GNSS-Denied Navigation for UAS

Satellite navigation is convenient, but not always available. The signal can degrade in urban areas, near metal structures, in gorges, indoors, under canopies or due to interference. Therefore, critical UAS missions require an architecture that is not dependent on a single coordinate source.

Navigation without GNSS does not mean magical autonomy. This is a set of sensors and algorithms that estimate position relative to the environment, trajectory, map or starting point. Such a system requires design, configuration and testing.

Start with the mission requirements

First you need to understand why the system should work without GNSS: to continue the route in the event of a short-term loss of signal, to carry out a mission in a closed area, to remain stable in the face of interference, or to ensure accurate positioning relative to an object.

• determine the permissible position error;
• understand the duration of work without a satellite signal;
• describe the environment: city, forest, premises, industrial facility;
• set requirements for return and emergency modes;
• understand whether a map is needed or the system builds it itself.

Components of a complete solution

Commonly used IMU, barometer, magnetometer, camera, optical flow, lidar, altimeter, terrain map and sensor fusion algorithms. Each source provides incomplete information, and the system’s task is to collect a stable estimate of the position.

• inertial sensors provide a fast response, but accumulate errors;
• camera and visual odometry are dependent on texture and lighting;
• lidar helps in geometrically rich environments;
• the map increases stability if the environment is known in advance;
• Filtering algorithms combine data and evaluate its credibility.

Key engineering constraints

The main risk is the accumulation of errors and degradation of individual sensors.

• identical surfaces impair visual navigation;
• dust, rain, snow and haze reduce the quality of sensors;
• lidar and camera increase mass and consumption;
• the magnetometer may make mistakes near metal;
• Without testing, it is impossible to honestly assess the duration of the autonomous section.

Practical application

In the working project, navigation without GNSS is tested in stages: test bench, short flights, degradation scenarios, error control, emergency modes, and only then a regular mission.

It is important for the operator to see not only the coordinate, but also confidence in the navigation. If the system understands that the quality of the assessment is falling, it should go into safe mode, and not continue the mission as if everything is normal.

Common mistakes

Errors arise when GNSS-free is treated as a separate checkbox in the characteristics.

• do not set the acceptable operating time without GNSS;
• consider visual odometry as universal for any environment;
• do not test the algorithm on real routes;
• do not take into account the computational load;
• do not design emergency scenarios.

Checklist before starting the project

Before implementation you need to describe:

• where and why GNSS is not available;
• what error is acceptable;
• what sensors are available in terms of mass and power supply;
• what environmental conditions are critical;
• how the quality of navigation is checked;
• what does the UAS do when the assessment is degraded.

Practical conclusion: Navigation without GNSS requires a system architecture. Reliability comes not from a single sensor, but from a combination of sensors, algorithms and proven failure scenarios.