Abstract
This paper presents the first application of Gaussian Mixture Copula Modelsto the statistical modeling of driving scenarios for the safety validation ofautomated driving systems. Knowledge of the joint probability distribution ofscenario parameters is essential for scenario-based safety assessment, whererisk quantification depends on the likelihood of concrete parametercombinations. Gaussian Mixture Copula Models bring together the multimodalexpressivity of Gaussian Mixture Models and the flexibility of copulas,enabling separate modeling of marginal distributions and dependencies. Webenchmark Gaussian Mixture Copula Models against previously proposed approaches- Gaussian Mixture Models and Gaussian Copula Models - using real-world drivingdata drawn from scenarios defined in United Nations Regulation No. 157. Ourevaluation across 18 million scenario instances demonstrates that GaussianMixture Copula Models provide a better fit to the data in terms of bothlikelihood and Sinkhorn distance. These results suggest that Gaussian MixtureCopula Models are a compelling foundation for future scenario-based validationframeworks.