Improved position estimation by fusing multiple inaccurate inertial measurement unit sensors

This paper presents improved position tracking and location estimation of dead reckoning, by combining the data from multiple inaccurate inertial sensors together using unscented Kalman filtering (UKF). Experimental test results using two 9-axis inertial measurement unit (IMU) sensors show that position estimation of each sensor achieves a 26 % decrease in max error (ME) and a 37% improvement in root-mean-square error (RMSE), when compared with a single independent sensor.