Safer Flight Systems using Dynamic Data-Driven Avionics
Speaker(s) : Dr. Carlos Varela
Sensor failures can lead to catastrophic accidents. For example, Air France 447’s accident was initially due to pitot tubes icing producing incorrect air speed data. Another example is Tuninter 1153’s accident caused by the installation of a wrong fuel quantity indicator. We study a dynamic data driven approach to detecting erroneous data produced by failing sensors, and when possible, we recover the data from redundant data streams. We describe physics-based mathematical concepts and software we have developed to tackle this challenge. We further discuss how to learn failure models from data, including previously unseen statistically significant failure modes. During flight emergencies, pilots need to make decisions very quickly to maximize the chance of a safe outcome. I will also briefly discuss preliminary results of our most recent work on creating safe gliding trajectories after engine failures, such as the US Airways 1549 incident over the Hudson River.
This research has been partially supported by the Air Force Office of Scientific Research and by the National Science Foundation. Publications are available at: http://wcl.cs.rpi.edu/bib/Keyword/DATA-STREAMING.html
More details here
cedric.herpson (at) nulllip6.fr