Fleet Capacity and Operational Coverage in Autonomous Transportation for Rural Healthcare Access

Abstract

Transportation barriers can limit rural healthcare access even when appropriate services are geographically available. This study develops a data-driven stochastic operational simulation for healthcare demand-responsive transportation in the 15-county South Plains region of Texas. Public aggregate data describe tract populations and disability, healthcare facilities, and road-network travel times; patient trips, appointment availability, and current transit waiting times remain assumed rather than observed. Eight primary scenarios and seven controls or sensitivities were evaluated with 100 replications each. The eight-vehicle autonomous reference completed 77.2% of requests with a 10.5-minute mean wait. Reducing the fleet to four vehicles decreased completion by 12.1 percentage points, while 12 vehicles increased it by 2.8 points. Restricting eligible routes by a travel-time coverage rule decreased completion by 13.1 points. An equal-size hybrid with six autonomous and two human fallback vehicles recovered 11.2 points relative to restricted autonomous service. Standard, wheelchair, and assistance-request completion in the autonomous reference was 81.5%, 66.0%, and 48.4%, respectively. When route coverage was matched, human and autonomous fleets produced identical service outcomes; their modeled cost difference depended entirely on hourly labor or supervision assumptions and disappeared at $25 per autonomous operating hour. The findings identify fleet coverage and accessible-service capacity as pilot-design priorities.