Abstract:
This paper analyzes the performance of an
Extended Kalman Filter (EKF) based attitude estimator for a
Quad Rotor Unmanned Aerial Vehicle (QRUAV). A non-linear
dynamic model of the QRUAV, simulated in Matlab was
complemented with an EKF based attitude estimator. The
estimator fuses measurements obtained from simulated
accelerometers and gyroscopes onboard the QRUAV. It is then
shown that this estimator only results in adequate representation
of the attitude when the translational velocity is large. This
behavior is explained considering the dynamics of the vehicle and
the onboard accelerometers. Also a possible method of exploiting
this behavior for improving the estimation accuracy is suggested.
