Unscented Kalman Filter Method for Speed Estimation Using Single Loop Detector Data

Abstract

The accurate estimation of flow speeds with single loop data is critical to freeway traffic management. Many existing methods cannot accurately estimate speed because of difficulty in determining vehicle lengths and because some assumptions can be applied only to certain limited conditions. Moreover, excessive experimentally determined parameters some proposed approaches difficult to implement. A new unscented Kalman filter method of speed estimation is presented. The algorithm of the proposed method is implemented and evaluated with the use of field data from Texas Transportation Institute's vehicle detector test beds on State Route 6 in College Station, Texas, and on Interstate 35 in Austin, Texas. Estimated speeds are compared with observed speed data and with results from other estimation methods. The results indicate that the proposed method has excellent estimation accuracy and outperforms other methods. In addition to its superior accuracy, the proposed method is fairly easy to implement and practical for real-system implementation.