Improvement in the Orientation and Position Estimation of Quadrotor UAV in Dynamic Motion

Kalman filter was commonly used to estimate the orientation and displacement of the quadrotor UAV. Meanwhile, inaccurate estimation of displacement often occurred especially in dynamic motion. This error generally caused by noise and limited sampling rate of the sensors that cause the inaccuracy acceleration data fed to the filter. This is exacerbated by the inaccuracies of accelerometer data when used under dynamic conditions for orientation determination. However, the accuracy of orientation significantly influences the accuracy of estimated displacement. The use of different weights for each sensor as a function of changing dynamic conditions of the quadrotor cannot always result in an accurate estimated displacement. This paper proposed a new method to compensate the inaccuracy of acceleration data fed to Kalman filter when in dynamic motion. This is done by calculating the forward thrust when the quadrotor tilted its body to create a pitch angle. From the forward thrust, the additional acceleration is calculated which is compensated for the quadrotor acceleration. With this new method, the estimate displacement can be corrected and results in more accurate estimated displacement and trajectory.