Exploring acceptable overtaking behaviour for automated vehicles

Modern transport research focuses on Automated Vehicle (AV) technologies, which are expected to significantly improve road safety, since most driver related accidents can be prevented. AVs will have to navigate in mixed traffic conditions, including vehicles of varying speeds, making overtaking an essential driving task. In order to encourage their uptake, the automated driving style will need to be acceptable to the user. Consequently, it is important to identify the parameters that affect overtaking acceptability, including manoeuvre characteristics, driving context and individual preferences. 

Driving situations of increased objective risk in manual driving constitute a priority for the implementation of automated features and are also likely to introduce perceived risk to AV passengers. Therefore, when investigating acceptable overtaking, test scenarios should account for situational factors related to objective risk. The range of variation of the overtaking characteristics that human drivers usually maintain can eliminate the feeling of a robotic operator and support the design of acceptable trajectories. Therefore, the current thesis aims to explore acceptable overtaking by (i) determining scenarios that are likely to introduce objective and thus perceived risk, (ii) investigating the parameters that affect acceptability in the identified driving situations and (iii) examining real-world overtaking behaviour. 

A cluster analysis was performed on the UK in-depth study RAIDS (Road Accident In-depth Studies) to mine combinations of situational factors related to overtaking crashes. Selected motorway overtaking scenarios were presented to 237 participants through an online survey to identify the impact of manoeuvre characteristics, driving context and driver characteristics on perceived risk and overall acceptability. The results indicated a strong impact of manoeuvre characteristics (pull-out distance, manoeuvre duration, speed) and a limited effect of the driving context and driver related characteristics. 

Real world overtaking behaviour under different traffic conditions was investigated by determining the variability of the pull-out comfort zone and modelling the examined manoeuvre characteristics of 1740 lane changes acquired from the naturalistic dataset highD. The results highlighted the significant impact of the surrounding traffic and the driving characteristics on or before the manoeuvre initiation point, which reflected the overtaking strategy selected (flying or accelerative).

The findings of this thesis can inform the design of user centred systems that assist or autonomously perform overtaking and thus create a positive passenger experience. Specifically, the research proposes context adaptive relationships of vehicle kinematic parameters to ensure pull-out manoeuvres meet the requirements of user acceptability.