The automotive industry is leaning heavily into virtual testing to speed up vehicle design and cut physical prototyping costs. However, checking if a simulation matches a real car still relies on older, rigid math formulae and manual checks. This makes virtual testing less reliable for highly dynamic driving maneuvers, required in vehicle design. To address this, following paper introduces an automated software framework built to accurately measure the differences between physical test data and virtual simulations.
The methodology was validated using a lightweight off-road prototype at the National Automotive Test Tracks (NATRAX) on a custom-engineered track inducing high-fidelity roll, pitch, and yaw excitations. To guarantee accuracy, an identical digital twin of this complex topography was constructed within a high-fidelity vehicle-dynamics simulation environment. The system features an algorithmic engine that automatically reads, cleans, and synchronizes distinct physical and virtual telemetry streams. Initial validation using standard metrics like NRMSE, RMSE, and S&G proved insufficient, failing to quantify errors in data exhibiting non-linear temporal shifts caused by stochastic human driver latencies. To overcome these limitations this study applies Windowed Dynamic Time Warping (DTW) to operationalize the ISO/TS 18571:2024 standard.
This approach leverages the optimal warping path to extract continuous, vector-based error profiles, isolating phase and amplitude inconsistencies across the entire event. This automated framework significantly reduces correlation time by transitioning from subjective scalar scores to high-resolution diagnostic error vectors. By quantifying dynamic fidelity through this automated, high-precision DTW computational tool, the methodology validates virtual models as robust predictive tools, allowing simulation data to yield results comparable to physical benchmarks.
This capability enables faster vehicle development by allowing rapid evaluation of multiple parameters in a virtual environment. It supports performance fine-tuning with reduced physical testing and can be integrated into automated workflows for repeatable data correlation, improving validation efficiency and reducing time.