M. Raubal (2000) Human wayfinding in unfamiliar buildings: a simulation with cognizing agents. in: M. Belardinelli (Ed.), International Conference on Spatial Cognition: Scientific Research and Applications (ICSC 2000), Cognitive Processing - special issue 2000, Rome, Italy, pp. 66.


Researchers in the areas of human wayfinding, spatial cognition, computer science, and artificial intelligence have developed cognitively based computer models for wayfinding. These models focus primarily on learning a spatial environment and on the exploration of mental representations rather than the information needs (Gluck 1991) for wayfinding. It is important to consider information needs because people trying to find their ways in unfamiliar environments do not have a cognitive-map-like representation but depend on external information. To simulate such wayfinding behavior in a cognitively plausible way it is therefore necessary to integrate structures of information perception and cognition in the underlying model.

In this work we use a cognizing agent to simulate people's wayfinding processes in buildings. The wayfinding model integrates an agent's cognitive schema and perceptual structures within an SPA (Sense-Plan-Act) approach. It focuses on external knowledge to explain actions of the agent performing wayfinding tasks. We use the concepts of information and affordances to describe the kinds of knowledge agents derive from the world by means of visual perception. Affordances (Gibson 1979) are possibilities for action with reference to the agent. Information (such as from signs) is necessary for the agent to decide upon which affordances to utilize. An internal cognitive schema (Neisser 1976) guides the agent's processes of perception, decision, and action during the wayfinding task. This schema includes information about the task and goal, and a minimum of wayfinding strategies and commonsense knowledge necessary to perform the task. The task description directs visual perception in such a way that the agent samples only task-relevant information and affordances. The wayfinding model concentrates on the actual information needs during wayfinding and does not focus on learning a spatial environment. Its fundamental tenet is that all information must be presented at each decision point as "knowledge in the world" (Norman 1988).

The proposed formal specification of the agent-based model within a functional programming environment can be used to simulate people's wayfinding behavior in spatial information and design systems. We employ the specific case of wayfinding in an airport to demonstrate the proposed model. The result is a test of the signage in the airport.