Table of Contents
Advertisement
Guidance
Terminology
acceleration. This fact makes using raw accelerometer data difficult. A navX-sensor's automatic
accelerometer calibration determines the component of measured acceleration which corresponds to
gravity, and uses this information together with gyroscope readings to calculate a gravity vector, which
represents acceleration due to gravity. Pitch and Roll angles are derived from this gravity vector.
Once the gravity vector is understood, this value is then subtracted from the raw accelerometer data to
yield the acceleration due to linear motion.
Velocity and Displacement
Acceleration is defined as the change in Velocity. Therefore, linear velocity can be calculated by
integrating linear acceleration over time.
Velocity is defined as the change in Position, otherwise known as Displacement. Therefore, linear
displacement can be calculated by integrating linear velocity over time.
Important Note: Using currently-available MEMS-based accelerometers to calculate linear velocity and
displacement is subject to large amounts of error primarily due to accelerometer "noise" (a difference
between the actual acceleration and the measured acceleration inherent with MEMS sensors). This noise
not only accumulates, but is also squared in the case of velocity, and is cubed in the case of
displacement. The significant amounts of error in displacement values mean they are not typically useful
for robotic navigation; the amount of error in displacement estimation can be several feet per second. As
MEMS sensors improve in the coming years and accelerometer noise is reduced, this technique will
become more useful for robotics navigation.
If you would like to experiment with using the navX-sensor to calculate displacement and velocity, you
can use the navXUI's "Experimental" button to bring up a dialog which displays the integrated velocity
and displacement values calculated in real-time by the navX-sensor.
World Reference Frame
Raw acceleration data measures acceleration along the corresponding sensor axis. This measurement
occurs in a reference frame known as "Body Reference Frame". This works well as long as the navX-
sensor circuit board is in it's original orientation. However as the navX-sensor circuit board rotates (e.g,
as the robot it is mounted to rotates), the X and Y accelerometer axes no longer point "forward/back" and
"left/right" with respect to the original orientation. To understand this more clearly, consider how the
meaning of the term "left" changes once a robot has rotated 180 degrees? Introducing a World Reference
Frame solves this issue by providing a reference upon which to measure "leftness".
To account for this, a navX-sensor's motion processing adjusts each linear acceleration value by rotating
it in the opposition direction of the current yaw angle. The result is an acceleration value that represents
acceleration with respect to the area in which the navX-sensor operates, which is known as "World
Reference Frame". This world-frame linear acceleration value is much simpler to use for tracking motion
of an object, like a robot, which might rotate while it moves.
66
Advertisement
Table of Contents
