Kalman-Filter-Based Vessel Position Prediction for Isolated Island Routes with Long Sampling Intervals

Abstract

We are aiming for predicting and imputation missing data when we use Kalman Filter for estimating vessel position. When dealing with a isolated island sea route like this one, the sampling interval for data acquisition may be long due to problems with equipment, etc. In this case, we would only use state extrapolation. In such case, when predicting the next position data, the accuracy of the prediction step alone will be greatly reduced if the sampling interval is long. Then, in this paper, we propose vessel position prediction method that uses sea route predefined for a vessel, and report on the improvement in the accuracy of position prediction under certain conditions.