Multidisciplinary Institute in Artificial intelligence
Prof Benjamin LECOUTEUX,
benjamin.lecouteux@univ-grenoble-alpes.fr
Prof Philippe MARIN,
philippe.marin@grenoble.archi.fr
The construction and architectural sectors are having difficulty to embrace digital transformation, supervisory bodies have identified the need to support the transformation of the sector, by encouraging the emergence of new knowledge and technical skills. Sector accounts for 30% of the european environmental footprint, making it the main contributors. It is essential to take greater account of environmental issues, right from the initial design phases.
AI plays a key role in these issues. In this context, the objectives of the project are twofold:
Contribute to the “capability” of designers, and meet the training challenges facing the schools of architecture and professionals;
Contribute to the design of a more environmentally-friendly built environment, by leveraging the qualities of AI systems.
Three main scientific challenges have been identified:
In terms of architectural sciences: Reframing the computational co-creativity, considering the effects and place of language in the process of designing, integrating environmental performance, right from the initial design phases.
In terms of computer sciences: Adapt language models (LLMs) to architecture-specific tasks by hybridizing them with domain-specific data and appropriately evaluate their performances and ethical issues.
In terms of engineering sciences: Optimizing solar productions and solar morphologies is a complex task, involving a multi-criteria optimization problem that considers heterogeneous parameters such as PV production, use, structure, comfort, energy, solar potential and local climat. AI facilitates these activities and speed up simulation processes.
The chair is organized around three packs:
The research pack is structured around two axes:
Computational Co-Creativity (CCC) : AI and creativity: How LLM can support humans in design (impacts and modalities of language interactions for representation productions).
Data Driven Design (DDD) : AI simulation and optimization: How predict the shape based on performance requirement, integrate environmental parameters into generative design processes.
The training pack has three components:
The innovation pack consider 4 main activities:
License:
