- July 11, 2017 at 8:13 pm #12359G. RoboticsParticipant
An Inertial Measurement Unit (IMU) is a common electronic component found in all sorts of modern embedded devices. This unit varies in degrees of freedom, which states how many sensors come attached. Some of the most popular IMUs come in 9 DOF (Degrees of freedom), which consists of a three-axis accelerometer, three-axis gyroscope, and three-axis magnetometer. However, how much of the IMU is actually useful when used in production?
While measurements of acceleration, tilt, and magnetic orientation or fields may be used as standalone units, they are almost always found packaged into their compressed form, which is the IMU. However, the usefulness of such a compact device is questionable, as there are rare applications which need all three forms of data provided by the IMU.
In the case of our particular project, Rover uses an IMU mainly for the magnetometer, as the robot needs to be able to orient itself to magnetic north, and uses this alignment in its subroutine of tracking. While the magnetometer provided microTesla* readings of nearby magnetic fields, a simple atan2 function easily converts those field readings to magnetic bearings.
More on atan2 functions: https://en.wikipedia.org/wiki/Atan2
*1 µT (microtesla) = 10 mG (milligauss)
The accelerometer and gyroscopic units largely have remained untouched during our development, due to the particular solution we’re taking with our project. The question is: Are there projects which would require the usage of all three types of sensors? Besides very precise vehicular products (cars, drones, boats, etc.) and fairly complex robotics projects with control systems, it seems that the IMU provided more than it’s useful for.
Thus, we must ask why these three sensors are consistently packaged into combined units instead of sold separately. Is it perhaps a convenience issue, where all three types of sensors use a similar underlying technology, and as thus makes it easy to wire them in conjunction? Or is it a product which is developed for applications which aren’t as frequently researched or developed now? Are we growing past the needs of IMUs, and now desire more compact, exact circuits with only the exact units we need?
I personally consider the IMU to be a nifty little sensing device, but overall its usage in design will largely be limited to only part of its complete circuit. While the trend in the industry right now is to create large sensing units with lots of peripherals which might or might not be used, I believe large-scale product development hinges on the efficiency of production, and that requires only getting the components which are absolutely necessary. As such, IMUs are destined for hobbyist researchers and prototypes, never destined to make the cut into final product development.
Care to disagree? Have you managed to come up with a cool project combining all three types of sensor information? Let us know!
Thank you to Yonah Elorza for this piece. Yonah is Goddard Robotics and Rover’s principle engineer.
You must be logged in to reply to this topic.