Modeling with Perceptual and Memory Constraints: An EPIC-Soar Model of a Simplified Enroute Air Traffic Control Task

摘要

This document reports the human performance model developed by Soar Technology, Inc. as part of the initial model comparison process of the Agent- Based Modeling and Behavioral Representation (AMBR) program being conducted by the Air Force Research Laboratory. The Soar effort is a simplified enroute air traffic control (ATC) task. A human subject (or the model) performs the task of a controller by communicating with adjacent traffic controllers and aircraft to maintain a smooth flow of traffic into and out of the center airspace. 'Smooth flow' is defined as performing aircraft and ATC transaction in a timely manner such that aircraft do not enter a 'hold' pattern because they have not been transacted. The architecture used is a hybrid system called EPIC-Soar. EPIC (executive Process-Interactive Control) is an architecture whose primary goal is to account for detailed human dual-task performance. Soar is a general architecture for building artificially intelligent systems and for modeling human cognition and behavior. EPIC-Soar is an integration of the perceptual and motor processor models of EPIC and Soar. It is an attempt to get the best of both worlds: the detailed predictions and explanations of sensory and motor systems from EPIC, and the broader, cognitive problem solving, planning, and learning capabilities of Soar.