12:37, 20 January 2026

Russia Develops Rescue Drone That Can Navigate Without GPS

Engineers at the Moscow Aviation Institute are building a navigation system that allows search-and-rescue drones to fly and locate people when GPS signals are unavailable.

At the Moscow Aviation Institute, researchers are developing a navigation system for search-and-rescue drones that can operate and locate people even when GPS signals are absent or unreliable.

The technology combines computer vision, inertial sensors (a gyroscope and accelerometer), and machine-learning algorithms. Instead of relying on satellites, the drone determines its position using its own motion data and images captured by an onboard camera.

This approach prevents the drone from getting lost indoors, in dense urban environments, tunnels, or disaster zones where satellite navigation signals are unavailable or unstable.

Reality Through Sensing

Pyotr Ukhov, an associate professor of computational mathematics and programming and deputy head of the MAI IT Center, explained how the system works.

“We’ve created a system that enables a drone to ‘understand’ where it is by relying on its own ‘sensations’ in the form of an inertial system combined with computer vision. A dedicated camera acts as the drone’s eyesight, while inertial sensors handle motion perception,” Ukhov said.

Mapping Itself

At the core of the technology is the SLAM method, which enables simultaneous mapping of the surrounding environment and self-positioning within it. Put simply, the drone builds a map of its surroundings as it flies and continuously refines its own location. The method has long been used in robotics and autonomous vehicles and is now being adapted for unmanned aviation.

Neural Network for Search

A key feature of the MAI project is the integration of a neural network. Its task is to analyze the video feed and detect people in real time. As a result, the system can serve both navigation and search-and-rescue operations. According to the developers, similar technologies are used abroad in drones designed to inspect damaged buildings and industrial sites.

Proprietary Controller

Unlike many existing solutions, the navigation system is fully integrated into the drone’s own flight controller rather than being housed in an external module. This reduces power consumption because the system no longer requires a separate power supply.

“The project was designed to be scalable from the outset. That makes it possible to adapt the technology for different types of drones and tasks—from territory monitoring to operations in emergency zones,” Ukhov noted.

The system is currently undergoing field tests. A team of students and engineers is refining the algorithms so the drone can automatically keep a detected person in view and determine whether a repeat flyover is needed. The developers plan to complete the project in the second quarter of 2026.

Science and new technologies