AutoHood3D: A Multi-Modal Benchmark for Automotive Hood Design and Fluid-Structure Interaction

2025-11-05

AI Summary

Existing datasets for automotive hood fluid–structure interaction (FSI) modeling suffer from dimensional limitations (2D-only), insufficient geometric diversity, absence of multimodal annotations, and lack of structured physical representations—hindering the development of physics-aware generative design. To address this, we introduce the first high-fidelity multimodal FSI benchmark dataset, comprising 16,000 parameterized hood geometries, each accompanied by STL meshes, time-resolved physical fields (pressure, displacement, stress), finite-element responses, and natural-language descriptions. We propose a physics-constrained multi-task learning architecture with a novel FSI joint loss function, integrating five neural network baseline models. Experiments establish systematic surrogate error benchmarks for displacement and load prediction, demonstrating that multimodal fusion significantly improves both generative design capability and physical consistency. The dataset, source code, and training pipelines are fully open-sourced.

Technology Category

• Machine Learning: Multimodal Learning
• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

This study presents a new high-fidelity multi-modal dataset containing 16000+ geometric variants of automotive hoods useful for machine learning (ML) applications such as engineering component design and process optimization, and multiphysics system surrogates. The dataset is centered on a practical multiphysics problem-hood deformation from fluid entrapment and inertial loading during rotary-dip painting. Each hood is numerically modeled with a coupled Large-Eddy Simulation (LES)-Finite Element Analysis (FEA), using 1.2M cells in total to ensure spatial and temporal accuracy. The dataset provides time-resolved physical fields, along with STL meshes and structured natural language prompts for text-to-geometry synthesis. Existing datasets are either confined to 2D cases, exhibit limited geometric variations, or lack the multi-modal annotations and data structures - shortcomings we address with AutoHood3D. We validate our numerical methodology, establish quantitative baselines across five neural architectures, and demonstrate systematic surrogate errors in displacement and force predictions. These findings motivate the design of novel approaches and multiphysics loss functions that enforce fluid-solid coupling during model training. By providing fully reproducible workflows, AutoHood3D enables physics-aware ML development, accelerates generative-design iteration, and facilitates the creation of new FSI benchmarks. Dataset and code URLs in Appendix.

Problem

• Addresses automotive hood deformation during rotary-dip painting processes
• Provides multi-modal dataset for fluid-structure interaction and machine learning
• Overcomes limitations of 2D cases and limited geometric variations

Innovation

• Large-Eddy Simulation coupled with Finite Element Analysis
• Multi-modal dataset with geometric variants and physical fields
• Structured natural language prompts for text-to-geometry synthesis