The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008 
Therefore the H 2x 4 in Equation (10) is 
[0^(1 
0 
[Q y x(l-%)-<p y Yr ff -i%) 
0 
(13) 
The stochastic model of the Kalman filter and the parameters 
for its tuning are designed according to the sensors’ 
specifications and the system configuration in the field tests. 
(m/s 2 ) 
Bias<12mg, 
DGPS 
(Novatel RTK) 
20 
Hz 
N, E, Alt 
(m); 
V xv , V z (m/s) 
-2 cm 
-0.05 m/s 
Optic Flow 
50 
Hz 
q q 
(deg/s) 
Scale <2% 
LRF 
25 
Hz 
LR (m) 
scale error <1% 
offset <10cm 
Table 1. Data collected for the integrated navigation system 
4. SYSTEM CONFIGURATION 
4.1 UAV (Helicopter) Platform 
Experiments were conducted on an RMAX unmanned 
helicopter manufactured by Yamaha. It has 30kg payload with 
endurance of approximately one hour, and comprising a 
conventional helicopter design with a single main rotor and a 
single tail-rotor for anti-torque compensation. 
An autopilot system is devised for conducting closed loop 
experiments, in which a control by telemetry approach is taken. 
The sensor information is sent to a ground computer using a 
radio link, which processes the sensor information, calculates 
corresponding control inputs and sends these back to the 
helicopter via another radio link. An onboard PC 104 computer 
based on the Pentium III chipset is for vision processing 
(Garratt, 2007). 
4.2 Sensors 
The sensors onboard the helicopter includes DGPS, INS 
magnetometer, LRF and CCD camera (imaging payload). Their 
locations are indicated in Figure 3. The data collected for the 
proposed integrated GPS/INS/Vision navigation system are 
listed in Table 1. 
Figure 3. UNSW@ADFA RMAX in flight 
Sensors 
Data 
Rate 
Parameters 
Accuracy 
INS 
(IMU700CB) 
50 
Hz 
p q r (deg/s) 
a x a v a z 
Gyro: Scale <2%, 
Bias <20deg/hr; 
Acce: Scale <1% 
An analogue Sony CCD camera is used as the image sensor. 
The camera outputs a PAL standard composite video signal 
which is captured by the frame grabber. The inertial sensors for 
the RMAX were pre-fitted with an isolation system, including 
three gyroscopes and three accelerometers with orthogonal axes. 
Highly accurate carrier phase DGPS measurements were 
available to provide a monitoring system to record the 
helicopter motion during closed loop. Novatel OEM4-G2L GPS 
cards are mounted adjacent to the RMAX flight computers. The 
OEM4 cards are used with differential corrections from a 
nearby base station fitted with another OEM4 card. The card 
operates in a Real-Time Kinematic (RTK) positioning mode; to 
provide 20Hz position and velocity with an accuracy of l-2cm. 
Helicopters are dynamically unstable and require constant 
control inputs to prevent them from diverging from a level 
flying attitude. Accurate knowledge of attitude (pitch, roll and 
yaw) is therefore vital to robust control. The RMAX helicopter 
is fitted with an in-house system for attitude determination 
(Garratt, 2007). 
Figure 4. Laser scanning system 
A LRF with a novel rotating mirror was integrated with the 
RMAX helicopter as shown in Figure 4.1. Due to the 
orientation of the axis of the mirror and the angle of the mirror 
to the axis, the laser traces an elliptical cone shape on the 
ground below. To obtain a sufficiently fast scan rate, the mirror 
is spun at least 1500RPM or 25 cycles per second. The 
rangefinder and encoder signals are read into a PC 104 with a 
sample rate of 2 KHz. An AccuRange 4000 LRF from Acuity is 
used for this project, with range accuracy of each point on the 
ground is better than 2cm (Garratt et al., 2007) 
4.3 Data Processing 
The data are processed in integrated GPS/INS/Vision navigation 
system described in Section 3. Two Kalman filters (KF) are