Driving Control Scenario Testing with Iterative State Coverage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing intelligent driving systems face challenges in comprehensive testing of various real-world driving scenarios, requiring complex and manpower-intensive analysis, with existing methods like fuzz testing being inefficient for automated and accurate scenario coverage.
Innovation Solution
A scenario testing method and apparatus that instrument a driving control system program to simulate real scenarios, generate random scenario samples, and iteratively refine test cases based on simulation outputs to achieve high test coverage and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If fuzz testing is used to automate scenario testing, then testing efficiency is improved, but test coverage accuracy deteriorates
Solution Approach 1:
The patent implements feedback mechanisms where test results from simulation are analyzed and used to generate improved test cases for subsequent iterations. The system continuously refines scenario parameters based on coverage analysis, allowing automated testing to progressively improve accuracy while maintaining high efficiency through iteration rather than manual intervention.
Solution Approach 2:
The testing system dynamically adapts scenario parameters and test case generation strategies based on real-time feedback from simulation results. Rather than using static random generation, the system evolves test scenarios iteratively, adjusting parameters to optimize both coverage accuracy and testing efficiency throughout the test process.
2Measurement precision
If manual analysis is used to construct comprehensive test scenarios, then test coverage accuracy is improved, but device complexity and manpower requirements worsen
Solution Approach 1:
The testing system performs self-analysis by automatically constructing state control diagrams from code analysis, generating test scenarios, and evaluating coverage without requiring manual expert intervention. The system serves itself by iteratively improving test cases based on automated feedback, reducing complexity and manpower requirements while maintaining high coverage accuracy.
Solution Approach 2:
The patent segments the complex testing process into distinct automated components: code analysis to extract state control diagrams, scenario generation from diagrams, simulation execution, and coverage evaluation. This segmentation allows each component to be independently optimized and automated, reducing overall system complexity while achieving comprehensive test coverage.
3Reliability
If more driving scenarios are tested, then reliability is improved, but loss of time increases
Solution Approach 1:
The system performs preliminary analysis of the driving control system code to automatically construct state control diagrams before testing begins. By pre-processing the code to extract state transitions and control logic, the system prepares targeted test scenarios in advance, enabling comprehensive reliability testing without proportionally increasing testing time through manual scenario construction.
Solution Approach 2:
The patent implements continuous iterative testing where each test cycle builds on previous results. Rather than performing discrete separate tests, the system continuously refines and re-executes scenarios based on coverage feedback, maximizing the useful action of each testing iteration and achieving comprehensive coverage more efficiently than traditional sequential testing approaches.
Data Source
Figure 1
Figure 2~3
AI summary
The present invention discloses a testing method and apparatus of a driving control system, a medium, and a device. According to the method, a state transition diagram is input, and then the test coverage of each state transition in the state transition diagram is gradually implemented on the basis of the state transition diagram and by means of loop iterative variation of a scenario sample. When the loop iterative variation is performed, a simulation output of the scenario sample is used as a reference, so that the iteration process can quickly converge to the corresponding state transition to achieve the goal of coverage testing. The present invention can intuitively test the processing capability and level of the driving control system for a driving environment, and exhibits high test efficiency.