Hybrid Navigation Method of INS/PDR Based on Action Recognition

Abstract

In allusion to indoor passive low-cost shoe-mounted personal navigation system (PNS), inertial navigation system (INS), and pedestrian dead reckoning (PDR) each has strengths and weaknesses. In this paper, an INS/PDR hybrid navigation method (IPH) is introduced based on action recognition aiming at large position divergence caused by heading drift. The PDR algorithm for improved heading estimation (IHE) is proposed. It can provide pseudoheading measurement according to the result of action recognition, so it can restrain the heading angle drift effectively. Besides, the INS/PDR IPH is proposed. In the situation of normal walking, the proposed method introduces the corresponding PDR pseudoposition measurement information to the PNS filter for correction. The proposed method not only can reserve the comprehensive and independent navigation information of INS but also can incorporate PDR location information with high precision when walking normally. Thus, the combination of the advantages of INS and PDR is achieved without adding any sensors. The effectiveness of the method is verified by the motion experiments. The navigation results of the PDR algorithm based on IHE are improved by around 57% compared with the traditional PDR. In addition, the navigation results of the INS/PDR hybrid method are improved by 42% on average compared with the INS.