Inertial Sensors and Their Applications - Signal Processing Systems - page 78

Tabl e 1 Example specifications for IMUs of different quality grades
Parameter

Unit Navigation Tactical Consumer

Accelerometer Pre-calibration bias m g 0 . 050 1 10

Scale factor error % 0 . 002 0 . 03 1

Misalignment mrad 0 . 05 0 . 5 -

Cross-axis sensitivity % - - 1

◦ /h

Gyroscope Pre-calibration bias 0 . 005 1 1 000

◦ /h

Bias instability 0 . 0035 1 25

◦ / √ h

Angular random walk 0 . 005 0 . 125 2

. .

Scale factor error % 0 001 0 015 1

. .

Misalignment mrad 0 01 0 5 -

Cross-axis sensitivity % - - 1

0.01 a

IMU Weight kg 5 1

c m 3 30 a Vo l u m e 5000 500

Power consumption W 10 5 0.3

a The physical dimensions of a MEMS IMU depend on the readout electronics etc

Consumer or automotive grade sensors are not capable of autonomous nav-
igation, but can be used for positioning temporarily, e.g., when satellite based
positioning is not available, such as when driving through an underpass. Consumer
grade sensors are primarily installed for other purposes than navigation; examples
of applications are given in Sect. 1.3 .
Tab le 1 shows example specifications of different grades of inertial measurement
units (IMUs); the values are examples only and should not be used as a definition
of the different quality levels. Anyway, it is clear that the gap between consumer
and navigation grades is large—the differences are in the order of many decades.
Misalignment errors have not been specified for consumer-grade units because it is
difficult, if not impossible, to separate their misalignment errors from other cross-
coupling effects such as inter-channel electromagnetic interference; hence, the total
cross-axis sensitivity is given for these IMUs instead. It should also be noted that
the size of a MEMS IMU depends on many factors, such as the readout electronics;
the sensor chip itself can be significantly smaller than the example values given in
Tab le 1 . Anyway, it is clear that MEMS instruments are smaller than, e.g., optical
gyroscopes by orders of magnitude.
When considering the performance parameters and requirements of sensors, it is
important to distinguish between errors before calibration and residual errors [ 25 ] .
For instance, the large bias of a consumer gyroscope can be mostly compensated for
by frequent calibration (e.g., whenever the IMU is stationary), but the bias instability
ultimately determines the attainable performance. On the other hand, with high-
quality IMUs it may be possible to calibrate out misalignment errors to an accuracy
better than the physically achievable sensor alignment precision.