Highway traffic state estimation and prediction considering measurement errors and traffic uncertainty

Most previous traffic state estimation and prediction approaches assume measurements from detectors are precise and represent traffic state of interest with a single value (e.g., the travel time from city A to B is 30-min). However, in most traffic systems, measurements are subjected to some level of errors. Further, single-value representation of traffic state does not adequately convey traffic uncertainty to road users nor effectively assist them in making departure time and route decisions. To deal with these two issues, this research uses intervals to represent uncertain measurement data and traffic states (e.g., the travel time from city A to B is between 25 and 35 minutes) and adopts the Interval Kalman Filter method to predict short-term highway traffic states. The method is applied to predict traffic density intervals on National Highway No. 5 in Taiwan. The result shows the proposed approach is effective in predicting highway traffic state intervals.