My thanks to Frederik Brudy for letting me know about a recent publication from our research team.
One of our summer internships from 2025 has resulted in a paper being published at EMNLP, the Conference on Empirical Methods in Natural Language Processing. This was a new conference to me - as many of them are, admittedly - but it's a monster: it's 30 years old and this year's event had 3000+ accepted papers. The table of contents in the published findings runs to 90 pages. Gulp.
Anyway, the paper our team helped author is called "IndoorWorld: Integrating Physical Task Solving and Social Simulation in A Heterogeneous MultiAgent Environment". Here's the abstract:
Virtual environments are essential to AI agent research. Existing environments for LLM agent research typically focus on either physical task solving or social simulation, with the former oversimplifying agent individuality and social dynamics, and the latter lacking physical grounding of social behaviors. We introduce IndoorWorld, a heterogeneous multi-agent environment that tightly integrates physical and social dynamics. By introducing novel challenges for LLM-driven agents in orchestrating social dynamics to influence physical environments and anchoring social interactions within world states, IndoorWorld opens up possibilities of LLM-based building occupant simulation for architectural design. We demonstrate the potential with a series of experiments within an office setting to examine the impact of multi-agent collaboration, resource competition, and spatial layout on agent behavior.
This paper marks an important step in our exploration around the potential for AI (in this case LLMs) to drive human simulation of design.
If either the abstract or the above figure seems of interest to you, be sure to take a look at the full PDF.
Leave a Reply