Inertial measurement unit (IMU) sensor plays a crucial role during the airborne phase of an unmanned aerial vehicles (UAVs), especially when passing through a precision-controlled area in performing their missions, such as wind turbine inspection or formation flight. To overcome this challenge, the statistics and control theory based unscented Kalman filter (UKF) exploits a deterministic sampling technique known as the unscented transformation (UT) and sigma point to estimate the non-linear system. Quadrotor UAVs are considered as a traditional system containing high complexity non-linear state. Therefore, this paper studies the IMU sensor fusion and predicts the flight trajectory through the UKF method, and treat them as a random bias issue. First, the kinematic model is established to obtain the Euler angles on the attitude of roll, pitch, and yaw for the first calculation and calibration. Subsequently, the noise and simulated fault are joined for computation. In the future, predictions from our ongoing experiments can be utilized to determine the threshold value for triggering a fault alarm.