DragSolver: A Multi-Scale Transformer for Real-World Automotive Drag Coefficient Estimation

Ye Liu, Yuntian Chen

TL;DR: We introduce a multi-scale Transformer approach for real-time automotive drag coefficient estimation, handling diverse 3D car meshes, and cutting error by 58.7% over prior methods.

Abstract:

Automotive drag coefficient ($C_d$) is pivotal to energy efficiency, fuel consumption, and aerodynamic performance. However, costly computational fluid dynamics (CFD) simulations and wind tunnel tests struggle to meet the rapid-iteration demands of automotive design. We present DragSolver, a Transformer-based framework for rapid and accurate $C_d$ estimation from large-scale, diverse 3D vehicle models.
DragSolver tackles four key real-world challenges:
(1) multi-scale feature extraction to capture both global shape and fine local geometry;
(2) heterogeneous scale normalization to handle meshes with varying sizes and densities;
(3) surface-guided gating to suppress internal structures irrelevant to external aerodynamics;
and (4) epistemic uncertainty estimation via Monte Carlo dropout for risk-aware design.
Extensive evaluations on three industrial-scale datasets (DrivaerNet, DrivaerNet++, and DrivaerML) show that DragSolver outperforms existing approaches in accuracy and generalization, achieving an average reduction of relative $L_2$ error by 58.7% across real-world datasets. Crucially, DragSolver is the first to achieve reliable, real-time $C_d$ inference on production-level automotive geometries.

Lay Summary: Cars and trucks waste a surprising amount of energy just by pushing air aside. Engineers capture this “air-slipperiness” with a single number—the drag coefficient (C_d). Each minor shape tweak has traditionally required hours-long computer-fluid simulations or costly wind-tunnel tests, bottlenecking the design loop. DragSolver is an AI system that learns the link between a vehicle’s 3-D surface and its drag coefficient from thousands of past designs. It examines shapes at two scales simultaneously: the global silhouette that steers the bulk airflow and the fine ridges and gaps that disturb it. The model automatically ignores interior parts that do not affect outside aerodynamics and flags when its own prediction may be unreliable. Across three industry-scale datasets, DragSolver cuts prediction error by about 50 % compared with previous methods while running in real time on a standard GPU. Designers can therefore evaluate far more concepts per day and ultimately build vehicles that travel farther on the same fuel or battery charge.

Application-Driven Machine Learning: This submission is on Application-Driven Machine Learning.

Primary Area: Applications->Chemistry, Physics, and Earth Sciences

Keywords: Aerodynamic Drag Prediction, Computational Fluid Dynamics, Transformer, Real-World Automobiles

Submission Number: 4352