A normal driving based deceleration behaviour study towards autonomous vehicles

Vehicle automation has recently attracted significant interest from the research community worldwide. Notwithstanding the remarkable development in autonomous vehicles (AVs), there is still a concern about the occupant's comfort since most research has mainly focused on the safety aspect. One of the most critical factors affecting the comfort level is the braking. It is however unclear which factors affect the braking behaviour and which braking profiles make the occupants feel safe and comfortable. This work therefore aims to thoroughly explore the deceleration behaviour of drivers using naturalistic driving study (NDS) data from two Field Operational Tests (FOT), the Pan-European TeleFOT (Field Operational Tests of Aftermarket and Nomadic Devices in Vehicles) project and the FOT conducted by Loughborough University and Nissan Ltd. A total of about 28 million observations were examined and almost 3,000 deceleration events from 37 different drivers and 174 different trips were identified and analysed. With the aid of a cluster analysis, a number of homogeneous scenarios based on human factors were formed. The scenarios have led to the application of multilevel mixed effect linear models to each cluster examining all influencing factors of the braking behaviour. The results indicate a dependence of the deceleration behaviour differing due to driver characteristics, initial speed and the reason for braking. Findings from this study will support vehicle manufacturers to ensure comfortable and safe braking operations of AVs