Behavioural Modelling of Automated to Manual Control Transition in Conditionally Automated Driving

Abstract

Human-vehicle interaction in the presence of automated driving features (ADFs) poses significant challenges in behavioural adaptation. At the current level of automation, the mixed right-of-control of connected and automated vehicle (CAV) with partial human involvement is of interest. This study explores human behaviour and the role of perception when regaining the control from the conditionally automated vehicle upon operational design domain (ODD) exit under different socio-demographic, mental workload, traffic flow, weather, and lighting conditions. Data collected in Virtual and Immersive Reality Environment (VIRE) (Farooq et al., 2018) is used to study the successful taking of the control back from automated driving. Apart from estimating binary and mixed logit models, latent structure was developed to incorporate the psychometric indicators in the integrated choice and latent variable (ICLV) model. Results suggest that participants were generally successful in safely regaining control. However, participants with more sensitive perceptions about CAV safety, were more likely to fail. Safe control transition became less successful as the mental workload and reaction times increased in worse weather conditions. Multitasking and congestion have a higher impact on the situational awareness in taking back control safely from CAV. Driver’s familiarity with the concept of CAVs was not enough for the safe transition of control.

Keywords: driver behaviour, conditional automation, connected and automated vehicles, hybrid choice, virtual immersive reality