Hi! I’m Thomas. I am currently finishing my third year as a PhD student at Polytechnique Montréal. My research interests include the application of deep reinforcement learning to the task of structural design. I am currently studying the combination of imitation learning and policy optimization to improve the performance of reinforcement learning agents in the task of structural design. I am also interested in the development of flexible vision models for representation learning of 2D and 3D meshes of variable resolutions.
Outside of school, I like to fly planes, fix coffee machines and go down bumpy hills on my mountain bike.
The study presents complexity-driven layout exploration for aircraft structures (CD-LEAS), a novel process for efficient topology optimization. Case studies confirm CD-LEAS’s ability to produce simple, light, stiff, and buckling-resistant layouts.
This research leverages a deep learning model to improve scalability in topology optimization, with a new approach that reduces computation time by 36.84% and maintains mechanical performance.
Comparison of first-order optimizers for the Moving Morphable Components topology optimization framework. Hybrid MMA-GCMMA optimizer is shown to have a better convergence behaviour for the MMC framework.