A multi-directional motion interacting fusion model for diver tracking

Authors

  • Feng Xu
  • Tao Wen
  • Yongqiang Ji
  • Juan Yang

Keywords:

moving diver, multi-direction motion model, time-varying transition probability, interacting fusion, parallel Kalman filter,

Abstract

According to the diver motion characteristics,which are low speed and rapid change of direction,a multi-directional motion model is presented.Then the motion model is introduced into aninteracting multiple model method, while the timevaryingmotion model transition probability wascorrected according to current measurements.Firstly, the predictive state was obtained by amulti-directional motion model. Secondly, theparallel Kalman filters were applied to estimatemulti-directional state. Finally, the interactivefusion processing for estimations from multidirectionalmotion model was conducted toimplement diver state estimation. The method wasverified by both simulation and experiment. Theresults show that the proposed method has highertracking accuracy and superior adaptability thanconventional interactive multiple model algorithmbased on single direction motion model. Theproposed method is effective for diver tracking.

Downloads

Published

2017-01-31