Simulation Surrogate Model for Traffic Scenario Validation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Simulating complex traffic scenarios in a resource-efficient and short-term manner, particularly for validating driver assistance systems and autonomous driving algorithms, is challenging due to the need for numerous simulation runs that are resource- and time-intensive.
Innovation Solution
A computer-implemented method generates change proposals for simulation runs by predicting results using a substitute model, such as a neural network, trained on completed runs to identify unnecessary or redundant simulations, allowing early termination or modification of not yet completed simulations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a large number of simulation runs are performed to cover the parameter space of complex traffic scenarios, then the completeness and reliability of validation results is improved, but the resource consumption and time required increases significantly
Solution Approach 1:
The system performs preliminary simulation runs to collect training data before the main validation phase. A surrogate model is trained in advance on this preliminary data, enabling it to predict outcomes of future simulation runs. This preliminary action allows the system to skip unnecessary full simulation runs later, significantly reducing total validation time while maintaining reliability through the surrogate model's predictions.
Solution Approach 2:
The system creates a surrogate model that copies the essential behavior and output characteristics of the complex simulation system. This surrogate model serves as a lightweight replica that can quickly predict simulation outcomes without requiring full simulation execution. The copying approach enables rapid evaluation of parameter space while preserving the reliability needed for validation.
2Measurement precision
If a large number of simulation runs are performed to cover the parameter space, then the accuracy and comprehensiveness of test coverage is improved, but the computational resources and cost increase
Solution Approach 1:
The system performs only the necessary subset of simulation runs required to achieve adequate validation coverage, rather than exhaustively running all possible parameter combinations. The surrogate model identifies which simulation runs provide valuable new information and which are redundant, enabling partial action that maintains measurement precision while reducing computational resource consumption.
Solution Approach 2:
The surrogate model acts as a computational copy that approximates the behavior of full simulation runs. Instead of repeatedly executing resource-intensive simulations, the system uses the surrogate model to predict outcomes, dramatically reducing energy and computational resource usage while maintaining sufficient test coverage accuracy for validation purposes.
3Productivity
If the simulation parameter space is reduced to save resources and time, then the efficiency of validation is improved, but the completeness of test coverage deteriorates
Solution Approach 1:
The system implements feedback loops where simulation results are continuously fed back to update the surrogate model and inform subsequent simulation run selections. This feedback mechanism ensures that the surrogate model learns from actual simulation outcomes and improves its predictions, allowing the system to maintain test coverage completeness even with reduced parameter space exploration by focusing computational resources on the most informative regions.
Solution Approach 2:
The system performs preliminary simulation runs to establish a foundation of training data before the main validation phase. This preliminary action enables the surrogate model to be pre-trained on representative samples, ensuring it can accurately predict outcomes in the reduced parameter space. The preliminary exploration preserves reliability by ensuring the surrogate model is trained on diverse, representative data before efficiency optimizations are applied.
Data Source
Figure 1
AI summary
The invention relates to a computer-implemented method for generating a change proposal (10) for at least one simulation run (12) of a simulation (14) comprising a plurality of simulation runs (12), comprising the steps of - receiving one or more simulation run results (16) of one or more completed simulation runs (12) of the simulation (14), - predicting at least one simulation run result (20) of a not yet completed simulation run (12) of the simulation (14) taking into account the received simulation run results (16), and - generating the change proposal (10) for not yet completed simulation runs (12) of the simulation (14), taking into account the at least one predicted simulation run result (20).Furthermore, the invention relates to a device for data processing comprising means for carrying out the above method, a computer program product and a corresponding computer-readable data carrier.