> AiDAPT Lab: TU Delft’s AI Lab for Design, Analysis, and Optimization
Faculty of Architecture & the Built Environment
Keywords
Artificial Intelligence, Optimization, Life-cycle Decision-Making, Design.
Initiated by
Charalampos Andriotis, Seyran Khademi
About
AiDAPT aims at developing AI methods able to support informed design decisions, through coupled data-driven and model-based approaches to the design-related optimization processes. Seeing design as an extended, multi-scale and dynamic life-cycle procedure, the two main research streams
AiDAPT pursues are:
- Dynamic learning and autonomous decision-making under uncertainty to enhance the life-cycle dimension of design. This includes reinforcement learning and deep learning methods, aiming at assisting architects and engineers to assess and control the long-term physics-driven responses of the built environment.
- Automatic recognition (classification, localization, matching, retrieval) and understanding (captioning, interpretation, summarization) the attributes of the visual data in various architectural representations and scales, aiming at assisting architects and engineers by providing relevant and interpretable visual analysis for the initial design process.
Overall, the developed frameworks will enable intelligent processes for abstraction and synthesis of structural and architectural decisions, ranging from the stage of initial design to future intervention and adaptation planning (e.g. maintenance, retrofits, form changes, etc.) and data collection scheduling. This integral approach will close the loop from data to design and vice versa, supporting adaptive and evidence-based choices for architects, civil engineers, designers, and policy-makers.
Funded by
TU Delft AI Labs Program
Contact