VISION-BASED WAYFINDING SIMULATION OF DIGITAL HUMAN MODEL IN THREE DIMENSIONAL AS-IS ENVIRONMENT MODELS AND ITS APPLICATION TO ACCESSIBILITY EVALUATION

摘要

Elderly and disabled individuals must be able to access the indoor and outdoor environment in an easy and safe manner. Accessibility must be assessed not only in terms of physical friendliness for users, but also of cognitive friendliness such as the ease of wayfinding. To ensure the ease of wayfinding, signage information available at every key decision point is essential because it enables people to find their way in unfamiliar environment. The aim of the present study is to develop virtual accessibility evaluation system that evaluates the environment accessibility from the cognitive friendliness aspect, such as the ease of wayfinding, by combining realistic human behavior simulation using a digital human model (DHM) with as-is environment models. To realize this system, we develop a vision-based wayfinding simulation algorithm for the DHM in textured three-dimensional (3D) as-is environment models. The as-is environment models are constructed with the structure-from-motion (SfM) technique. During the wayfinding simulation, the visibility and legibility of each sign are evaluated on the basis of the visual perception of the DHM and its visibility catchment area (VCA). The DHM walking trajectory is dynamically generated depending on the perceived sign. When a disorientation place is detected where the DHM cannot find any sign indicating its destination, plans for rearranging the signs are proposed by the simulation user and then examined using a developed virtual eyesight simulator (VES). The VES enables the user to check the DHM eyesight virtually during the wayfinding simulation through a head-mounted display. To mimic visual impairments, visual impairment filters are introduced into the VES. In this paper,we demonstrate the process of detecting the disorientation place, and planning and evaluating the rearranged signage.