Real-to-Sim Dynamics Fidelity Assessment Framework for High-Fidelity Virtual Vehicles
Year: 2026
Type: conference-paper
Abstract: The validation of autonomous driving systems necessitates safety assessments across extensive driving scenarios, thereby elevating the technical importance of simulation-based testing. However, dynamic discrepancies between virtual environments and physical systems constitute a primary factor undermining the objective reliability of simulation outcomes. To address this issue, this paper proposes a systematic framework for quantitatively verifying the physical fidelity of simulation models based on high-precision real-world driving data. By acquiring field data in compliance with international standard protocols and replicating identical scenarios within the virtual environment, time-series synchronization and comparative analyses were performed on the longitudinal and lateral behavioral characteristics of both real and virtual vehicles. Experimental results demonstrate that the proposed verification scheme empirically substantiates the dynamic reproducibility of virtual vehicle models, thereby contributing to the enhancement of reliability in simulation-based safety assessments for autonomous vehicles.
Institution: Gachon University
Language: English
Subfield: Automotive Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7202293804