The aim of this study is to establish the method for improving the performance of body motion analysis by inertial sensors. The position vector must be measured before the experiment in body motion measurements using inertial sensor. In the previous study, we developed the estimation method for the position information of the attached sensors. The method estimates the position vectors from the joint position to the sensor system by the centrifugal and tangential accelerations included to the accelerometer output. The method is established by the Extended Kalman filter. However, the method must hold on the base link for estimation of the position vectors. In this study, we developed a new estimation method of the position vectors from the joint to the sensor systems. This method is established by focusing to local coordinate system. The Unscented Kalman fileter algorithm is used for the estimation of the position vectors from the joint to the sensor using the two inertial sensor systems attaching the body segments. The proposed method can be used in actual body motion measurement. The method can be used to the development of more simply motion measurement method and the improvement of accuracy in the motion analysis by inertial sensor system.