22
UM001
3.3.6
World Gravity Model
The world gravity model (WGM) is a large spatial-scale representation of the Earth’s gravity potential as a
function of position on the globe. The internal model used on the VN-100 is consistent with the Earth
Gravity Model (EGM96), which consist of a spherical-harmonic expansion of the Earth’s geopotential. By
default the world gravity model on the VN-100 is turned off, allowing the user to directly set the reference
gravity vector. Control of the world gravity model is performed using the Reference Vector Configuration
Register.
3.4
Vector Processing Engine
The Vector Processing Engine (VPE) is a collection of sophisticated algorithms which provide real-time
monitoring and simultaneous estimation of the attitude as well as the uncertainty of the input
measurements used by the attitude estimation algorithm. By estimating its own input measurement
uncertainty the VPE is capable of providing significantly improved performance when compared to a
traditional statically tuned EKF AHRS attitude estimation algorithm. The estimated measurement
uncertainty is used too in real-time at the NavFilter rate (default 200 Hz) adaptively tune the attitude
estimation Kalman filter. This adaptive tuning eliminates the need in most cases for the user to perform
any custom filter tuning for different applications. It also provides extremely good disturbance rejection
capabilities, enabling the VN-100 in most cases to reliably estimate attitude even in the presence of
vibration, short-term accelerations, and some forms of magnetic disturbances.
3.4.1
Adaptive Filtering
The VPE employs adaptive filtering techniques to significantly reduce the effect of high frequency
disturbances in both magnetic and acceleration. Prior to entering the attitude filter, the magnetic and
acceleration measurements are digitally filtered to reduce high frequency components typically caused
by electromagnetic interference and vibration. The level of filtering applied to the inputs is dynamically
altered by the VPE in real-time. The VPE calculates the minimal amount of digital filtering required in order
to achieve specified orientation accuracy and stability requirements. By applying only the minimal amount
of filtering necessary, the VPE reduces the amount of delay added to the input signals. For applications
that have very strict latency requirements, the VPE provides the ability to limit the amount of adaptive
filtering performed on each of the input signals.
3.4.2
Adaptive Tuning
Kalman filters employ coefficients that specify the uncertainty in the input measurements which are
typically used as “tuning parameters” to adjust the behavior of the filter. Normally these tuning
parameters have to be adjusted by the engineer to provide adequate performance for a given application.
This tuning process can be ad-hoc, time consuming, and application dependent. The VPE employs adaptive
tuning logic which provides on-line estimation of the uncertainty of each of the input signals during
operation. This uncertainty is then applied directly to the onboard attitude estimation Kalman filter to
correctly account for the uncertainty of the inputs. The adaptive tuning reduces the need for manual filter
tuning.
3.4.3
VPE Heading Modes
The VectorNav VPU provides three separate heading modes. Each mode controls how the VPE interprets
the magnetic measurements to estimate the heading angle. The three modes are described in detail in
the following sections.
