My research focuses on representing and solving spatial problems in a general-purpose cognitive architecture. This takes place in the context of an existing system, Soar, but the underlying research questions are independent of it. My approach centers on representing space both abstractly (e.g., "block A is on the table"), and in terms of more detailed, sensory-level information (e.g., pixels or 3d coordinates). By using both forms of representation in concert, an agent can achieve behavior that is both precise within a given problem, and general across problems. This approach to spatial representation is inspired by the psychological phenomenon of mental imagery-- the ability for humans to create low-level sensations corresponding to hypothetical or previously-encountered situations. The goal of my research is not to simply mimic this, though, but instead to examine how it can provide functional benefits to an agent.

publications

miscellaneous other projects:

balloonsat	dome drawing	universal text imitator