As driving automation systems become increasingly integrated into modern on-road
vehicles, ensuring their safety through comprehensive testing is critical. Virtual Testing
Environments (VTEs) have emerged as a promising cost-efficient, scalable and safe
augmentation to traditional test-track and naturalistic testing to create the test coverage
needed to prove these systems’ safety case. These environments incorporate both dynamic
and static elements, making their accuracy in replicating the Operational Design Domain
(ODD) vital to meaningful evaluations.
To ensure the reliability of VTEs, it is essential to closely replicate the ODD and quantify
measurable deviations. Despite advancements in VTE platforms, a robust, geometry-aware
metric for assessing how well these environments represent the real world has yet to be
established. In this work, we present preliminary results of a graph-based approach,
alongside Chamfer distance, as a novel metric to evaluate the reliability of VTEs in faithfully
representing real-world conditions.