, data loss, etc.); 5) Parameter definition list (each parameter includes name, type, value range, step size, etc.).

[0053] For example, in CANoe, test XML scripts are written in the manner described above. Taking the power failure signal as an example, the FaultCategory field in the XML file gives the power failure signal as BCM_1_KeySts==0. When migrating to other platforms, the power failure signal changes. At this time, you only need to change the signal name and parameter value of the power failure signal.

[0054] XML parsing module: Based on an XML parser, it reads and parses template files in real time, extracts injection points and parameter information, and forms an injection matrix data structure; the built-in XML parsing module parses template content in real time and extracts the information required for the injection matrix. Among them, the injection matrix generation method supports single-variable scanning, full permutation combination and orthogonal design, which can meet different testing needs and resource constraints, and improve testing efficiency and effectiveness.

[0055] Multi-platform adaptation module: Based on the injection matrix and the selected target test platform, the module calls the corresponding template generation function to automatically output the control script format for the corresponding platform; the multi-platform adaptation module automatically generates fault injection control scripts adapted to platforms such as ETACICA / LAB-5850, dSPACE ControlDesk, Vector CANoe, and NI VeriStand based on the parsing results. Script display and export: Enables previewing, editing, and file export of generated scripts, supporting common script file formats; The user interface provides an interactive interface that supports template loading, parameter editing, platform selection, and script generation. It supports script preview, saving, and exporting, improving script writing efficiency and reducing human error rates. The multi-platform automatic script generation engine incorporates script templates from multiple platforms, including ETAS INCA, dSPACE ControlDesk, Vector CANoe, and NI VeriStand, enabling one-click conversion and seamless integration with various hardware platforms. The multi-platform automatic script generation engine also supports online debugging and verification command generation for different platforms, ensuring the controllability and accuracy of the injection process.

[0056] This application achieves full-process automation from parameter input to multi-platform script output, significantly improving the efficiency and reliability of functional safety fault injection testing and meeting the needs of large-scale autonomous driving safety testing.

[0057] Furthermore, this application also automatically injects coverage detection, integrates log comparison and analysis tools, automatically determines whether the injection is effective, and marks uncovered or abnormal injection items in real time to ensure the integrity of test coverage.

[0058] Furthermore, by traversing the parameter space, a complete injection matrix is ​​generated based on the input parameter range, step size, and injection combination strategy (such as single variable, full combination, orthogonal design, etc.), and the injection order is optimized according to the test priority (ASIL level).

[0059] Furthermore, corresponding script generation logic is implemented for different testing platforms, including: 1) ETAS INCA macro script format; 2) dSPACE ControlDesk experimental script format; 3) Vector CANoe CAPL script fragment; and 4) NIVeriStand fault injection configuration script. The generated script syntax conforms to the interface specifications of each platform and supports parameter replacement and dynamic injection instructions.

[0060] In this embodiment, during the script preview and export steps, the generated script content is displayed in real time in the text area of ​​the graphical interface, supporting text editing and direct export and saving, which facilitates user verification and subsequent use.

[0061] Therefore, the implementation method of this embodiment is to extract relevant information fields and their parameter values ​​from the XML file (e.g., information fields, and the parameter in this information field is the TCS failure signal), and then extract the TCS failure signal (different signal) of the same field from the new platform.

[0062] This application greatly simplifies the fault injection configuration process by defining a unified XML template and automatically generating multi-platform scripts, avoiding repetitive manual maintenance of multi-platform scripts and improving test configuration efficiency and accuracy. Users can quickly complete complex parameter configurations through an intuitive graphical interface, significantly improving test preparation speed and coverage. The integrated multi-platform script generation function effectively supports collaborative work in heterogeneous testing environments, adapting to the diverse needs of autonomous driving functional safety testing.

[0063] like Figure 3 As shown, this application also provides an integrated fault injection automatic configuration method for autonomous driving, characterized by comprising: It can receive multi-dimensional fault parameters, including at least fault type, injection point path, injection mode, parameter range and step size, and supports signal tree navigation and automatic protocol type identification. A structured fault injection template file conforming to a predefined schema is generated based on multidimensional fault parameters. The template embeds security context fields, associated security target IDs, corresponding security mechanism state transition paths, and upstream driving signals. The structured fault injection template file is parsed in two layers: syntax and semantics. The syntax layer verifies the compliance of the schema. The semantic layer calls the vehicle communication matrix to identify the upstream and downstream signal dependencies of the injection point, extracts multi-dimensional fault parameters, generates a fault propagation path diagram, and thus forms the injection matrix data structure. Based on the injection matrix data structure and the selected target test platform, the corresponding template generation function is called to automatically output the control script format for the corresponding platform. Render the generated script content in real time and perform static syntax checks.

[0064] like Figure 4 As shown, a third objective of this application embodiment is to provide 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 computer program, it implements the aforementioned integrated fault injection automatic configuration method for autonomous driving. It also includes a communication interface and a bus.

[0065] The fourth objective of this application is to provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned integrated fault injection automatic configuration method for autonomous driving.

[0066] A fifth objective of this application is to provide a computer program product comprising computer instructions that instruct a computer to execute the aforementioned integrated fault injection automatic configuration method for autonomous driving.

[0067] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0068] These computer program instructions can also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0069] This application may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application may take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, readable storage media, optical storage, etc.) containing computer-usable program code.

[0070] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0071] Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this application.

[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application and not to limit them. Although this application has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation methods of this application. Any modifications or equivalent substitutions that do not depart from the spirit and scope of this application should be covered within the protection scope of this application.

