Spotting sheep in Yorkshire: using eye-tracking for studying situation awareness in a driving simulator

The paper describes an experiment designed for the EASY (Effects of Automated Systems on safetY) project, in which the effect of highly automated driving on driver situation awareness was studied. Driving was either manual, with all manoeuvres controlled by the driver, or highly automated, where lateral or longitudinal control of the car was maintained by an ‗automated system‘. In both cases, drivers were requested to complete a number of secondary tasks during driving, such as answering questions over a mobile telephone or watching a video. Drivers‘ eye movements towards the sudden appearance of objects (sheep, police cars) at the side of the road were recorded, and used to infer situation awareness. This methodology was favoured over freeze frame techniques which interrupt the driving task. Whilst eye tracking is a well established technique, its use for interpreting visual attention in dynamic scenes is more problematic. This paper discusses the different analysis techniques used to investigate visual attention to events in a dynamic driving simulator environment, comparing frame by frame playback of video scenes with summary charts based on predefined areas of interest. Lessons learnt from each technique are outlined.