Particle Filter for Reliable Bus Travel Time Prediction Under Indian Traffic Conditions

In recent times, traffic congestion has been increasing rapidly and deteriorating the quality of traffic systems in urban areas of many developed and developing countries. This became a serious problem faced by society, as many people are using private vehicles while commuting from one place to the other. One of the reasons people are shifting towards private transportation is due to lack of reliability of the public transportation systems. Attracting more travelers towards public transportation using Intelligent Transportation Systems (ITS) technologies is one way to reduce the negative impacts. In this context, prediction of bus travel time and providing information about bus arrival time to passengers accurately is a potential solution, which will help in reducing the uncertainty and waiting time associated uncertainties with public transit systems. However, for this solution to be effective, the information provided to passengers should be highly reliable. The present study proposes a model based prediction method that uses particle filtering technique for accurate prediction of bus travel times for the development of a real time passenger information system under heterogeneous traffic conditions that exist in India. The results obtained from the implementation of the above method are validated using the measured travel time. The prediction accuracy is quantified using the Mean Absolute Percentage Error (MAPE) and the performance is compared with a base approach namely, the historic average method. The quantified error in terms of MAPE is 20% for the proposed method and 37% for the historic average method, indicating the superiority of the proposed method over historic average method. Thus, it can be concluded that particle filter is a viable tool in the prediction of bus travel times.

Language

  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01642011
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Jul 26 2017 4:21PM