Preferences for shared automated vehicles: A hybrid latent class modeling approach

We model the mode choice between three novel emerging transportation services and identify latent market segments not studied yet in the context of shared automated vehicles, including ridesharing, car sharing and automated transit, using a hybrid choice model. We use a discrete distribution to capture taste heterogeneity of distinct latent classes. Latent variables, sociodemographics and travel habits inform latent class assignment estimated simultaneously with a discrete choice kernel. Respondents chose their preferred mode for going to work in a set of stated preference choice tasks, based on the attributes of their current commutes using a Bayesian DEfficient design. Users were segmented into two latent classes based on latent factors that capture time style orientation and public transit dislike. The effects of seating designation, not modeled previously in this context, trip cost and travel times in a shared ride, were estimated for the two classes. Users who neither like transit nor ridesharing with strangers are less likely to choose a shared ride if their designated seat is the middle seat, and overall less likely to choose automated transit. Individuals who have more organized time styles demonstrate higher marginal sensitivity to travel times and costs and are more likely to choose automated transit. Value of time analysis reveals that wait time of services that offer a convenient home pickup is valued lower than invehicle time. The implications for future adoption of shared automated vehicles is further discussed.

Automated summary

This study models the choice between different shared automated transportation services using a hybrid choice model, identifying two latent classes of users based on factors like time style orientation and seating designation. The research shows that factors such as time style and seating designation play a significant role in determining users' preference for shared rides and automated transit.