Global Positioning System Reference
In-Depth Information
used to increase reliability. In fact, the use of redundant inertial components for fail-
ure detection and isolation predates GPS RAIM work by nearly two decades [3].
9.2.2 Inertial Sensor Performance Issues
Before addressing the sensors utilized in all inertial systems, a few remarks about the
distinction between the two essential types of inertial systems are needed. INSs can
be broadly classified as either gimbaled or strapdown [4]. The basic distinction
between the two lies in the method by which the coordinate frame utilized for navi-
gation is maintained: In gimbaled systems, the frame is mechanized physically by
preserving a platform that is generally either the navigation frame itself or a frame
related to the navigation frame by a known transformation (e.g., the azimuth in a
wander azimuth mechanization of a gimbaled system). The platform is usually kept
locally level (i.e., level with respect to the horizon), where the accelerometers are
able to directly sense the horizontal components of host vehicle acceleration. How-
ever, use of a so-called space stable gimbaled orientation (e.g., as was used for the
Space Shuttle's inertial system) is an example of a gimbaled system that is not locally
level. To summarize, in a gimbaled inertial system, the sensors are maintained in a
preferred orientation and generally isolated from the vehicle's changes in attitude. In
a strapdown mechanization, on the other hand, the instruments are fixed in the vehi-
cle (e.g., along the nose of an aircraft, out the left wing, and with third axis complet-
ing the set). The navigation frame is maintained mathematically, not physically, by
the calculation of a transformation between the vehicle's body frame (where the
instruments reside) and the navigation frame. This transformation is most com-
monly referred to as a direction cosine matrix, but its mechanization is usually as a
quaternion or rotation vector [4] for improved efficiency.
The relative advantages and disadvantages of the two types of systems are fairly
well known. The gimbaled systems tend to be more expensive, due to the additional
hardware required for maintaining the physical platform, while the computational
requirements for the strapdown system (largely for maintenance of the direction
cosine matrix) are higher. Historically speaking, gimbaled systems were used almost
exclusively over the last several decades in navigation systems where accuracy was a
significant driver, while strapdown systems were relegated to applications with very
short flight times (e.g., a missile interceptor problem). However, advances in micro-
processor and inertial sensor technology have changed this trend, making
strapdown inertial systems the selection in most applications except those with the
most demanding requirements (e.g., submarine use). Microprocessor improvements
have made the high-rate computation of the direction cosine matrix relatively easy,
and the advent of optical gyros (i.e., ring laser and fiber optic) has produced designs
without the significant acceleration sensitivity of their mechanical counterparts.
This is quite important since the strapdown sensors see the full vehicle dynamics,
which
leads
to additional
errors
relative
to their
gimbaled
counterparts in
high-dynamic applications.
Returning now to the inertial sensors, there are two types, gyroscopes and accel-
erometers . The output of a gyroscope is a signal proportional to angular movement
about its input axis (
), and the output of an accelerometer is a signal proportional
to the change in velocity sensed along its input axis (
∆θ
∆ν
). A three-axis IMU would
 
Search WWH ::




Custom Search