Karl-Popper-Kolleg on Networked Autonomous Aerial Vehicles (KPK NAV) is a research group focused on real-world applications, in this case to realize the 3D reconstruction of a real environment using collaborative autonomous vehicles. Among four research areas, which are navigation and pose estimation, coordination and mission planning, wireless communications and networking, and distributed autonomous network synchronization, two master thesis topics are derived.
The student will get a chance to work with a real drone, learn how to program it, control it, and work with the sensor data. The environment will be placed indoor, together with the motion tracking system able to give locations of the corresponding entities in the system in submillimeter accuracy.
There are several challenges a student will have to handle in order to successfully understand and complete the project. One of them is to explore the unknown environment in order to find objects of interest. The student will have to be able to read object’s specifications based on the sensor data, by applying computer vision algorithms.
In addition, the student will have to use the images of detected objects, together with a 3D reconstruction software, in order to create a 3D map of the environment.
- Setup the Intel Aero drone for indoor flights
- Perform literature research on object detection algorithms and 3D reconstruction tools
- Deploy an object detection algorithm based on camera data
- Create a 3D map of the environment using the 3D modelling tool
- Asses the implemented algorithms and results
M1: Drone setup and literature study
M2: Algorithm implementation
M3: Experimental evaluation
- Good knowledge in Python/C++, Linux
- A plus is the knowledge of Robot Operating System (ROS), GIT
- English (thesis has to be written in English)
- Interest for new software and hardware aspects of drone research
Period and Contacts
Time period: 6 months, beginning SS19
- Bernhard Rinner (firstname.lastname@example.org)
- Petra Mazdin (email@example.com)
External supervisor(s): n/a