Integrated Modeling of Cognition and the Information Environment: Closed-Loop, ACT-R Modeling of Aviation Taxi Errors and Performance Technical rept

摘要

This report describes the progress made, findings, and lessons learned in modeling the T-NASA2 scenario within the ACT-R/PM cognitive architecture. The report begins with a brief overview of the motivation for integrated modeling of cognition and the information environment in the domain of aviation safety. The ACT-R/PM modeling framework is then introduced, along with a description of the ecological, rational and environmental analysis methods needed to put flesh on the bones of this cognitive architecture. The application to the T-NASA2 scenario is described next, first in terms of overall flow of control (goal selection), and at more detail in terms of the goal decomposition around which ACT-R/PM productions were constructed (e.g., look for incursion, maintain speed). Analysis and modeling revealed that an accurate representation of turn-related decision strategies was crucial to the success of the model, in terms of replicating human performance and error in the T-NASA2 scenario. As a result, a study was performed with an SME, a working airline pilot, to assess any high level (i.e., airport neutral) knowledge pilots might use in deciding what turns to make during taxi operations. The results of this rational analysis study, which was based on a sample of 284 candidate taxi routes over eight major U.S. airports, suggested that taxiway geometry and airport layout was much more regularly structured than we had originally thought. This finding suggested that it is likely that pilots make turn-related decisions using a combination of specific information gained in real time from perception of airport layout, and from longer term and more general knowledge of regularities in the likely paths of travel between touchdown and gate.