Public transit systems play an important role in the alleviation of traffic congestion in urban road networks. The same vehicle type and a fixed departure timetable are usually applied to a bus route in the conventional public transit systems. They fail to cater to the time-varying travel demand or the diversified characteristics of transit passengers. To this end, this study proposes a demand-responsive public transit (DRPT) system consisting of a fixed bus route and demand-responsive stops with multiple vehicle types. The vehicle types of dispatched buses and the ride-matching schemes are optimized to serve transit passengers in real-time. Due to the non-convexity, Deep Q-Network (DQN), a reinforcement learning (RL) algorithm, is applied to the dynamic dispatching problem in the proposed DRPT system. The numerical studies validate the advantages of the proposed DRPT system and the RL-based dispatching algorithm.
Reinforcement Learning Based Demand-Responsive Public Transit Dispatching
21st COTA International Conference of Transportation Professionals ; 2021 ; Xi’an, China
CICTP 2021 ; 387-398
2021-12-14
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
A Threshold Policy for Dispatching Vehicles in Demand-responsive Transit Systems
DOAJ | 2019
|Metropolitan Transit Agency's Experience Operating General-Public Demand-Responsive Transit
Transportation Research Record | 2013
|Metropolitan Transit Agency's Experience Operating General-Public Demand-Responsive Transit
Online Contents | 2013
|Fleet scheduling and dispatching for demand-responsive passenger services
Online Contents | 2002
|