This paper presents a fuzzy alpha-beta (a-b) filter as an intelligent vehicle tracking method. This filter is compared with other methods such as a-b filter and Kalman filter. Each method is briefly reviewed with its own properties. Simulation shows that a fuzzy a-b filter is as accurate as the Kalman filter in tracking vehicles. Considering computational cost, a fuzzy a-b filter has advantage over the Kalman filter because its basic structure is built on the a-b filter. Intelligence comes into play to modify the fixed-coefficients, a and b, to the current difference between measurement and prediction. The cause of failures in vehicle tracking are described and reconstruction of missing data is demonstrated using a fuzzy a-b filter.
Abstract