Predicting lane changes in real-world driving: analysing explicit, implicit and contextual cues on the motorway

Abstract

Communication of intention between drivers relies on explicit cues and implicit cues. While these cues have been extensively studied in urban environments, their application to motorway driving remains underexplored. This gap is particularly evident in lane change scenarios, where communication failures such as the failure to accurately predict an imminent lane change can result in significant safety risks. Furthermore, although contextual cues are theorised to play an important role, empirical studies of their influence are limited. Results of our research start to fill both gaps by investigating how drivers predict lane changes on motorways, considering explicit, implicit and contextual cues. We utilized a Naturalistic Driving Study (NDS) approach, wherein 30 participants documented 798 relevant lane change situations during their daily drives. Data was collected through smartphones equipped with a custom app for real-time voice and video recording. The results highlight that effective communication relies primarily on implicit cues, such as longitudinal and lateral vehicle motion and movement patterns, which drivers use to predict driving behaviour. Contextual cues, including dynamic cues such as vehicle spacing and static cues such as the wider traffic environment, play a secondary role in shaping drivers’ predictions. Interestingly, explicit cues were rarely used by drivers in their decision making, highlighting their limited role in communication in motorway scenarios. These results support the development of automated driving styles that emulate human-like behaviour which could lead to automated vehicles that are more intuitive and predictable for human drivers, both within the automated vehicle and in the surrounding traffic.