Methodologies to Understand the Road User Needs when Interacting with Automated Vehicles

Interactions among road users play an important role for road safety and fluent traffic. In order to design appropriate interaction strategies for auto-mated vehicles, observational studies were conducted in Athens (Greece), Mu-nich (Germany), Leeds (UK) and in Rockville, MD (USA). Naturalistic behav-iour was studied, as it may expose interesting scenarios not encountered in con-trolled conditions. Video and LiDAR recordings were used to extract kinematic information of all road users involved in an interaction and to develop appropriate kinematic models that can be used to predict other’s behaviour or plan the behav-iour of an automated vehicle. Manual on-site observations of interactions pro-vided additional behavioural information that may not have been visible via the overhead camera or LiDAR recordings. Verbal protocols were also applied to get a more direct recording of the human thought process. Real-time verbal re-ports deliver a richness of information that is inaccessible by purely quantitative data but they may pose excessive cognitive workload and remain incomplete. A retrospective commentary was applied in complex traffic environment, which however carries an increased risk of omission, rationalization and reconstruction. This is why it was applied while the participants were watching videos from their eye gaze recording. The commentaries revealed signals and cues used in interac-tions and in drivers’ decision-making, that cannot be captured by objective meth-ods. Multiple methods need to be combined, objective and qualitative ones, de-pending on the specific objectives of each future study.