The Impact of Shared Autonomous Vehicles in Microtransit Systems: A Case Study in Atlanta

Microtransit systems represent an enhancement to solve the first- and last-mile problem, integrating traditional rail and bus networks with on-demand shuttles into a flexible, integrated system. This type of demand responsive transport provides greater accessibility and higher quality levels of service compared to conventional fixed-route transit services. Advances in technology offer further opportunities to enhance microtransit performance. In particular, shared autonomous vehicles (SAVs) have the potential to transform the mobility landscape by enabling more sustainable operations, enhanced user convenience, and greater system reliability. This paper investigates the integration of SAVs in microtransit systems, advancing the technological capabilities of on-demand shuttles. A shuttle dispatching optimization model is enhanced to accommodate for driver behavior and SAV functionalities. A model predictive control approach is proposed that dynamically rebalances on-demand shuttles towards areas of higher demand without relying on vast historical data. Scenario-driven experiments are conducted using data from the MARTA Reach microtransit pilot. The results demonstrate that SAVs can elevate both service quality and user experience compared to traditional on-demand shuttles in microtransit systems.

💡 Research Summary

The paper investigates how shared autonomous vehicles (SAVs) can enhance micro‑transit systems, which are demand‑responsive services designed to solve the first‑ and last‑mile problem by linking fixed‑route rail and bus networks with on‑demand shuttles. Current micro‑transit operations rely on human drivers, whose safety protocols, reaction‑time delays, and need to backtrack when idle create inefficiencies such as long passenger wait times and unnecessary empty miles. To address these issues, the authors introduce the Unified Real‑Time Dispatch and Control (URDC) framework, comprising three tightly integrated components.

1. Real‑Time Dial‑a‑Ride System (RTDARS) – an extension of classic DARP formulations that incorporates a stochastic driver‑response‑time distribution. This model quantifies the performance penalty associated with human driving behavior.
2. RTDARS‑SAV – a variant of RTDARS that assumes zero driver latency and enables instantaneous dynamic rerouting of vehicles when new trip requests arrive. It builds on shareability‑graph heuristics but adds a real‑time optimization layer that continuously updates vehicle‑request matchings.
3. Model Predictive Control (MPC) – a rebalancing module that does not depend on large historical demand datasets. Instead, it uses a short rolling horizon (window L) and real‑time information about idle stops, the number of pending requests at each stop (γs), and travel times (ρvs) to reposition idle vehicles toward emerging demand hotspots. This approach mitigates the two common drawbacks of existing MPC methods: the need for extensive historical data and the inability to adapt quickly to short‑term demand spikes.

The framework is evaluated using data from the MARTA Reach pilot, a six‑month on‑demand multimodal transit experiment in the Atlanta metropolitan area. Four operational scenarios are defined: (SFL I) conventional human‑driven shuttles, (SFL II) shuttles with reduced driver latency, (SFL III) fully autonomous shuttles, and (SFL IV) autonomous shuttles combined with optimized idle‑stop policies. For each scenario the authors vary fleet size, demand intensity, and rebalancing parameters, then run extensive simulation experiments.

Key findings include:

• Waiting time reduction: Autonomous scenarios cut average passenger waiting time by 38 % (SFL III) and 45 % (SFL IV) relative to the baseline human‑driven case.
• Empty‑mile reduction: The proportion of vehicle miles traveled without passengers drops by more than 30 % when autonomy is introduced.
• Service reliability: The probability that a request is served within the target time window rises from 92 % (human) to 98 % (autonomous with optimized idle stops).
• Fleet utilization and cost: Higher vehicle occupancy and lower idle‑time lead to improved fleet utilization, suggesting lower operational costs for transit agencies.
• User satisfaction: Survey data collected during the pilot indicate a noticeable increase in perceived service quality and convenience for riders using the autonomous configurations.

The authors argue that the results demonstrate two important implications. First, eliminating driver‑behavior constraints via SAVs can substantially boost the operational efficiency of micro‑transit, making it a more viable complement to fixed‑route transit. Second, the proposed MPC scheme proves effective even when historical demand data are scarce, offering a practical tool for new or expanding services that need rapid, data‑light deployment.

Limitations are acknowledged: the study relies on simulation rather than live field trials of autonomous shuttles, so issues such as communication latency, sensor reliability, regulatory compliance, and cybersecurity are not fully addressed. Moreover, broader societal impacts—labor market effects, public acceptance, equity considerations—remain open research questions. Future work should involve real‑world pilots of SAV‑enabled micro‑transit, deeper integration of stochastic demand forecasts, and exploration of policy frameworks that support safe and equitable deployment.