UAV Navigation in GNSS Denied Environments

July 4, 2025

Unmanned Aerial Vehicles (UAVs) are increasingly deployed in mission-critical scenarios where conventional satellite-based navigation systems, such as GPS, become unreliable or unavailable. These GNSS-denied environments pose significant challenges for accurate and continuous state estimation. This work presents a robust navigation solution based on sensor fusion, integrating data from an Inertial Measurement Unit (IMU) and monocular Visual Odometry (VO) through an Extended Kalman Filter (EKF). The system is designed to maintain accurate estimates of position, velocity, and orientation even in the absence of GNSS signals. The VO module extracts camera motion by detecting and matching visual features between sequential frames, while the IMU provides high-rate inertial data for motion propagation. The EKF fuses these data streams, correcting drift and enhancing robustness. The proposed method is validated using both the EuRoC MAV dataset and the AirSim simulation environment. The results demonstrate improved accuracy compared to standalone IMU or VO systems, with Root Mean Square Error (RMSE) significantly reduced across multiple sensor configurations. The system shows promise for low-cost, infrastructure-independent navigation in challenging environments.

Authors in the community:

• 

  D

  Diego Zarco

  ist198096

Supervisors of this institution:

• 

  A

  Alexandra Moutinho

  ist13959

• 

  A

  Afzal Suleman

  ist13672

mechanical-engineering - Mechanical engineering

eng - English

Instituto Superior Técnico