A comparison of four filters that use angle data from a fixed target

Abstract

Four filters used to smooth the angle-only data received by a sensor on a moving platform are compared. A least-squares filter and a Kalman filter in X-Y coordinates are compared. Two Kalman polar filters (a conventional one and a modified one) are included in the comparison. Comparative position errors are shown, and the shapes of the error curves as a function of measurement error and time are discussed. In general, the Kalman filters were easier to control and adapt to the environment and provided smaller position errors, particularly with larger measurement errors. The modified polar filter was better than the conventional polar filter because of a greater stability in the presence of large measurement errors. The modified polar filter was slightly better than the X-Y filter.<>