Advisory committee

Researchers in the areas of human wayfinding, spatial cognition, computer science, and artificial intelligence have developed cognitively based computational models for wayfinding. These models focus primarily on learning a spatial environment and on the exploration of mental representations rather than the information needs for wayfinding. It is important to consider the information needs because people trying to find their ways in unfamiliar environments do not have a previously acquired mental representation but depend on external information. The fundamental tenet of this work is that all such information must be presented to the wayfinder at each decision point as knowledge in the world . Simulating people’s wayfinding behavior in a cognitively plausible way requires the integration of structures for information perception and cognition in the underlying model. In this thesis we use a cognizing agent to simulate people’s wayfinding processes in an unfamiliar building. The agent-based model is grounded in the ontology and epistemology of the agent and its environment. Both are derived from human subjects testing using an ecological approach. This leads to two tiers in the conceptual model: simulated states of the environment and simulated beliefs of the agent. The agent is modeled with state, an observation schema, wayfinding strategies, and commonsense knowledge. The wayfinding environment is modeled as a graph, where nodes represent decision points and edges represent lines of movement. The perceptual wayfinding model integrates the agent and its environment within a Sense-Plan-Act framework. It focuses on knowledge in the world to explain actions of the agent while performing a wayfinding task. We use the concepts of affordance and information to describe what kinds of knowledge the agent derives from the world by means of visual perception. Affordances are possibilities for action with reference to the agent. Information such as from signs is necessary for the agent to decide which affordances to utilize. During the navigation process the agent accum ulates beliefs about the environment by observing task-relevant affordances and information at decision points. The utilization of a so-called “go-to” affordance, i.e., following a pathway, leads the agent from one node to another where it is again provide d with percepts. A successful navigation corresponds to the agent’s traversal from a start to a goal node. The perceptual wayfinding