HTN planning for dynamic vehicle scheduling with stochastic trip times

Due to the stochastic nature of traffic flow and passenger demand at stops, bus service is constantly subject to delays and disruptions. Therefore, dynamic vehicle scheduling is essential to reduce the negative effects of service disturbance. This paper proposes a novel hierarchical task network (HTN) planning approach for dynamic vehicle scheduling with stochastic trip times. In the approach, a hybrid dynamic control strategy is devised to achieve headway adherence. Experimental results demonstrate that the proposed approach can flexibly handle a variety of abnormal operating conditions and maintain a stable headway. Due to the consideration of stochastic trip times, the HTN-based dynamic vehicle scheduling can simulate real bus operations. It can not only flexibly simulate different scenarios, such as inserting new trips or vehicle breakdown, but also easily test different control strategies. This will assist dispatchers handling disruption agilely and enhance the service quality of public transport.