Eye movements of cyclists when interacting with automated vehicles: What can static images tell us?

The transition period towards large-scale or full automated driving will pose specific challenges. One of the challenges concerns the interaction of automated vehicles with vulnerable road users. So far, most studies into this type of interactions took the perspective of the car. The current study, however, takes the perspective of the vulnerable road user. More specifically, it explores how cyclists perceive and expect automated vehicles to ‘behave’ and how cyclists would react. Expectations are important determinants of traffic behaviour. Incorrect expectations could lead to overly trustful or hesitant behaviour and subsequent unsafe interactions. Recently, Hagenzieker and colleagues (2017) conducted a photo experiment in which regular cyclists had to judge photos
of traffic situations where they encountered manually-driven cars and automated cars (recognisable by either a sticker on the side of the car or a roof name plate on top of the car). Participants judged 30 photos twice, in random order. In that study a subset of nine participants were equipped with an eye tracker in order to study their eye movements while judging the photos. This study further examined these eye tracking data, comparing timeto- first-fixation, dwell time from the start of the first fixation, and total number of revisits in interactions with automated and with manually-driven cars. Results indicate no differences in time-to-first-fixation nor in the number of revisits between situations with automated cars and traditional cars. Dwell times revealed an effect of familiarity, showing that cyclists spent more time looking at cars during the first round of photos compared to the second round. In particular, they spent more time looking at automated cars which were identifiable by a ticker on the side. The results are discussed and suggestions for future research are given.