M/sup 3/ filter with adaptive mode probabilities control for low sampling rate tracking

This paper proposes the Multiple Maneuver Model (M/sup 3/) filter with recalculation of residual covariance in mode probabilities calculation. The conventional M/sup 3/ filter had a problem that mode probabilities oscillate in the long range radar with long sampling period. The oscillation is caused by extremely low likelihood functions of target models at the measurement in mode probabilities calculation. For controlling the oscillation under clutter and frequent miss detection environment, the M/sup 3/ filter that calculates the residual covariance in mode probabilities calculation is proposed. The residual covariance is calculated in advance using the overlap coefficient that indicates how error ellipses are overlapped. It is shown that the proposed M/sup 3/ filter has capability of controlling the oscillation without residual information and improving the tracking quality for maneuvering target.