Prediction of Path Travel Time Using Kalman Filter

On-line application adaptability of Kalman filter to predict path travel times is investigated. The filter predicts link travel times for an hour with 5 minute intervals from probe vehicle data on two paths of Seoul Metropolitan where travel times randomly fluctuates all the time. Then, two different types of path travel times, instantaneous and experienced path travel times, are estimated and compared with real path time data. The filter quickly follows the fluctuation trend of the real path travel times. The result shows that the filter maintains a sound application possibility with minimal computation time. More paths in different areas need to be studied in next.