The last mile in a public transport trip is known to bring a large disutility for passengers, because the conventional transport modes for this stage of the trip can, in many cases, be rather slow, inflexible and not provide a seamless experience to passengers. Fully automated vehicles (AVs), that is, those which do not need a driver, could act as a first mile/last mile connection to mass public transport modes. In this paper, we study a system that we call Automated Last-Mile Transport (ALMT), which consists of a fleet of small, fully automated, electric vehicles to improve the last mile performance of a trip done in a train. An agent-based simulation model was proposed for the ALMT whereby a dispatching algorithm distributes travel requests amongst the available vehicles using a FIFO sequence and selects a vehicle based on a set of specified control conditions (e.g. travel time to reach a requesting passenger). The model was applied to the case-study of the connection between the train station Delft Zuid and the Technological Innovation Campus (Delft, The Netherlands) in order to test the methodology and understand the performance of the system in function of several operational parameters and demand scenarios. The most important conclusion from the baseline scenario was that the ALMT system was only able to compete with the walking mode and that additional measures were needed to increase the performance of the ALMT system in order to be competitive with cycling. Relocating empty vehicles or allowing pre-booking of vehicles led to a significant reduction in average waiting time, whilst allowing passengers to drive at a higher speed led to a large reduction in average travel time, whilst simultaneously reducing system capacity as energy use is increased.
|Journal||International Journal of Transportation Science and Technology|
|Publication status||Published - 2017|
- Automatic vehicles
- Public transport
- Feeder service
- Agent-based simulation