Skip to content
Blog

GPS-denied navigation: methods compared

Published:

GPS-denied navigation means estimating an aircraft's position and velocity when satellite signals are unavailable, blocked or unreliable. No single method solves it on its own; practical systems combine several on-board techniques, each with its own strengths and failure modes.

Why GPS is not always available

Satellite navigation signals arrive at the receiver very weak. They can be blocked by buildings, terrain or foliage, degraded by interference, or deliberately disrupted by jamming and spoofing. Indoors, underground and in dense urban canyons the signal may never be usable at all.

An aircraft that depends only on GPS loses its position the moment the signal goes. GPS-denied navigation is the set of techniques that keep a usable position estimate from the aircraft's own sensors.

The main methods

The methods below differ in which sensors they use, how fast their error grows and what the environment has to offer:

  • Inertial navigation (INS): integrates accelerometer and gyroscope measurements from an IMU. Works anywhere and needs no outside signal, but its error grows steadily over time.
  • Visual odometry (VO): estimates motion by tracking features between consecutive camera images. Low drift over short distances; needs light and texture.
  • Visual-inertial odometry (VIO): fuses camera and IMU data. The IMU covers fast motion and recovers scale; the camera limits IMU drift.
  • SLAM: builds a map while localizing in it and can correct accumulated error when it recognizes a place it has seen before (loop closure).
  • Terrain-referenced navigation (TRN) and scene matching: compare what the aircraft measures or sees with a stored elevation model or reference imagery to get an absolute fix.
  • Barometric and magnetic aids: an altimeter bounds vertical error and a magnetometer helps with heading, though both need care around disturbances.

Comparing the methods

A useful way to compare them is by the kind of error they leave behind. Relative methods such as INS, VO and VIO estimate motion step by step, so small errors accumulate into drift. Absolute methods such as TRN and scene matching give an independent position, but only where the terrain or imagery is distinctive and a reference is available.

Environment matters as much as the algorithm. Visual methods struggle at night, over water or snow, and in fog. TRN needs terrain with enough relief. Inertial navigation is indifferent to the environment but cannot correct itself.

Sensor fusion: putting them together

Because each method fails in different conditions, robust systems fuse several of them, typically with a Kalman filter or an optimization-based estimator. The IMU provides a high-rate backbone, visual methods limit its drift, and absolute fixes reset accumulated error when they are available.

The design question is not which method is best, but which combination fits the aircraft's size, power budget, flight profile and the environments it will fly in.

IRA-2: GPS-denied navigation kit →

Frequently asked questions

There is no single most accurate method. Accuracy depends on the environment and flight time: visual methods drift little over short distances in good light, while absolute methods such as terrain-referenced navigation bound error over long distances where terrain is distinctive. Combining methods through sensor fusion gives the most consistent result.

For short periods, yes. An IMU alone gives position by integrating acceleration, but the error grows quickly with low-cost sensors. Over longer flights an IMU needs help from other sources such as a camera, an altimeter or an absolute reference.

GPS is one satellite system; GNSS is the general term covering GPS, GLONASS, Galileo and BeiDou. GNSS-denied is the broader and more precise term, but the two are often used interchangeably.

It depends on the method. Inertial and terrain-referenced methods do not need light. Camera-based methods need enough light or a sensor suited to low light, so night operation is usually handled by combining methods.