This Repo contains everything related to my master thesis.
- Title: Autonomous Modeling of a 3D Environment with Drones
- First reviewer: Prof. (FH) PD Dr. Mario Döller
- Second reviewer: Sebastian Danninger, MA
Within the context of autonomous exploration and modeling of a 3D environment with drones, this work targets two distinct research questions. Most existing autonomous exploration frameworks divide the exploration task into three subproblems: localization, mapping, and path planning [1]. The location and mapping problem, which refers to the task of generating a local map of the environment while localizing the drone within it, is addressed by several already existing SLAM algorithms. The first examined research question to answer is what the best suited open-source visual SLAM algorithm for the exploration task is. Three algorithms were evaluated using predefined criteria addressing trajectory accuracy, point cloud accuracy and computation time. The results showed, that ORB SLAM outperforms other algorithms. The second research task was to develop a framework, that enables users to test and implement fully autonomous exploration systems within a virtual environment. The framework was developed in ROS and provides the possibility to navigate a drone within a simulation while the ORB SLAM algorithm is applied on the drone’s camera output. With the help of suited transformations on the output of the ORB SLAM algorithm, ground truth data is then streamed within the framework and enables users to apply and test flight path planning algorithms in order to complete the autonomous exploration task.
Contains everything related to the first chapter, the evaluation
Contains everything related to writing the actual thesis including latex setup
Contains everything related to the proposed framework in section two.