Mar 29, 2016: Dennis Guck: Exploring Model Conformance Relations for ACAS X

March 29, 2016Exploring Model Conformance Relations for ACAS X
Room: HB 2FDennis Guck

The next generation airborne collision avoidance system, ACAS X, targets robustness through a probabilistic model that represents sources of uncertainty. Dynamic programming then produces a look-up table, used during deployment to produce advisories. In this presentation, we investigate relations between the probabilistic model and the real world, in an attempt to characterize the quality of the model for the pur- pose of building ACAS X. Our purpose is to develop model comparison metrics. The current measure of model quality is for the resulting system to perform better than the state-of-the-art, TCAS. Our work aims at building confidence beyond this measure, by characterizing the relative quality of different models as used within this context. We discuss conformance relations that we have studied, their application to ACAS X, and the results that we obtained.