Abstract
Robo-taxis or shared-use automated vehicle-enabled mobility-on-demand services (SAMSs) are now in operation in the US and China. By removing drivers’ labor costs, SAMSs promise to provide significantly lower-cost transportation than human-driven mobility-on-demand services. Under this assumption, prior research indicates SAMS can provide sizable employment accessibility benefits to workers. The current paper aims to analyze the distribution of SAMS accessibility benefits across segments of the population (i.e., perform an equity analysis) using an agent-based travel modeling approach. The study’s methodology (i) clusters workers by their socio-demographic and -economic characteristics using latent class analysis, (ii) estimates hierarchical work location and commute mode choice models for four worker segments, and (iii) obtains logsum-based monetary measures of accessibility for each worker in a synthetic population of Southern California. Using this information, we analyze the distributions of SAMS accessibility benefits across several population segmentations. We utilize box plots to visualize the distributional differences across population segments. Additionally, we use ANOVA and post-hoc Tukey’s Honestly Significant Difference tests to analyze the overall and inter-group statistical significance of the distributional differences, respectively. The results indicate that low-income, Black, and Hispanic workers receive larger SAMS accessibility benefits on average than high-income, White, and Asian workers. Additionally, workers in zero-car households benefit more from SAMSs than one- and multi-car households, particularly after conditioning on the transit accessibility of the worker’s residence. The study also aggregates the agents into their origin census tracts, classifies the census tracts based on agent socio-demographic attributes, and then analyzes the distribution of SAMS accessibility benefits across census tract designations (e.g., low median income tracts vs. high median income tracts). The study finds that if analysts were to make individual-level inferences based on the spatial analysis, the inferences would be inaccurate in the case of household income and age, thereby misinforming policymakers regarding who benefits more/less from SAMS.
