Assessment of Landmark Visibility for Automated Wayfinding Route Instructions

Abstract

Landmarks are used by people in unfamiliar spatial environments as points of reference with which to construct a mental comprehension of space. Supplementing route directions with landmark knowledge therefore improves cognitive ergonomics by increasing people’s ability to use, remember and communicate routes. This research addresses the problem of incorporating landmark knowledge into automatically generated route instructions by exploring the full lifecycle of application development. Two development stages are identified, an initial offline phase and a subsequent online phase. Offline methods are developed to increase LiDAR (Light Detection and Ranging) building representational accuracy, and to populate landmarks with Pointclouds representing their structural morphology, allowing quantification of landmark saliency. Viewshed analysis is performed and precalculated saliency values are stored as a static road network representative of any potential egocentric viewpoint, allowing landmarks with the greatest saliency at a point in space to be selected. The second online phase explores the delivery of landmark information by allowing user route searches, such that supplementary landmark information is compiled on-the-fly, and is route dependant. A prototype application, delivered as an online web mapping service utilising an open source, online routing system demonstrates practicality and scalability of the approach.