Intent-Based Mode-Confusion Detection (Learning-based)

Detailed mathematical models of the autopilot and flight management system logics with known system parameters are difficult to obtain
An intent-based Generalized Fuzzy Hidden Markov Model (GFHMM) using a Viterbi-like algorithm is developed to learn model parameters for mode confusion detection

Proposed Architecture

Demonstration

Demonstrated with data of ~600 flights generated by the Multi-Aircraft Control System (MACS)
Illustrative Example

Vertical Profile

Zoomed-in view from 60 sec to 150 sec

Inferred intents (top: automation; middle: pilot) and detected mode confusion (bottom)

Related Publication

H. Lyu, J.S. Nandiganahalli, S. Lee, and I. Hwang, “Automation Intent Inference Using the GFHMM for Flight Deck Mode Confusion Detection,” AIAA Journal of Aerospace Information Systems, Vol. 15, No. 3 (2018), pp. 172-177.

Google Sites

Report abuse