Travel time modeling for bus transport system in Bangalore city

Abstract

The aim of this work is to model travel time of buses in Bangalore city. Travel time estimation and prediction is inevitable for planning and operation of bus transport system. Moreover, accurate travel time prediction is an important requirement in most of the Intelligent Transport System (ITS) applications. Previous approaches on travel time prediction have been mostly done in “lane disciplined” and “homogeneous traffic conditions.” In this work, an attempt is made to predict travel time of public transport buses using Kalman filtering technique in Indian conditions where the traffic is heterogeneous and not lane disciplined. The additional challenge faced in the Indian conditions is the unavailability and/or ineffectiveness of automated data collection devices such as loop detectors. Data collection was done in Bangalore, India for different times of a day and also for different days of a week for understanding the travel time variation on different days. A prediction model was developed using Kalman filtering technique. The validation of the results shows good comparability of modeled values with the actual travel time values.