SparkFun Forums

[Schema]

I've been experimenting with various IMUs like the MPU9150, but I'm finding it simply isn't up to the task. It's noisy, non-linear, and not repeatable. I've been using the excellent RTIMULib and calibrated the magnetometers, but it's still not what I need. I'm using it indoors which may be an issue.

You get what you pay for. I have a bigger budget for this project, so I'm looking for an IMU that can give me...

-Output of absolute orientation in Quaternions
-10Hz update rate
-Accurate within 5 degrees
-Repeatable within 2 degrees
-Ideally, serial output (to interface to an XBee/XN-XV)
-Physically, smaller than a deck a cards if possible

Sparkfun's highest-end seems to be the Razor IMU, is it up to the task? Other alternatives? Hoping to spend under $200, but could stretch it to $500 if absolutely necessary.

[Cannibal]

Hi there,

With all things inertial, there are serious limitations placed on what the systems can do based on what sort of external information you can feed to them, as well as what periods of time are being considered. For example, if you have an external update which can tell you your heading accurately every few seconds, even cheap inertial sensors can probably handle the updates over the in between periods. However, if you have one update per hour of heading, only very expensive tactical or strategic grade inertial units will do (and those can have ITAR restrictions).

Magnetometers as an update are relatively nice when outdoors or in wood framed houses, but in buildings that are concrete+steel, or around machinery they alternate between 'ok' and 'totally misleading'. Detecting when it's being a compass and when it's being a random number generator is an unsolved problem.

If you can stick the sensor on the shoe you can get away with low cost hardware and get pretty good performance for periods of up to 15 minutes, *but* the heading accuracy of 5 degrees may not be met, nor may the 2 degree repeatability.

http://www.openshoe.org/

Best of luck.