Istanbul 2004
*ceiver is the
in orientation
J is largely
bed by a bias
and the angle
distinguished
e per month)
r degree per
' degree per
following.
d in the early
z (1990) for
- land-vehicle
y integrating
iccelerometer
by including
and azimuth)
igh GPS was
oy using GPS
2 enough to
perational the
ated with a
em were the
see Cosandier
for Remote
-Sheimy and
ental systems
land vehicles.
t that time is
of the geo-
as due to the
GPS-receiver
t IMU's that
zu the Tatter
| in the early
le systems,
type. When
and attitude
saged at that
derable price
apidly falling
me the most
ystem. Since
€ bulk of the
ence of low-
ition to this
yf solid state
ompensation
ut 10-2 ms-2
US$ 10 000
rtion and the
»bviously not
of their rapid
rovided with
erential GPS
ld be kept in
he integrated
m accuracy
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part BS. Istanbul 2004
applications.; for details on system design and performance, see
Báumker and Matissek (1992), Lipman (1992), Bader (1993),
and Knight et al (1993) among others.
With the rapid improvement of fibre optic gyro performance,
the sensor accuracy of a number of these systems has improved
by about an order of magnitude (1 deg/h and 10-1ms-2) in the
past five years. Typical cost are about USS 30 000. Beside the
increased accuracy, these systems are more user friendly and
offer a number of interesting options. When integrated with a
DGPS phase solution the resulting position and attitude are
close to what is required for the high-accuracy class of
applications. When aiming at highest possible accuracy these
systems are usually equipped with a dual-antenna GPS, aligned
with the forward direction of the vehicle. This arrangement
provides regular azimuth updates to the integrated solution and
bounds the azimuth drift. This is of particular importance for
flights flown at constant velocity along straight lines, as is the
case for photogrammetric blocks. Commercialization of the
mobile mapping system concept for all application areas has
been donc by the Applanix Corporation (now a subsidiary of
Trimble — www.applanix.com). In general, the position and
orientation accuracy achieved with these systems is sufficient
for all but the most stringent accuracy requirements; for details
see chapter 7)
During the past few years a new technology has rapidly
changed manufacturing processes in engineering, specifically in
sensor design and telecommunications. It is called MEMS
technology for its products, which are Micro Electronic
Mechanical Systems (MEMS). Accelerometers and gyros are
among the early products manufactured in this way. They are
micro-machined and, when produced in large quantities, will be
extremely inexpensive. Current prices per sensor range from
USS 20 — 150, depending on accuracy, but predictions are that
they will get into the range of dimes rather than dollars. T The
inertial sensors produced until recently by MEMS were for the
mass market and were of poor quality when compared to
navigation-grade inertial sensors. Gyros had constant drift rates
of thousands of degrees per hour. However, results recently
presented at IEEE PLANS 2004 indicate that companies are
actively working on MEMS-based tactical gyros, see for
instance Hanse (2004) and Geen (2004). Considering that the
production processes for MEMS inertial sensors are relatively
new and that the improvement potential is considerable, can it
be expected that at some point in the future the accuracy of
these sensors may be sufficient to support navigation-type
applications? At this point it is not possible to answer this
question in an unequivocal way. However, two arguments will
be given, one in favor, the other against. They may be helpful
for forming an opinion.
The argument against is based on some interesting empirical
results that the authors received by courtesy of Dr. Robert J.
Smith at the Honeywell Technology Center. They have been
partly reproduced in Figure 1. Gyro performance (measured by
long-term bias stability) is plotted vs. the nominal size of the
gyro on a log-log scale. It should be noted that the figure is not
based on a comprehensive market analysis, but is an in-house
study conducted by Dr. Smith. This is the reason why only
Honeywell gyros are shown. The gyros represented in this
figure vary in terms of size (between 120 mm and 4 mm) and
principle used (RLG, ESG, HRG, FOG, 2DF rotor, QRS). All,
*Xcept one, are production-type systems. Each gyro is
lepresented by an ellipse showing the performance range in
horizontal direction and the variability in size in vertical
761
direction. It is remarkable that the line N=4 gives such a close
fit to most gyros presented in the chart. This indicates that gyro
accuracy, independent of the principle used, is determined by
the size of the sensor. The gyros above the line fit are typically
not pressing the state of the art, because of other considerations
(cost, lifetime). For the one gyro below the line, the H-ESG,
which seems to outperform the general trend, only bias stability
values in a benign temperature environment were available. It
might therefore not be directly comparable to the other
performance values which cover a wide range of production
environments.
Excluding these special cases, the N=4 line can be considered
as an empirical law for gyro performance which is independent
of the principle used to build the gyro. This means that it can be
used as a predictor for gyro performance in cases where the size
of the gyro is given by other considerations. When applying this
principle to the MEMS gyro environment it would mean that a
gyro with a nominal size of 2mm would perform at the 10 000
deg/h level, while a tactical grade gyro with a performance of 1-
10 deg/h should have a minimal size of about 20-mm. Chip size
will essentially limit the accuracy of the IMU-on-a-chip.
Similarly, the likelihood that MEMS-based gyroscopes will
reach navigation-grade performance is tied to the nominal size
of the gyro which has to be about 6 cm to achieve the
requirements.
Gyro Scaling Laws: Bias Stability vs. Size
Power Law: B = 1/D" (normalized ta D=100, B=0.001)
1000.0
| Low performance & large |
Other performance advantages?
a RLG IFOG Lower cost? Lifetime? |.
E 100.0 0 i E ATE EE Re RR
E exiis REG. HRG
a EN THEE ex» FOG (tactical)
5 H-ESG
s (stable temp RLG —. «m 2DF Rotor
10.0 & platf E oci
s on QRS
t (fmm) OR >
2 (7 mm) QRS—. |
2 1.0 dmm) NA
i. Beyond state of the art = |
0.1
0.0001 0.001 0.01 0.1 1.0 10.0 100.0 1000.0 10000.0
Long-Term Bias Stability, B [Deg/Hr]
Courtesy: Honeywell
Figure 1: Bias Stability vs. Nominal Size for Mature Gyro
Technology.
The argument in favor of MEMS gyro usage for navigation-
type applications is based on publications recently presented at
the IEEE PLANS 2004 and results obtained by the Mobile
Multi-Sensor Research Group at the University of Calgary,
Canada. The latter were obtained in a land-vehicle test using a
MEMS-based IMU developed by employing off-the-shelve
MEMS sensors with an average cost 20$ per sensor (see Figure
2). The test also included the Honeywell CIMU, a navigation
grade inertial navigation system, and DGPS. Both DGPS and
CIMU trajectories were available throughout the whole test and
were used as an accurate reference for the MEMS IMU results.
Inertial measurements of MEMS sensors were integrated with
the single point positioning GPS output (accurate to 10 -30 m)
and processed through the INS Tool Box (Shin and El-Sheimy,
2003) Kalman filter software. GPS signals from a minimum of
seven satellites were available throughout the test. In order to
assess the performance of the integrated system, GPS outages
of 30 seconds were simulated, by removing GPS data, along
