Stadler C, German R, Djanatliev A (2022)
Publication Language: English
Publication Type: Conference contribution, Original article
Publication year: 2022
Pages Range: 2508-2513
Conference Proceedings Title: IEEE 25th International Conference on Intelligent Transportation Systems (ITSC)
DOI: 10.1109/ITSC55140.2022.9922082
Due to the immense testing effort, the development and release of automated driving functions is increasingly dependent on simulations in addition to real world testing. In order to trust the simulation results, the deviations have to be quantified compared to tests on real roads. That is a prerequisite to decide if the required simulation fidelity is achieved or not. Therefore, several single quantitative and qualitative metrics are combined in this work to a multivariate scoring value for scenario-based testing in virtual environments. The approach focuses on deviations of the driven trajectories as well as the conducted maneuvers and the criticality in the traffic situations. However, the built framework for quantifying the gap compared to real world testing is generally formulated to be extendable for further features. Furthermore, scoring levels are introduced for classification. The practicable usage of the defined metric is demonstrated based on a resimulation of a scenario extracted from real driving data. The results are classified and thus the deviations are interpretable according to the used weighting factors.
APA:
Stadler, C., German, R., & Djanatliev, A. (2022). Towards a Multivariate Scoring Metric for the Evaluation of Simulation Scenarios in Automated Driving. In IEEE 25th International Conference on Intelligent Transportation Systems (ITSC) (pp. 2508-2513).
MLA:
Stadler, Christoph, Reinhard German, and Anatoli Djanatliev. "Towards a Multivariate Scoring Metric for the Evaluation of Simulation Scenarios in Automated Driving." Proceedings of the IEEE 25th International Conference on Intelligent Transportation Systems (ITSC) 2022. 2508-2513.
BibTeX: Download