Download PDF Button


Attitude is all about heading, roll and pitch. GNSS-only systems with multiple receivers and antennas are able to calculate all three; but when you lose satellite connection, you lose all that data. That’s not a problem if the task is open sky all the time. But, what happens in urban environments with tall and reflective buildings or agricultural sights obstructed by heavy foliage.

For these tasks, the operator would need a GNSS receiver with an inertial measurement unit (IMU) with its 6 degree of freedom accelerometers and gyros to produce continuous roll, pitch, azimuth and velocity measurements even when satellite connection is lost. Until recently, a typical GNSS/INS system, the  kind used on tactical systems such as aircraft, were well out of the price point for conventional mapping or agricultural applications, upwards of tens of thousands just for the IMU.

That’s changed considerably in recent years, thanks in large part to the continuing advancement of low-cost, high performance microelectro-mechanical systems (MEMS) IMU sensors. MEMS IMUs are orders of magnitude cheaper, smaller and easier to deploy, making it possible for operators in cost-sensitive applications such as ground based surveys
or agricultural tasks to gain positioning advantage. Low cost MEMS IMUs are even incorporated into ‘backpacks’ that map hiking trails with a 3D camera.

Of course, there’s a catch. While buying a complete system with a MEMS IMU incorporated is likely the best option, what if you want to buy one yourself? In essence, are all MEMs created equal?

The answer is no. Today’s market is saturated with low cost MEMS—but no standard way to measure and compare performance. Manufacturer specifications are necessarily skewed to feature the best aspects of their solutions and price point is not a reliable indicator.

There are common specifications such as the in-run bias stability (the amount the sensor drifts over time). However, there are several ways to specify a MEMS in-run bias stability. Is it a  room temperature specification or is it done in a temperature chamber? Is it an Allan Variance or is it compared against a calibrated control? How long and over how many units did the qualification occur?

Other specifications will have similar statistical variations. Is the specification quoting Angular Random Walk or Noise? What are the quoted units? Bottom line, read the fine print and ask the questions.

There is really only one way to determine if your $50 or $2000 MEMS IMU is effective for your application—and that’s to try it out. A better option is to rely on manufacturers who thoroughly vet MEMS IMU sensors before they are deployed in GNSS receivers.