The main issue addressed here is that of estimating the kinematic state components of a vehicle in autonomous navigation using landmark angle-only measurements from an onboard passive sensor. The estimates of the absolute position and velocity of the vehicle are provided by a hybrid coordinate fusion filter. The hierarchical architecture of the filter which consists of a group of local processors and a global processor is developed for improving estimation accuracy. In each local processor, an extended Kalman filter uses hybrid information from the reference Cartesian coordinate system and the modified polar coordinate system for state and state error covariance extrapolation and updating. In the global processor, a weighted least squares estimator is utilized to combine the outputs of local processors to form a global estimate. By using only two landmarks simulation results show that proposed algorithm improves the estimation accuracy drastically.