Claims

1. An integrated fault injection automatic configuration system for autonomous driving, characterized in that, include: A graphical user interface module is used to receive multi-dimensional fault parameters, including at least fault type, injection point path, injection mode, parameter range and step size, and supports signal tree navigation and automatic protocol type identification; The template generation module is used to generate a structured fault injection template file that conforms to a predefined schema based on multi-dimensional fault parameters. The template embeds security context fields, associated security target IDs, corresponding security mechanism state transition paths, and upstream driving signals. The XML parsing module is used to perform two-layer parsing of the structured fault injection template file: the syntax layer verifies the schema compliance. The semantic layer calls the vehicle communication matrix to identify the upstream and downstream signal dependencies of the injection point, extracts multi-dimensional fault parameters, generates a fault propagation path diagram, and then forms the injection matrix data structure. The multi-platform adaptation engine module is used to call the corresponding template generation function based on the injection matrix data structure and the selected target test platform, and automatically output the control script format of the corresponding platform. The script preview and debugging module is used to render the generated script content in real time and perform static syntax checks.

2. The integrated fault injection automatic configuration system for autonomous driving according to claim 1, characterized in that, The automatic output of the control script format corresponding to the platform includes the automatic generation of ETAS INCA macro scripts, dSPACEControlDesk experimental configurations, Vector CANoe CAPL code snippets, and NI VeriStand configurations; wherein: Insert the WaitForSignal() instruction into the INCA script to match the timing constraints of the upstream signal; Enable cascading injection logic in the CANoe script. When the main injection point is triggered, the synchronous injection of downstream related signals will be automatically activated. A virtual ECU consistency comparison mechanism is introduced in ControlDesk and VeriStand. Before exporting, it is loaded into the simulation environment and run to compare the fault effective time deviation ≤5ms, safe state entry delay error ≤10ms, and key signal transition edge jitter <3 sampling cycles under each platform. If the conditions are not met, a correction prompt is returned. All scripts come with a traceable JSON-formatted metadata package.

3. The integrated fault injection automatic configuration system for autonomous driving according to claim 1, characterized in that, The security context field includes the security target ID, the corresponding FMEA failure mode number, the expected security status code, and the minimum response time threshold required by the security mechanism.

4. The integrated fault injection automatic configuration system for autonomous driving according to claim 1, characterized in that, The script preview and debugging module also learns from historical injection results online and dynamically adjusts the weight and sorting priority of similar faults in subsequent matrices.

5. An integrated fault injection automatic configuration system for autonomous driving according to claim 4, characterized in that, The online learning of historical injection results to dynamically adjust the weights and sorting priorities of similar faults in subsequent matrices includes: The online learning model takes injection point ID, platform type, test priority, whether the injection point field is triggered, response latency, and number of resets as input features, and outputs effective fault weight coefficients to guide the sorting priority of fault combinations in the injection matrix.

6. An integrated fault injection automatic configuration system for autonomous driving according to claim 1, characterized in that, The template file adopts a unified naming structure, which includes: metadata, fault category, injection point information, injection mode, and parameter definition list.

7. An integrated fault injection automatic configuration system for autonomous driving according to claim 1, characterized in that, The process of extracting multi-dimensional fault parameters, generating a fault propagation path graph, and then forming an injection matrix data structure includes: During the analysis process, the vehicle communication matrix is ​​invoked to identify the direct upstream drive signals and downstream dependent signals of the injection point, and to construct a fault propagation path diagram. Traverse the parameter space, generate a complete injection matrix data structure based on the fault propagation path graph, and optimize the injection order according to the test priority.

8. An integrated fault injection automatic configuration system for autonomous driving according to claim 1, characterized in that, The process of extracting multi-dimensional fault parameters, generating a fault propagation path graph, and then forming an injection matrix data structure specifically includes: Read the fault injection points from the XML template, call the propagation path graph G, and automatically identify all associated nodes, including upstream driving signals and downstream dependent signals; determine the path weights and dimensions of the injection matrix based on the associated nodes; The path weights are mapped to parameter constraint rules, and combined constraints are applied to each row of the matrix based on the dimension of the injection matrix. Based on the propagation path graph G, the injection points are divided into levels according to their levels, and a matrix row structure is generated based on the combination constraints. After the matrix row structure undergoes consistency verification, the injected matrix data structure is output.

9. An integrated fault injection automatic configuration system for autonomous driving according to claim 8, characterized in that, After consistency verification, the matrix row structure outputs the injected matrix data structure, including: Load all matrix row structures using a unified virtual ECU model; Under the same operating conditions, after running and comparing, the following conditions must be met: Fault activation time deviation ≤ 5ms; The time error for entering a safe state is ≤ 10ms; The jitter of the critical signal transition edge is less than 3 sampling periods; If all the above conditions are met, insertion is recommended; otherwise, the row is removed from the matrix, and the propagation graph is re-evaluated. Only rows that pass the validation are included in the final output injection matrix data structure.

10. An integrated fault injection automatic configuration method for autonomous driving, characterized in that, include: It can receive multi-dimensional fault parameters, including at least fault type, injection point path, injection mode, parameter range and step size, and supports signal tree navigation and automatic protocol type identification. A structured fault injection template file conforming to a predefined schema is generated based on multidimensional fault parameters. The template embeds security context fields, associated security target IDs, corresponding security mechanism state transition paths, and upstream driving signals. The structured fault injection template file undergoes a two-tiered parsing process: syntax and semantics. The syntax layer verifies schema compliance. The semantic layer calls the vehicle communication matrix to identify the upstream and downstream signal dependencies of the injection point, extracts multi-dimensional fault parameters, generates a fault propagation path diagram, and then forms the injection matrix data structure. Based on the injection matrix data structure and the selected target test platform, the corresponding template generation function is called to automatically output the control script format for the corresponding platform. Render the generated script content in real time and perform static syntax checks.

Images