What Is an IMU?

An IMU, or inertial measurement unit, is a sensor package that measures how an object accelerates and rotates. A typical IMU has a three-axis accelerometer, measuring linear acceleration along the x, y and z axes, and a three-axis gyroscope, measuring the rate of rotation about them. Together they are called a 6-axis IMU. Many add a three-axis magnetometer, which senses the Earth’s magnetic field as a heading reference, making a 9-axis IMU.

An IMU does not measure position directly. A processor integrates its readings: rotation rate gives orientation, and acceleration, once gravity is removed, gives velocity and then position. That process is called dead reckoning, and it works anywhere, with no external signal, which is exactly what satellite positioning cannot do.

The catch is drift. Small errors in the sensors add up as they are integrated, so an IMU’s position estimate wanders further from the truth the longer it runs alone. That is why IMUs are almost always combined with something that gives an absolute reference, most often GNSS.

Most IMUs today are MEMS devices: microscopic mechanical structures etched in silicon. A MEMS accelerometer measures how far a tiny proof mass moves on its springs; a MEMS gyroscope measures the Coriolis force on a vibrating structure as it turns. They are small, cheap and robust, which is why they are in every phone and drone.

Higher-performance IMUs use optical gyroscopes. A fibre optic gyroscope (FOG) sends light both ways around a coil of fibre and measures the tiny phase shift rotation causes; a ring laser gyroscope (RLG) does the same with laser beams in a sealed cavity. They drift far less, at a much higher cost.

IMUs are usually grouped by how stable their gyroscopes are, because gyro drift dominates how long the IMU can navigate alone. Figures vary by source and are approximate.

For most vehicle and robot work, an industrial MEMS IMU fused with GNSS is enough; survey and mapping systems move up to tactical grade to hold accuracy through longer outages.

The terms are related but not interchangeable:

GNSS alone gives absolute position but loses it in tunnels, deep cuttings, under bridges and between tall buildings. An IMU tracks movement between fixes, and fusing the two gives a smooth, continuous position that a mapping or inspection system can rely on. GNSS corrects the IMU’s drift; the IMU fills the gaps and smooths the noise in GNSS. Adding RTK corrections takes the combined system to centimetre level outdoors. Our sensor fusion FAQ covers how the estimates are combined, and GNSS vs GPS explains the satellite side.

The IMU also tells the system which way cameras and sensors are pointing, which matters for mapping and for stabilising images, and it records shocks and vibration along a route.

Getting good results depends as much on installation as on the sensor. Mount the IMU rigidly, close to the vehicle’s centre of rotation where possible, and measure its offset from the GNSS antenna and cameras accurately, because the fusion software uses those lever arms in every calculation. Isolate it from high-frequency vibration, such as an engine or pump, which can saturate MEMS sensors. Temperature changes shift MEMS bias too, so better IMUs are calibrated across their temperature range at the factory.

IMUs are in drone flight controllers, where they keep the aircraft level hundreds of times a second (see how a Jetson connects to a flight controller), in vehicle navigation and ADAS, in robots and AGVs, in rail and road survey systems and in antenna and camera stabilisers. The railway track inspection case study uses GNSS and IMU cards inside one of our railway computers to position every image along the track.

We fit GNSS and IMU modules in rail, vehicle and edge AI computers as standard, set up the drivers and time sync, and test the inputs before each system ships. Tell us the accuracy you need and the environment, and we will specify the sensors and computer.