Collaborators: Glaang Therakomen, Sitt Therakomen, Ellen Yi-Luen Do

Mousehaus simulates pedestrian activities in a virtual environment to help architects and urban designers think about how people inhabit public outdoor space in cities. The program uses multi-agent technology in which each autonomous agent (“mouse”) employs individual behaviors and rules selected from a database obtained from empirical observations to perform its role as a pseudo-pedestrian. A “mouse” moves through a model of an urban environment, reacting to buildings and open space as well as to other mice. The system then tracks each mouse’s path, recording the overall movement patterns, which urban designers can use to understand relationships between pedestrians and the built environment.

The design of pedestrian friendly urban places is the starting point for the Mouse.class project. Where will people tend to sit? Where will they walk? Where will they stop to talk? In Mouse.class, local behavior rules programmed into simulated pedestrians (“mice”) cause them to move about, responding to internal needs and environmental stimuli. A hierarchy of rules—from reflexive and reactive walking, to goal seeking, to imitative and inductive behavior—results in complex and seemingly life-like group dynamics. Simulating pedestrian behavior can help urban designers identify problems and opportunities in a design proposal before it is built.

Mouse.class: The Experiments for Exploring Dynamic Behaviors in Urban Places, Preechaya Therakomen, M.Arch thesis, University of Washington 2001 [pdf]