Skip to Main content Skip to Navigation

Capteur de stéréovision hybride pour la navigation des drones

Abstract : Knowledge of attitude, altitude, ground plane segmentation and motion is essential for an Unmanned Aerial Vehicle during critical maneuvers such as landing and take-off. In this thesis we present a hybrid stereoscopic rig composed of a fisheye and a perspective camera for vision-based navigation. This sensor is then exploited by systemic methods. In contrast to classical stereoscopic systems based on feature matching, we propose methods which avoid matching between hybrid views. A plane-sweeping approach is proposed for estimating altitude and detecting the ground plane. Rotation and translation are then estimated by decoupling : the fisheye camera contributes to evaluating attitude, while the perspective camera contributes to estimating the scale of the translation. The motion can be estimated robustly at the meter scale, thanks to the knowledge of the altitude. Our method uses a 2-point algorithm complemented by a Kalman filter. We propose a robust, real-time, accurate, exclusively vision-based approach with a C++ implementation. Although this approach removes the need for any non-visual sensors, it can also be coupled with an Inertial Measurement Unit.
Complete list of metadata

Cited literature [77 references]  Display  Hide  Download
Contributor : Damien Eynard <>
Submitted on : Thursday, December 15, 2011 - 11:07:58 PM
Last modification on : Thursday, September 9, 2021 - 4:52:02 PM
Long-term archiving on: : Sunday, December 4, 2016 - 4:26:29 PM


  • HAL Id : tel-00652615, version 1



Eynard Damien. Capteur de stéréovision hybride pour la navigation des drones. Robotique [cs.RO]. Université de Picardie Jules Verne, 2011. Français. ⟨tel-00652615⟩



Record views


Files downloads