Artificial Intelligence Technique for Pilot Approach Decision Aid Logic (PADAL) System

Abstract

Abstract : In this project the objective was to improve Landing Signal Officer (LSO) decision making by using Artificial Intelligence (Al) and other techniques to develop pilot trending and ship oscillation decision support aids. During the pursuit and satisfaction of the primary objective, several sub-objectives were met. The project developed pilot trending and ship oscillation recognition techniques and software by investigating the use of Fourier, wavelet, neural networks, fuzzy logic and other transform techniques in conjunction with the application of decision-centered design methodologies from cognitive psychology; the research determined that a combination of neural networks and fuzzy logic applied under a decision-centered design approach proved most useful and was developed. We determined the significant aircraft approach parameters and similarity measures and important pilot considerations and similarity measures. We also developed pilot trending techniques and software using case-based reasoning and combinations of other Al techniques. In addition, in conjunction with many LSOs, we determined the best display options and most appropriate display logic for the information produced by the pilot trending and oscillation recognition modules, and designed and implemented the resulting LSO interface. Then the design concepts were implemented and tested, in an iterative fashion. The decision aid prototypes were evaluated and critiqued by active LSOs with enhancements based on feedback from the LSOs.