Investigating users' attitudes towards conventional and automated buses in twelve European cities

Driverless buses running in urban low-speed, mixed-traffic conditions are the subject of current research and demonstration in Europe. The paper aims to assess how automation fares with respect to conventional services in terms of users' attitudes. Stated preference data, based on a questionnaire administered in twelve cities, are used for the estimation of logit models providing preference shares for the conventional and the automated bus. The correlation among errors in repeated measurement data is tackled using a multivariate version of the Farlie-Gumbel-Morgenstern copula, yielding a closed form for the choice sequence probability. Estimation results show a relatively higher preference for automation across the cities where the automated bus is implemented inside a major facility. Without provision of information on how the system works and with no experience, users tend not to trust automation in the normal mixed-traffic conditions that are found in cities. The impacts of socio-economic variables are heterogenous across cities. Comparison between the logit based on the independence assumption and the logit with correlated observations shows deviations in coefficient estimates as large as twenty per cent with a typical sample size of about two hundred respondents and four choice tasks per individual. Deviations decrease to few percentage points when sample size is tripled. Deviations in preference shares are, however, negligible, a result which, in terms of policy implications, is indicative of the validity of the independence assumption in applied work.