Assessing the impacts of automated mobility-on-demand through agent-based simulation: A study of Singapore. (August 2020)

• Record Type: Journal Article
• Authors: Oh, Simon; Seshadri, Ravi; Azevedo, Carlos Lima; Kumar, Nishant; Basak, Kakali; Ben-Akiva, Moshe
• Is Part Of: Transportation research. Volume 138(2020)
• Page Start: 367
• Page End: 388
• Publication Date: 2020-08
• Subjects: Automated Mobility-on-Demand (AMOD) -- Smart mobility -- Agent-based simulation -- Activity-based model; Transportation -- Research -- Periodicals; 388.011
• DOI: 10.1016/j.tra.2020.06.004
• Languages: English
• ISSNs: 0965-8564

Abstract

Highlights: Assessed impacts of AMOD through high-fidelty activity and agent-based simulation. Captured future demand using ABM that draws on data from smartphone-based SP survey. Modeled operations of AMOD fleet within multimodal mesoscopic traffic simulator. Simulated and evaluated future scenarios using a model of Singapore 2030. Introduction of AMOD may result the cannibalization of PT and increase congestion. Abstract: The advent of autonomous vehicle technologies and the emergence of new ride-sourcing business models has spurred interest in Automated Mobility-on-Demand (AMOD) as a prospective solution to meet the challenges of urbanization. AMOD has the potential of providing a convenient, reliable and affordable mobility service through more competitive cost structures enabled by autonomy (relative to existing services) and more efficient centralized fleet operations. However, the short and medium-term impacts of AMOD are as yet uncertain. On the one hand, it has the potential to alleviate congestion through increased ride-sharing and reduced car-ownership, and by complementing mass-transit. Conversely, AMOD may in fact worsen congestion due to induced demand, the cannibalization of public transit shares, and an increase in Vehicle-Kilometers Traveled (VKT) because of rebalancing and empty trips. This study attempts to systematically examine the impacts of AMOD on transportation in Singapore through agent-based simulation, modeling demand, supply and their interactions explicitly. On the demand side, we utilize an activity-based model system, that draws on data from a smartphone-based stated preferences survey conducted in Singapore. On the supply side, we model the operations of the AMOD fleet (including the assignment of requests to vehicles and rebalancing), which are integrated within a multimodal mesoscopic traffic simulator. Comprehensive simulations are conducted using a model of Singapore for the year 2030 and yield insights into the impacts of AMOD in dense transit-dependent cities from the perspective of the transportation planner, fleet operator, and user. The findings suggest that an unregulated introduction of AMOD can cause significant increases in network congestion and VKT, and have important policy implications that could potentially inform future deployments of AMOD.