CarBench
A common benchmark for neural surrogates of 3D car aerodynamics
Paper title CarBench: A Comprehensive Benchmark for Neural Surrogates on High-Fidelity 3D Car Aerodynamics
Paper arXiv 2512.07847PDF
Trains and scores eleven surrogate configurations (a voxel FNO, PointNet, RegDGCNN, PointMAE, PointTransformer, Transolver, Transolver++, AB-UPT, TripNet and larger variants) on the same DrivAerNet++ splits, with bootstrap uncertainties on the metrics, three scoring protocols (on the sampled points, interpolated to the full surface mesh, or predicted node by node over the full mesh), and tests that train on some car body types and predict an unseen one.
- Published
- Communications Engineering
- Data
- DrivAerNet++
Share on LinkedIn, X or Bluesky, or or .
Know related work that belongs here? Suggest a link
Something wrong on this page? Tell the editor