Understanding Heterogeneity in Autonomous Taxi Preferences in Luxembourg: A Latent Class Choice Model Approach

Abstract

Autonomous taxis and shared autonomous mobility are expected to transform urban transport systems, yet their adoption depends on user acceptance, particularly of ride-sharing services. This study investigates preferences for conventional taxi, autonomous taxi, and shared autonomous taxi for home-to-airport trips in Luxembourg, with a focus on preference heterogeneity. Using data from a stated preference experiment with 1,373 respondents, we estimate both a Multinomial Logit (MNL) model and a Latent Class Choice Model (LCCM) to identify distinct behavioral segments. The results show that cost, travel time, and waiting time significantly influence choices, with waiting time having a stronger impact than in-vehicle time. The LCCM reveals two segments: a privacy-sensitive group that is more cost-sensitive and strongly reluctant to use shared autonomous taxis for longer trips, and a sharing-oriented group that is more open to shared autonomous mobility. Trip length plays a critical role, with shared options becoming less attractive for longer journeys. These findings highlight the importance of accounting for preference heterogeneity and suggest that shared autonomous mobility may face adoption barriers for longer, purpose-driven trips such as airport access.