Aerospace and Electronic Systems Magazine October 2017 - 46

Electronics and Vision System of a Projectile-Drone Hybrid System
the deep enhancement of the Kalman
filter to fit the GLMAV application. In
addition, errors of the GPS and sensors
are better modeled, which improves the
angle stability and optimizes the filter
convergence speed.

The 3D Motorized Roll-Pitch-Yaw
Device
A 3D motorized roll-pitch-yaw device
was designed and manufactured at ISL
to reproduce the vehicle motions during its flight and to complete the sensor
calibration of the IMU when embedded
in the GLMAV. The device has the ability of positioning the central mounting
support with 0.2° of accuracy. Static
positioning tests allow the calibration
of the accelerometers and magnetometers, as well as the misalignment between the IMU and the body frame.
Figure 13 shows a picture of the GLMAV mounted in the device.

Figure 12.

Nonlinearity of the accelerometers of the IG-500N-B-G4A3.

Optimization of the Data Fusion filter
The IMU merges several sources of information (speeds, acceleration, GPS data, and barometer) to calculate the best estimate of attitude, speed, and position of the vehicle. The mathematical algorithm used by the IG-500N to make this estimate is an extended
Kalman filter adapted to inertial navigation. This algorithm is based
on a relatively simple principle of statistics, but its implementation
and in particular the setting up of the various parameters are very
complex. The accuracy improvement of the IG-500N required an
almost complete rewriting of the algorithm to better take into account the modeling of sensor errors and the vehicle dynamics.
The vibration tests allowed the enhancement of the Kalman
filter. Hence, the complete acquisition and processing chain of the
accelerometers and gyroscopes was revised to better filter the vibration phenomena. The rotation speeds and accelerations are integrated at a rate of 1 kHz instead of 100 Hz to limit and to better
optimize the quality of the angles and the position calculated by
the IMU and, in particular, in the presence of oscillatory motions.
Gyroscopes and accelerometers are very sensitive sensors and
bias (offset of zero) may change over time and environmental conditions. For example, an accelerometer bias drift of 10 mGal was
measured after a gun launching of 2500 g of acceleration. The bias
of the gyroscopes and accelerometers are to be estimated by the
Kalman filter to provide an angle estimation accuracy better than
the degree. This estimation was significantly improved and, in
particular, when the IMU receives GPS correction data. The latest
algorithms are able to estimate a gyroscope bias better than 40°/h
and an accelerometer bias less than 0.5 mGal.
Each application is very different, and it is impossible to use the
same Kalman filter set in the same way for all vehicles (boat, car,
aircraft, and helicopter). The GLMAV is a very particular vehicle,
as it passes from a projectile configuration to a configuration similar to a helicopter. The new Kalman filter uses specific dynamic
profiles for each vehicle, which means that the tuning of the filter
is different from one application to another. This principle allows
46

COMPLETE VISION SYSTEM WITH IMAGE TRANSMISSION
The focus is on an image pickup device in two orthogonal directions, processing and management of the software installed in the
vehicle and in the ground station, and the transmission of images
and control commands to and from the ground station by radio link.

Figure 13.

GLMAV mounted in the 3D motorized roll-pitch-yaw device.

IEEE A&E SYSTEMS MAGAZINE

OCTOBER 2017

Table of Contents for the Digital Edition of Aerospace and Electronic Systems Magazine October 2017