ETH Design++ Seminar Series 2024

Representations and Learning in Computer Aided Design

Event source: FCL · Category: Public

Computer-Aided Design (CAD) systems use two representation schemes, parametric geometry and procedural modelling, to achieve precision, accuracy, and editability. However, applying machine learning to CAD modelling has been difficult due to these constructs' complex and heterogeneous nature, resulting in a lack of AI in mechanical design. In this talk, Ben will discuss overcoming these challenges and enabling learned CAD model analysis. Ben will also present new and ongoing work on generating procedural representations of CAD models using semantic information from foundational text and image models.

Benjamin Jones is a PhD student at the Paul G. Allen School of Computer Science & Engineering at the University of Washington, advised by Adriana Schulz. His research focuses on bridging geometry, machine learning, and computational fabrication. He is interested in exploring complex design spaces and is currently working on neural representations for CAD geometry and AI exploration of CAD design spaces. Before joining the University of Washington, he earned bachelor's degrees in physics and joint mathematics and computer science from Harvey Mudd College. He helped develop a wireless power transmission array for space-based solar power and built distributed systems for web analytics at Quantcast. He also worked on computational imaging (Fourier ptychography) in the Biophotonics Laboratory at Caltech.


Future Cities Laboratory Global

Welcome to FCL Global, an interdisciplinary research programme that seeks to address the worldwide circumstances of rapid urbanisation. Our ultimate goal is to promote more equitable and livable urban futures, by bringing together Science, Design, Engineering and Governance.

Contact Us

Stay informed

Visit the outreach section to stay on top of on recent news, developments, publications and upcoming events at Future Cities Laboratory Global.

Outreach

Newsletter Signup

Subscription details are not shared with any 3rd parties.