The Transfer–Collection Trade-off: When Does a One-Seat Ground–Air Ride Beat Transfer-Based Air Mobility?

Advanced Air Mobility (AAM) promises to lift travellers above road congestion, yet most proposed services require passengers to reach a vertiport, transfer into an aircraft, and transfer again at the destination. These transfers, together with vertiport processing, can absorb much of the time saved in the air. This paper examines a mass-transit alternative, the Drive–Fly–Drive (DFD) bus. A road-capable vehicle collects passengers at neighbourhood stops, converts and flies the line-haul segment, then drives again to distribute passengers, so that each traveller keeps one seat from origin to destination. Removing the transfer is not free. The vehicle must tour the neighbourhood to fill its seats, and every additional kilometre of catchment radius adds collection time. We name this tension the Transfer–Collection Trade-off and formalise it with a continuum-approximation model of door-to-door travel time for four modes: a ground express bus, transfer-based AAM with ride-hail access, transfer-based AAM with feeder-bus access, and the DFD bus. Under illustrative assumptions, three results follow. (i) Among mass-transit options, the DFD bus is fastest from about 10 km up to a distance limited only by its air-speed penalty, for example about 250 km when that penalty is 20%. Collection radius plays no role in this comparison because every mass mode must collect. (ii) Against ride-hail-access AAM, each kilometre of collection radius costs about nine minutes. The one-seat ride therefore wins only in compact catchments unless vertiport processing approaches airport-like durations. (iii) Because a high-capacity vehicle uses one take-off and landing slot for many passengers, the concept also eases vertiport throughput, which we argue is the binding constraint of future AAM networks. We map these results onto four corridor archetypes and invert them into minimum performance requirements for any future one-seat ground–air vehicle. Working paper (not peer-reviewed). A Korean abstract is included at the end of the PDF. All figures are reproducible with the accompanying model.py.