Predicting pilot behavior during midair encounters using recognition primed decision model

Next Generation Air Transportation System is designed to deal with the dramatic increase of air traffic. To achieve this goal, new technologies and operation concepts are to be introduced. In this paper, a novel pilot behavior model is developed to support design and analysis of new technologies and operation concepts that are to be used in midair encounter scenarios. In this method, Bayesian network is used to build a probabilistic framework for human and automation components. To model human pilot, different human information processing stages which include perception, decision making and response are formulated. Especially, this paper firstly proposes a recognition primed decision model to describe human pilot's decision making process. In addition, a structural pilot model is introduced to promote the accuracy of existing pilot behavior models. Finally, this paper shows how this method could be used to study midair encounter scenarios. Meanwhile, the validity and feasibility of this method are investigated and discussed.