Type: Bachelor-/ Master Thesis Studien-/ Diplomarbeit
Die nachfolgende Aufgabenstellung ist in englischer Sprache verfasst.
Die Bearbeitung der Themen kann, je nach Wunsch des Studenten,
wahlweise in Englisch oder Deutsch erfolgen.
Loop Closure in Urban Environments based on Omnicam
- Source: Oxford Mobile Robotics lab
One of the main requirements for any robot navigating in an Urban environment is being able to robustly distinguish new places from places that have been already visited. The main aim of this project is for the robot to be able to distinguish places already visited using an Omni camera as it provides the benefit of 360 degree view of the surrounding environment. Loop closure can be addressed in many ways, such as image to image correspondance which involves matching salient features accross images however it suffers from degeneration in case of repeating visual structures. Some researchers have addressed this issue by weighting features based on occurance. Other possible techniques involve inclusion of certain geometric information from the environment.
The main steps involved in this Thesis are:
1) A Detailed Literature review of existing techniques used for loop closure.
2) Implementation of Bag of Words approach to cluster all possible features in the environment (Accumulation of Visual words).
3) Implementation of Image to Image comparison techniques (for Urban environments)
Prerequisites:
- Good programming skills C / C++
Helpful but not required:
- Acquainted with Linux
- Experience with ROS
Supervisors:
1) Sheraz Khan (contact: sheraz@lsr.ei.tum.de)
2) Roderick de Nijs (contact: rsdenijs@lsr.ei.tum.de)
Literature:
- Mark Cummins and Paul Newman FAB-MAP: Probabilistic Localization and Mapping in the Space of Appearance. The International Journal of Robotics Research, 27(6):647-665. 2008
- Kin Leong Ho and Paul Newman Combining Visual and Spatial Appearance for Loop Closure Detection in SLAM. In 2nd European Conference on Mobile Robots {ECMR}. Ancona, Italy. September 2005