Autonomous navigation with quadcopters in unknown 3D environments requires accurate and robust solutions to localization and mapping. In the first part of my talk, I will present our current work on monocular visual SLAM on a Parrot Ardrone. In the second part, I will show a novel method for computing highly accurate, real-time visual odometry from RGB-D data that we plan to employ onboard on the AscTec Pelican platform. Finally, I will present our efforts on creating an RGB-D dataset with ground-truth camera poses obtained from a motion capture system that allows the accurate evaluation of such systems.
Since March 2011, Dr. Jürgen Sturm is a Post-Doc in the Computer Vision and Pattern Recognition group of Prof. Daniel Cremers at the Department of Computer Science of the Technical University of Munich. His major research interests lie in real-time 3D perception, localization, mapping and safe navigation for micro aerial vehicles. He obtained his PhD in 2011 from the Autonomous Intelligent Systems lab headed by Prof. Wolfram Burgard at the University of Freiburg. In his PhD thesis, he developed novel approaches to estimate kinematic models of articulated objects and mobile manipulators as well as for tactile sensing and imitation learning.
Universität Bonn, Institute for Computer Science, Departments: I, II, III, IV, V, VI