Bridging Design and Fabrication via Neuro-Symbolic Visual Program Synthesis and Rendering
Research Team
Our Motivation:
“Our research will bring together recent advances in artificial intelligence (AI), computer vision, graphics, and fabrication, leading to both fundamental innovations in neuro-symbolic AI and neural/inverse rendering, and significant industrial impacts in applications that facilitate designers, builders, and owners..”
Research Contribution
Seamless integration of multiple data modalities for bridge owners, designers, and builders.
Use of inverse and neural rendering to connect 3D shapes/scene layouts and 2D images.
Develop neuro-symbolic program synthesis and execution algorithms to infer vectorized CAD models for fabrication from raw shapes.
Problem
Practical Problem
Owners, designers, and builders are three key players in the design and fabrication processes; however, the type of data that each group works with is fundamentally different, due to various engineering and cognitive constraints.
Owners may suggest to designers the preferred layout, furniture shape, texture, or fabric via 2D raw images, sketches, and line drawings, designers mostly work with 3D data, such as meshes and primitives, and builders would need detailed fabrication instructions, such as CAD and BIM models.
Solution
Bridge design and fabrication by inferring such sequence information from raw data via neuro-symbolic visual program synthesis and rendering.
Added Value For The Industry
Our research will directly benefit owners, designers, and builders, by significantly reducing their collaboration and communication cost and workload, and improving the flexibility and richness in design and fabrication. Ultimately, we hope our research will lead to the fundamental integration of the three parties.
Cooperation Partner
Chin-Yi Chang
Research Manager and Principal Research Scientist at Autodesk AI Lab
Ran Zhang
Sr. Research Scientist - Machine Learning at Autodesk
Timeline
|
Date |
Activity |
|
|---|---|---|
|
Year 2021 |
Research became awarded: Bridging Design and Fabrication via Neuro-Symbolic Visual Program Synthesis and Rendering |
|
|
Fall 2021 |
Start working with industry partners on case-studies |
|
|
Fall 2021 |
Literature review to understand product metrics |
|
|
Winter 2021 |
Problem formulation and pilot data collection |
|
|
Winter 2021 |
Building initial prototypes |
|
If you want to participate in the project please reach out to the Research Team.
Seed Research is in progress and will be continued in the academic year 2022/23.