Markerless Indoor Localization of a Drone without GNSS
Abstract
The drone used at the AerodrOHM currently determines its position via an external, marker-based tracking system. While this provides very accurate poses, it requires a permanently installed, instrumented flight space and is not available outside of it. The goal of this thesis is to enable the existing drone to localize itself indoors without GNSS and without markers, using only its onboard sensors.
The work builds directly on the existing drone and the established ROS-based system architecture. In a first step, suitable onboard sensors (camera, IMU, optionally an optical-flow or depth sensor) are selected and integrated. Based on this, a markerless localization method is implemented — for example visual-inertial odometry (VIO) or a visual-SLAM approach — that continuously estimates the drone’s ego-motion and feeds it into the flight controller.
Finally, the achieved accuracy and robustness are systematically evaluated. The existing marker-based tracking system at the AerodrOHM serves as ground truth, against which the markerless solution is compared in flight experiments.
Description
The drone used at the AerodrOHM currently determines its position via an external, marker-based tracking system. This provides very accurate poses but requires a permanently installed, instrumented flight space. The goal of this thesis is a markerless, GNSS-free indoor localization based solely on onboard sensors, so that the drone can fly independently of the external tracking. The work builds on the existing drone and the established ROS system architecture.
Work Packages
- Familiarization with the existing drone platform and the marker-based tracking system at the AerodrOHM
- Selection and integration of suitable onboard sensors (camera, IMU, optionally optical-flow or depth sensor)
- Implementation of a markerless, GNSS-free localization method (e.g. visual-inertial odometry or visual SLAM)
- Integration into the existing ROS system architecture and flight control
- Evaluation of accuracy and robustness against the marker-based tracking as ground truth
- Flight experiments and documentation of the results at the AerodrOHM
Requirements
- Programming skills (Python and/or C++)
- Ideally some experience with ROS / ROS 2
- Basic knowledge of computer vision and/or state estimation (e.g. Kalman filter)
- Interest in drones, sensor integration and localization
- Independent and diligent way of working
This topic can be completed as a project or bachelor’s thesis subject to agreement.
Supervision
| Role | Name | |
|---|---|---|
| Supervisor | Prof. Dr. Christian Pfitzner | christian.pfitzner@th-nuernberg.de |
Prof. Dr. Christian Pfitzner
Professor of Mobile Robotics
Professor of Mobile Robotics at TH Nürnberg Georg Simon Ohm, Faculty efi. Director of the TTZ Nürnberger Land. Research interests in autonomous navigation, drone technology and sensor fusion.