Navigation Functionalities for an Autonomous UAV Helicopter
Navigation Functionalities for an Autonomous UAV Helicopter
Navigation Functionalities for an Autonomous UAV Helicopter
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
5.1. FILTER ARCHITECTURE 63<br />
north, east <strong>an</strong>d down velocity error; roll, pitch <strong>an</strong>d heading error), <strong>an</strong>d<br />
three accelerometer biases modeled as first order Markov processes. The<br />
Kalm<strong>an</strong> filter algorithm estimates the errors of the INS mech<strong>an</strong>ization by<br />
fusing the estimate of these errors provided by <strong>an</strong> internal linear model <strong>an</strong>d<br />
the measurements from the vision system. The theory behind the Kalm<strong>an</strong><br />
filter c<strong>an</strong> be found in [12].<br />
The linear model used <strong>for</strong> the INS errors is represented by the system<br />
of differential equations in 5.2:<br />
δ ˙r = −ωen×δr+δv<br />
δ ˙v = −(ωie+ ωin)×δv−ψ×f+δa (5.2)<br />
˙ψ = −ωin × ψ<br />
δ ˙a = −βδa<br />
where δr is the position error, δv is the velocity error <strong>an</strong>d δψ is the<br />
attitude error. δa represents the three accelerometer biases. In addition<br />
ωen is the rotation rate vector of the navigation reference system relative<br />
to the Earth reference system, ωin is the rotation rate vector of the navigation<br />
reference system relative to the inertial reference system <strong>an</strong>d ωie<br />
is the rotation rate vector of the Earth reference system relative to the<br />
inertial reference system. The update of the filter is per<strong>for</strong>med when a<br />
new measurement from the image processing system is available. The raw<br />
measurement delivered by the image processing is the 3D helicopter position<br />
relative to the pattern. The image processing measurement covari<strong>an</strong>ce<br />
used in the filter is scheduled with the dist<strong>an</strong>ce to the pattern. The relation<br />
between the uncertainty of the vision measurement <strong>an</strong>d the dist<strong>an</strong>ce to the<br />
pattern is found in Paper III.<br />
The Kalm<strong>an</strong> filter is implemented in discretized <strong>for</strong>m <strong>an</strong>d the recursive<br />
implementation of the discretized Kalm<strong>an</strong> filter equations c<strong>an</strong> be found<br />
easily in the literature. A detailed implementation of a nine state Kalm<strong>an</strong><br />
filter c<strong>an</strong> be found in [23].