Simplest instructions: Finding easy-to-describe routes for navigation

Abstract

Current applications for wayfinding and navigation assistance usually calculate the route to a destination based on the shortest or fastest path from the origin. However, numerous findings in cognitive science show that the ease of use and communication of route instructions depends on factors other than just the length of a route, such as the number and complexity of decision points. Building on previous work to improve the automatic generation of route instructions, this paper presents an algorithm for finding routes associated with the "simplest" instructions, taking into account fundamental principles of human direction giving, namely decision point complexity, references to landmarks, and spatial chunking. The algorithm presented can be computed in the same order of time complexity as Dijkstra's shortest path algorithm, O(n 2). Empirical evaluation demonstrates that the algorithm's performance is comparable to previous work on "simplest paths," with an average increase of path length of about 10% compared to the shortest path. However, the instructions generated are on average 50% shorter than those for shortest or simplest paths. The conclusions argue that the compactness of the descriptions, in combination with the incorporation of the basic cognitive principles of chunking and landmarks, provides evidence that these instructions are easier to understand. © 2008 Springer-Verlag Berlin Heidelberg.