DrivAerNet++: A Large-Scale Multimodal Car Dataset with Computational Fluid Dynamics Simulations and Deep Learning Benchmarks

2024-06-13

Neural Information Processing Systems

AI Summary

A critical bottleneck in automotive aerodynamics research is the lack of large-scale, multimodal benchmark datasets. Method: We introduce AutoAero—the largest open-source multimodal dataset to date (39 TB)—comprising 8,000 diverse vehicle body configurations (e.g., fastback, notchback), systematically incorporating design distinctions between electric and internal-combustion powertrains (e.g., underbody and wheel-arch geometry). It provides high-fidelity CFD simulation outputs (aerodynamic coefficients, volumetric flow fields, surface fields), parametric CAD models, 3D meshes, part-level semantic segmentation, and annotated point clouds. Contribution/Results: AutoAero enables cross-task learning—including generative modeling, surrogate model training, and CFD acceleration. We establish a baseline for drag coefficient prediction, empirically verifying physical consistency and geometric diversity. Experiments demonstrate significantly improved model generalization and design optimization efficiency. All data and code are publicly released.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Application Domains: Mobility, Driving & Flight
• Computer Vision: Multi-modal Vision

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Systems and Infrastructure for Web, Mobile and WoT: Web performance, measurement, and characterization
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

We present DrivAerNet++, the largest and most comprehensive multimodal dataset for aerodynamic car design. DrivAerNet++ comprises 8,000 diverse car designs modeled with high-fidelity computational fluid dynamics (CFD) simulations. The dataset includes diverse car configurations such as fastback, notchback, and estateback, with different underbody and wheel designs to represent both internal combustion engines and electric vehicles. Each entry in the dataset features detailed 3D meshes, parametric models, aerodynamic coefficients, and extensive flow and surface field data, along with segmented parts for car classification and point cloud data. This dataset supports a wide array of machine learning applications including data-driven design optimization, generative modeling, surrogate model training, CFD simulation acceleration, and geometric classification. With more than 39 TB of publicly available engineering data, DrivAerNet++ fills a significant gap in available resources, providing high-quality, diverse data to enhance model training, promote generalization, and accelerate automotive design processes. Along with rigorous dataset validation, we also provide ML benchmarking results on the task of aerodynamic drag prediction, showcasing the breadth of applications supported by our dataset. This dataset is set to significantly impact automotive design and broader engineering disciplines by fostering innovation and improving the fidelity of aerodynamic evaluations. Dataset and code available at: https://github.com/Mohamedelrefaie/DrivAerNet.

Problem

• Large-scale multimodal car dataset
• High-fidelity CFD simulations
• Aerodynamic drag prediction benchmarking

Innovation

• High-fidelity CFD simulations
• Extensive multimodal dataset
• Machine learning benchmarking