Testing Landmark Identification Theories in Virtual Environments

Landmarks are fundamental elements for people to learn an environment. People use these landmarks to enrich their route descriptions (for example, when anchoring movements at decision points). Several automatic landmark identification and selection theories have been suggested in recent years. This paper evaluates these theories by comparing the landmarks identified by automatic landmark selection with landmark choices and behavioral characteristics of human participants moving in a virtual environment. This comparison of automatic selection and human behavior will improve our understanding of automatic landmark identification theories, and will facilitate a weighting of methods for navigation services to generate more human-like route descriptions.