Computational Cognitive Modeling of Pilot Performance in Pre-flight and Take-off Procedures

Abstract

Verbatim from Embry-Riddle Scholarly Commons. Not paraphrased, not summarized.

While the current practice of pilot training relies on flight instructors’ subjective assessment, computational cognitive modeling may be used to support future objective assessment and diagnosis of pilot performance. We built two models in a cognitive architecture to simulate pilot flight performance during pre-flight and take-off tasks. Modeling results were compared with human results collected from the same tasks using X-Plane 11 flight simulator. The models were able to capture human pilot performance and workload results from both tasks with good levels of fitness (percentage errors ranging from 0.8% to 13.2%). This work demonstrated the capability and advantage of this theory-driven modeling approach for supporting general aviation pilot training. We expect that this type of cognitive model will be complementary to data-driven machine learning models, and the current work provides the foundation for future work to expand the modeling capability and test practical applications in general aviation.

Authors

• Xu, Rongbing, M.A.Sc Embry-Riddle Aeronautical University
• Cao, Shi, Ph.D. Embry-Riddle Aeronautical University
• Kearns, Suzanne K, Ph.D. Embry-Riddle Aeronautical University
• Niechwiej-Szwedo, Ewa, Ph.D. Embry-Riddle Aeronautical University
• Irving, Elizabeth, Ph.D. Embry-Riddle Aeronautical University

Keywords

• cognitive modeling
• QN-ACTR
• pilot performance
• workload
• competency-based training
• Systems Engineering