Abstract: A method for estimating target parameters of a target with respect to a platform such as ownship includes assigning predetermined initial values for the target parameters, wherein the target parameters are part of models having respective model probabilities. The models may include a Kalman filter. The models are updated in response to measured parameters of the target, wherein the measured parameters are different from the ones to be estimated. After a predetermined number of updates, the model having the highest updated model probability may be selected as the winning model. The updated values of the target parameters in the winning model at the time of winner selection are used as the estimated values for the parameters sought to be estimated. The updated parameter values may be refined by restarting the process using the updated values of the winning model as the basis for assigning new initial values for the target parameters.