Pedestrian Navigation Simulation in Virtual Environment

Pedestrian navigation studies in real environments have shown consistent statistical relationships between ‘configuration’ properties of virtual layouts and movement flows, facilitating prediction of movement from designs. However, these studies are at an aggregate level and say nothing about how individuals make the micro-scale decisions producing these emergent regularities. They do not therefore ‘explain’ movement. Progress on this is difficult since decision-making mechanisms are hard to observe in the real world and the ‘experimenter effect’ is ever-present. If it could be shown that movement in VEs was analogous to movement in real environments then micro behavior data (direction of gaze, visual move behaviors) could be obtained through virtual experiments. The aim of this thesis is to explore this possibility by constructing experimental worlds with virtual properties varied to reflect those known to relate to movement in the real world, and asking individuals navigate through them immersive. Powerful analogies are initially demonstrated between virtual and real behavior. In this analysis, ‘subjective’ movement behavior is related to ‘objective’ properties of environments. The experiments show results strongly suggesting how noted aggregate regularities are produced: linearity is strongly conserved, usually following long sight lines, with pauses in configurationally ‘integrated’ locations offering strategic visual properties, long lines of sight, and large areas.