MEMs IMU Calibration Bill Vaglienti, Five by Five Development Page 1
MEMs Inertial Measurement Unit Calibration
1. Introduction
Inertial Measurement Units (IMUs) are everywhere these days; most particularly in smart phones and
other mobile or handheld devices. These IMUs are typically constructed from MEMs
inertial sensors.
Although the sensors are inexpensive they suffer from accuracy problems. This document describes a
simple process for calibrating an IMU to improve its accuracy.
2. Errors in the sensors
An IMU consists of six individual sensors: three orthogonal angular rate gyros, and three orthogonal
accelerometers
.
2.1. Individual sensors
Each individual sensor is characterized with a linear equation:
1)
Where:
Output is the data of interest in rad/s for the gyro, or m/s
2
for the accelerometer.
Measurement is the raw reading from the sensor (in Volts or counts).
Bias is the sensors offset in the same units as the measurement.
Gain is the linear conversion from the raw reading to the output, with units (for example) of
rad/s/Volt.
Each individual sensor has a sensitive axis, where load (acceleration or rate of rotation) on that axis
produces the highest output, and load on the two orthogonal axes produce zero output.
2.2. Sensor triad cross coupling and alignment
Together the three gyros or three accelerometers are called a triad. Ideally the sensitive axes of the
three sensors are perfectly orthogonal and aligned to the reference axes of the IMU. In reality
manufacturing limitations prevent this ideal from being realized. Imagine an individual sensor (let’s use
the x-axis rate gyro as an example) whose physical alignment has the sensitive axis tilted towards the Y
axis by an angle α and towards the Z axis by an angle β. This sensor would now give a measurement in
response to rotation about Y and Z as well as X according to:
Micro Electromechanical Machines: small mechanical devices that are built using the same deposition and
photolithographic techniques used to make integrated circuits.
Despite their name accelerometers do not measure total acceleration. Like everything else an accelerometer
cannot “feel” a body force such as gravity. Hence an accelerometer is actually a specific force sensor. One must
add gravity to get the total acceleration.
评论3