Abstract: In this work, Kalman filter is designed with the help of C programming language and compared experimentally based on the actual Encoder values with Complementary estimation for Inertial Measurement Unit (IMU) sensor fusion. As IMU sensor, ADXL345 accelerometer and L3G4200D gyroscope are utilized in this analysis. For the actual rotation, the Quadrature encoder is applied with Arduino UNO by interrupt reading. Complementary filter is intended for human arm movement and Kalman filter is emphasized derived from the experimental covariance such as Qangle for accelerometer trust level, Qbias for gyroscope and Rmeasure for the measurement. The sable solution to this system is obtained via a chart display and root mean square error checking method. For the various changing to Kalman coefficient, the experimental test is performed and optimal estimation could be founded out at 0.001, 0.1 and 0.0001 for Qangle, Qbias and Rmeasure for Kalman filter.