Universal Design of Signage Through Virtual Human Simulation

Intuition behind sign placement and wayfinding features of architectural design can rarely encompass the needs of a wide range of building users. To help in automating sign placement, recent research has combined the use of agent-based simulation with optimization algorithms for maximizing visibility and wayfinding throughout a building model. As with many instances of machine learning applications, these too have unfortunately been dominated by an assumed young, healthy, and perfectly sighted virtual human. In this paper, we present an analysis of virtual human agents exploring a digital space using a combined vision and modified A* algorithm across multiple postures and visual impairments common amongst building occupants. We show how the inclusion of head angle and limited sights can change the results of what may be considered an optimal sign location.