Traditional taxi systems in metropolitan areas often suffer frominefficiencies due to uncoordinated actions as system capacity and customerdemand change. With the pervasive deployment of networked sensors in modernvehicles, large amounts of information regarding customer demand and systemstatus can be collected in real time. This information provides opportunitiesto perform various types of control and coordination for large-scaleintelligent transportation systems. In this paper, we present a recedinghorizon control (RHC) framework to dispatch taxis, which incorporates highlyspatiotemporally correlated demand/supply models and real-time GPS location andoccupancy information. The objectives include matching spatiotemporal ratiobetween demand and supply for service quality with minimum current andanticipated future taxi idle driving distance. Extensive trace-driven analysiswith a data set containing taxi operational records in San Francisco shows thatour solution reduces the average total idle distance by 52%, and reduces thesupply demand ratio error across the city during one experimental time slot by45%. Moreover, our RHC framework is compatible with a wide variety ofpredictive models and optimization problem formulations. This compatibilityproperty allows us to solve robust optimization problems with correspondingdemand uncertainty models that provide disruptive event information.
展开▼
