Integrating Road Surface Condition Data into OpenDRIVE Models for Autonomous Vehicle Simulations

Abstract. ASAM OpenDRIVE is widely used as a standardized geometric road network description format supporting simulation-based development of advanced driver assistance and autonomous driving systems. While the standard primarily focuses on static road geometry and topology, the increasing need for realistic simulation inputs highlights the importance of incorporating empirically observed pavement conditions into digital road models. This paper proposes a conceptual framework for integrating automatically detected pavement defects into OpenDRIVE without introducing any schema extensions. Road surface anomalies identified from smartphone-based image acquisition using deep neural networks (DNN) are represented as JSON-based intermediate data and transformed by a dedicated mapping module into OpenDRIVE-compliant structures. The approach formalizes the geometric transformation from WGS84 coordinates to the OpenDRIVE curvilinear reference system and encodes defects either as discrete object-level entities or as lane-level material modifications along the road reference line. The main contribution lies in demonstrating that OpenDRIVE, although originally conceived as a geometric exchange format, inherently supports condition-aware road modeling when its existing hierarchical elements are used consistently. The proposed integration pathway establishes a structured link between perception-level data acquisition and simulation-compatible digital road descriptions, thereby contributing to more realistic and context-aware virtual testing environments.