A Gaze- vs. Joystick-Based Interaction Method for a Remote Reconnaissance Task

Abstract

Increasing automation and artificial intelligence in every domain requires resolving complex human machine interactions. In the context of remote reconnaissance, an operator is likely to be supported by computer vision technology, whilst still making the final assessment, with the speed and precision of the classification being key. Within a vast design space with multiple design options, different interaction concepts were explored. One concept is based on gaze, which is tracked using eye tracking technology. It is then indicated if the operator's gaze is on a software identified object and allows an interaction through a secondary input device. To evaluate the concept, a simulator experiment was designed, comparing it to a joystick input. Expert users will complete two equal scenarios using both interaction methods, whilst metrics such as task success are recorded and interviews are conducted, for a balanced analysis of qualitative and quantitative, as well as subjective and objective data.