Computer Aided Engineering (CAE) is at the core of modern industrial engineering and manufacturing. However, the current CAE applications suffer from significant time and human resource expenses. Our goal is to leverage deep learning techniques to automate the CAE process and reduce the R&D costs for the industry.
At CVLab, we target at aeronautics engineering and focus on three main problems:
- Object geometry modelling and parameterization [1,2,3]
- Physical performance evaluation and surrogate modeling [4]
- Gradient-based optimization [1,2]
We employ deep learning as a data-driven approach to learn human prior knowledge, which can minimize human intervention in the design-and-validation loop. We also use deep learning to tackle high dimensional problems and empower traditional CAE applications to address more complicated problems. Meanwhile, the application of deep learning presents new challenges that give rise to academic problems such as volumetric mesh regularization and manipulation, data-efficient training, uncertainty estimation, etc.
Students will work on one of the problems we provide. The work will be done in collaborations with aerodynamics experts at ISAE-SUPAERO in Toulouse, France.
Prerequisites:
-
Proficiency in Python
-
For computer science students: Familiarity with deep learning and the PyTorch library. Knowledge of 3D geometry and meshes would be advantageous.
-
For mechanical engineering students: Familiarity with computational fluids dynamics. Knowledge of machine learning, and skills of meshing and fluid simulation would be advantageous.
Contact:
If you are interested, please write to [email protected]
Reference:
[1] Wei, Z., Guillard, B., Fua, P., Chapin, V., and Bauerheim, M., “Latent Representation of CFD Meshes and Application to 2D Airfoil Aerodynamics,” AIAA Journal, 2023
[2] Wei, Z., Fua, P., and Bauerheim, M., “Automatic Parameterization for Aerodynamic Shape Optimization via Deep Geometric Learning,” AIAA Aviation Forum, 2023
[3] Remelli E., Lukoianov A., Richter S., Guillard B., Bagautdinov T., Baque P., Fua P., “MeshSDF: Differentiable Iso-Surface Extraction,” NeurIPS 2020
[4] Baqué, P., Remelli, E., Fleuret, F., and Fua, P., “Geodesic Convolutional Shape Optimization,” ICML, 2018