Model Based Analysis of Two-Alternative Decision Errors in a Videopanorama-Based Remote Tower Work Position

Initial analysis of Remote Control Tower (RTO) field test with a prototype videopanorama system under quasi operational conditions (Friedrich and Möhlenbrink in ATM, 2013) has shown performance deficits quantified by two-alternative aircraft maneuver discrimination tasks (Fürstenau et al. in EPCE/HCII 2013, Part II, Lecture Notes in Artificial Intelligence (LNAI), pp. 105–114, 2013). Here we present the quantitative analysis of these results using the complementary methods of Bayes inference, signal detection theory (SDT) with parametric and non-parametric discriminabilities d’ and A, and application of time pressure theory (Fürstenau et al. in EPCE/HCII 2013, Lecture Notes in Artificial Intelligence (LNAI), pp. 143–154, 2014). RTO-controller working position (CWP-) performance was directly compared with that one of the conventional tower-CWP with direct out-of-windows view by means of simultaneous aircraft maneuver observations within the control zone at both operator positions. For this analysis we took into account correct (Hit rate), incorrect (False Alarms, FA) answers to discrimination tasks, and we took into account non-answers for a pessimistic quantification of RTO-performance. As initial working hypothesis this lead to the concept of time pressure (TP) as one major source of the measured response errors. A fit of experimental error rates with an error function derived from the Hendy et al. information processing (IP/TP)-hypothesis (Hendy et al. in Human Factors 39:30–47, 1997) provides some evidence in support of this model. We expect the RTO-performance deficits to decrease with the introduction of certain automation features to reduce time pressure and improve the usability of the videopanorama system.