Convergence of extended Kalman filter to locate a moving target in wild life telemetry

In wildlife telemetry and several other applications, some or all of the measurements consist of angles made with very poor accuracy. In these cases classical extended Kalman filtering, where the angles are treated as linear variables, tend to introduce errors in estimation on a sporadic basis due to incorrect handling of the angles. We propose a correction term in the Kalman filtering equation which will ensure that the angular terms are treated in an invariant fashion