Presenting uncertainty in general aviation UIs for supporting decisions in safety critical situations.

Abstract

This PhD research aims at studying the presentation of uncertain data to the operators of highly automated and safety critical systems such as pilots and air traffic controllers. Despite the efforts for increasing accuracy and predictability, uncertainty remains an intrinsic component in aviation and air traffic management (ATM) systems. Moreover, systems with high levels of automation, i.e. prediction algorithms, can produce further information characterized by uncertainty. In this work I present the design and development of an interface focusing on the uncertainty visualization method and solutions. A literature review explores the relationships between the data, the highly automated systems and the human. The aspects embraced in the review describe, in addition, the nature of the visualization problem for the uncertain data. Experimental results from three different case studies centred on the light general aviation domain are presented and discussed. Results from user observations and human in the loop simulation analysis indicate that uncertain data can be used and may provide concrete benefits, in the specific case on the situational awareness component. An accurate design process that supports the introduction of features involving uncertain data must be is in place, however, to observe the desired benefits. Such process must involve users and/or domain experts as it requires a set of specific information that it is unlikely to be elicited by other means. To support this process, a framework for the visualization of uncertain data is described at the end of the thesis. The aim is to demonstrate that raising awareness about the uncertainty through the presentation on the user interface can be an effective approach to cope with the side effects of automation and may allow an effective usage of such information.