Explore the scholarly record.

Abstract

The impact of autonomous vehicles (AVs) on travel behaviour remains uncertain, largely due to two key factors: first, how individuals inherently perceive different privately owned and shared AV forms, and second, whether time is valued differently across these modes. The current study addresses both issues by developing a Bayesian panel mixed multinomial logit model based on a stated choice experiment conducted in the UK. The model empirically quantifies the inherent attraction and reservation effects associated with AVs, beyond those captured by service-level factors such as travel time and cost, and conceptualises them as the rest of convenience (ROC) value. ROC values suggest that AV modes are inherently less preferred than conventional alternatives, albeit with substantial heterogeneity among them. AV ride-hailing exhibited the lowest inconvenience barrier per trip (-£6.30), followed by AV minibuses (-£8.43), while private AVs (-£14.03) remained the least attractive AV option. Considerable disparity was also observed across socio-demographic groups and trip contexts, with a sizeable segment of potential early adopters (19-27%) already displaying positive perceptions towards AV services.At the same time, we estimate the value of time (VOT) across conventional and three AV modes (private AV, AV ride-hailing, and AV minibus). Results suggest that, compared with conventional cars, travel time is perceived as significantly more useful (30%) in private AVs, implying a lower disutility of travel time. However, this advantage diminishes as the mode becomes more shared, with AV ride-hailing exhibiting a VOT similar to that of manually driven cars, whereas AV minibuses exhibit a 47% higher VOT. The findings, therefore, indicate that automation alone does not eliminate the perceived inconvenience associated with shared travel. Taken together, the combined evidence on ROC and VOT provides a robust behavioural basis for assessing AV adoption and evaluating transport policies.