Bus Information Service: Optimal Arrival Time Release Considering En Route Behaviour

Abstract

Bus operations are subject to various internal and external factors, making accurate prediction of bus arrival times inherently challenging. Furthermore, passenger behaviour en route is influenced by the information provided about arrival time. Consequently, how to release predicted bus arrival times to passengers becomes critical for Advanced Traveler Information Systems (ATIS). This study focuses on passengers' journey from receiving arrival time information to boarding, defined as the "catching process". To minimize the cost of the catching process for passengers, we develop an out-of-vehicle travel time cost model using the released bus arrival time as the decision variable, considering passenger behaviour feedback, i.e., the choice of short-term additional activity en route based on the received bus arrival time. The theoretical framework guides the optimal release time (ORT) within the bus arrival time prediction confidence interval. Under the given travel scenarios, numerical experiments reveal that the optimal release strategy reduces the expected out-of-vehicle time cost per passenger by 2.86 CNY (a 9.1% saving) compared to releasing the expectation value. Finally, we conduct sensitivity analyses of the factors affecting ORT. The study results show that arrival time prediction accuracy and trip purpose are critical in determining the optimal release strategy, whereas factors such as bus headway, the distance of the expected trip to the boarding stop, and passenger access distance from origination to the boarding stop exhibit relatively low sensitivity. The optimal release strategy will especially benefit ATIS in case of poor prediction accuracy performance and passengers needing strict demand for punctual arrival.

Keywords: Advanced Traveler Information Systems, Bus arrival time release, Behaviour feedback, Additional activity choices, Out-of-vehicle travel time cost