Skip to content
Blog

What is inertial navigation (INS)? IMU, drift and limits

Published:

Inertial navigation (INS) estimates a vehicle's position, velocity and orientation from accelerometer and gyroscope measurements alone. It needs no outside signal, so it keeps working when GPS is unavailable, but its error grows over time. That trade-off shapes how every GPS-denied system is built.

What is an IMU?

An inertial measurement unit (IMU) combines accelerometers, which measure specific force along three axes, and gyroscopes, which measure rotation rate about three axes. Many IMUs used on drones are MEMS devices: small, light and inexpensive.

An IMU on its own only reports how the vehicle is accelerating and rotating. Turning those measurements into a position is the job of the inertial navigation algorithm.

How inertial navigation works

The algorithm first integrates the gyroscope readings to track orientation. Using that orientation, it rotates the accelerometer readings into a navigation frame, removes gravity, and integrates acceleration once to get velocity and again to get position.

This process is called dead reckoning: each new estimate is built on the previous one. It runs at a high rate and responds instantly to motion, which is why the IMU is the backbone of most navigation systems.

Why inertial navigation drifts

Because position comes from integrating measurements twice, small sensor errors grow quickly. The main sources are:

  • Bias: a small constant offset in the accelerometer or gyroscope that slowly changes with temperature and time.
  • Noise: random variation in each measurement that accumulates as random-walk error.
  • Scale factor and alignment errors: the sensor reads slightly more or less than the true value, or its axes are not perfectly aligned.
  • Orientation error: a small tilt error makes part of gravity look like horizontal acceleration, which turns into position error.

How drones limit inertial drift

In practice an IMU is rarely used alone for long. Other sources correct its drift through sensor fusion, usually in a Kalman filter:

  • A camera, through visual odometry or VIO, constrains velocity and position.
  • A barometric altimeter bounds vertical error.
  • A magnetometer helps keep heading stable.
  • Absolute references such as terrain-referenced navigation or GNSS, when available, reset accumulated error.
GPS-denied navigation: methods compared →

Frequently asked questions

An IMU is the sensor that measures acceleration and rotation. An INS is the complete system: the IMU plus the algorithm that turns those measurements into position, velocity and orientation.

It depends on sensor quality. With the low-cost MEMS IMUs common on small drones, position error grows quickly, so they are normally combined with other sources. Higher-grade inertial sensors drift much more slowly but cost and weigh more.

Yes. Inertial navigation uses only on-board sensors and needs no outside signal, which is why it is the backbone of GPS-denied navigation. Its limitation is drift, not dependence on signals.

Dead reckoning is estimating the current position from a known earlier position plus measured motion since then. Inertial navigation is a form of dead reckoning, as are wheel odometry and visual odometry.