Modelling drivers’ braking behaviour and comfort under normal driving
thesisposted on 14.05.2020, 15:44 by Penny Deligianni
The increasing growth of population and a rising number of vehicles, connected to an individual, demand new solutions to reduce traffic delays and enhance road safety. Autonomous Vehicles (AVs) have been considered as an optimal solution to overcome those problems. Despite the remarkable research and development progress in the area of (semi) AVs over the last decades, there is still concern that occupants may not feel safe and comfortable due to the robot-like driving behaviour of the current technology. In order to facilitate their rapid uptake and market penetration, ride comfort in AVs must be ensured.
Braking behaviour has been identified to be a crucial factor in ride comfort. There is a dearth of research on which factors affect the braking behaviour and the comfort level while braking and which braking profiles make the occupants feel safe and comfortable. Therefore, the primary aim of this thesis is to model the deceleration events of drivers under normal driving conditions to guide comfortable braking design. The aim was achieved by exploiting naturalistic driving data from three projects: (1) the Pan-European TeleFOT (Field Operational Tests of Aftermarket and Nomadic Devices in Vehicles) project, (2) the Field Operational Test (FOT) conducted by Loughborough University and Original Equipment Manufacturer (OEM), and (3) the UDRIVE Naturalistic Driving Study.
A total of about 35 million observations were examined from 86 different drivers and 644 different trips resulting in almost 10,000 deceleration events for the braking features analysis and 21,600 deceleration events for the comfort level analysis. Since deceleration events are nested within trips and trips within drivers, multilevel mixed-effects linear models were employed to develop relationships between deceleration value and duration and the factors influencing them. The examined factors were kinematics, situational, driver and trip characteristics with the first two categories to affect the most the deceleration features. More specifically, the initial speed and the reason for braking play a significant role, whereas the driver’s characteristics, i.e. the age and gender do not affect the deceleration features, except for driver’s experience which significantly affects the deceleration duration.
An algorithm was developed to calculate the braking profiles, indicating that the most used profile follows smooth braking at the beginning followed by a harder one. Moreover, comfort levels of drivers were analysed using the Mixed Multinomial Logit models to identify the effect of the explanatory factors on the comfort category of braking events. Kinematic factors and especially TTC and time headway (THW) were found to affect the most the comfort level. Particularly, when TTC or THW are increased by 1 second, the odds of the event to be “very comfortable” are respectively 1.03 and 4.5 times higher than being “very uncomfortable”. Moreover, the driver’s characteristic, i.e. age and gender affect significantly the comfort level of the deceleration event. Findings from this thesis can support vehicle manufacturers to ensure comfortable and safe braking operations of AVs.
- Architecture, Building and Civil Engineering