Robust autonomous vehicle navigation using adaptive interacting multiple model estimator

The Global Positioning System has been widely used for autonomous navigation applications in dynamic environments. Recently, an interacting multiple model estimator was tried to adapt for the improvement of the GPS positioning performance in various uncertain dynamic conditions. The estimation performance of an interacting multiple model estimator, however, may be degraded conspicuously when the actual motions of a vehicle are in discord with the motion models of the filter bank of the interacting multiple model estimator. In order to complement this shortage, this paper presents an efficient and robust navigation algorithm which integrates an interacting multiple model with a dynamic-free estimator in a form of analytic solution. Computational simulation clearly shows that the proposed navigation algorithm provides robust estimates within bounded errors whenever the autonomous vehicle's motions are incongruous with the motion models of the interacting multiple model estimator's filter bank in dynamic environments.