Output details
11 - Computer Science and Informatics
Imperial College London
Inverse Depth Parametrization for Monocular SLAM
<23>This highly cited work, now a standard reference, is the culmination of several years of collaboration with the University of Zaragoza. It generalises "Unified inverse depth parametrization for monocular SLAM", the most highly cited paper ever published at the elite Robotics: Science and Systems Conference. We solved a significant problem in initialising features in monocular SLAM via an elegant new parameterisation which efficiently represents depth uncertainty. The technique is widely used, including in a major forthcoming consumer home robotics product (currently confidential). Our open source implementation (http://www.robots.ox.ac.uk/~SSS06/Website/index.html) has been downloaded thousands of times and used in international summer schools.