THESIS
2020
viii leaves, 31 pages : illustrations ; 30 cm
Abstract
Invent of GPS enabled mobile technology has made dynamic ridesharing popular in these
recent years. In dynamic ridesharing program, a platform (mobile app run by a ridesharing
company) finds matching between riders and drivers using dynamic matching algorithm taking
real-time GPS location of riders and drivers, origin and destination of trips, flexible start time
of trips as argument. Then rider is notified with contact of matched driver to decide the
convenient pickup point through conversation. Usually this pickup point is near to the rider’s
origin so that rider don’t need to walk much. For a shared ride with multiple passengers, a
driver usually has to pick up passengers from different pickup locations one by one which leads
to an increase in detour time for the driver and f...[
Read more ]
Invent of GPS enabled mobile technology has made dynamic ridesharing popular in these
recent years. In dynamic ridesharing program, a platform (mobile app run by a ridesharing
company) finds matching between riders and drivers using dynamic matching algorithm taking
real-time GPS location of riders and drivers, origin and destination of trips, flexible start time
of trips as argument. Then rider is notified with contact of matched driver to decide the
convenient pickup point through conversation. Usually this pickup point is near to the rider’s
origin so that rider don’t need to walk much. For a shared ride with multiple passengers, a
driver usually has to pick up passengers from different pickup locations one by one which leads
to an increase in detour time for the driver and first passenger in most cases. Meeting points
are introduced with a view to avoiding extra detour time and increasing matching probability,
where riders might need to walk to that meeting point from where driver can pick up all
passengers at once given riders and driver satisfy time and distance feasibility.
To understand the complex interaction among the stakeholders of ridesharing system with
meeting points, this study has proposed an optimization based simulation model that matches
riders and drivers continuously at a constant interval in a rolling horizon way. In our
framework, we consider solo and dedicated drivers simultaneously and willingness of rider
whether joining meeting point or not. The initial feasible match list obtained from match
finding algorithm are passed into the optimization problem that determines the best match
ensuring system-wide maximum driving distance saving as well as the maximum number of
matched participants. A numerical study with the Didi GAIA dataset shows that the increase of
number of meeting points improves occupancy rate and matching rate initially and stabilizes
after further increase. Sensitivity test with the increase of maximum allowed walking distance
with a fixed number of meeting point results in improvement of objective value in the
optimization.
Post a Comment