A ZUPT method based on SVM regression curve fitting for SINS

Zero velocity update (ZUPT) is an important error control method for strap-down inertial navigation system (SINS). Conic fitting as a frequently-used ZUPT method is simple and effective, but makes low use of the velocity samples gained during the stops of SINS, and requires SINS to stop frequently. In order to solve these problems, a new ZUPT method based on support vector machine (SVM) regression curve fitting for SINS is proposed. The simulation results demonstrate that the new ZUPT method can improve the positioning accuracy and maneuverability of SINS, and has high robustness compared with conic fitting.