Autonomous long-range navigation in GNSS-denied environment with low-cost UAV platform

Abstract

The capability of an Unmanned Aerial Vehicle to navigate in GNSS-denied environments is desired in many situations, especially in situation in which the GNSS signal is not trustable. In this work is presented a solution based on a low-cost UAV platform, that uses a Landmark Recognition algorithm to compensate the inertial sensors errors cumulation, aiming to perform a long-range navigation, independently of GNSS information. The results in a real-time simulation demonstrated the effectiveness of the integration between the UAV control and the computer vision, performing a successful navigation of 20.7 km long at 400 meters high, where the system has been able to estimate the coordinates during the whole route, with an average error lower than 6 meters.