Multi-UAV real-time collaborative aerial mapping and 3D reconstruction
UAVs are increasingly used to create 2D and 3D maps to support monitoring and emergency scenarios. However, the spatial range that a single UAV can cover is limited due to battery life or regulations prohibiting a drone from flying above sensitive structures or leaving the pilot’s line of sight. Therefore, multiple drones are used to cover a larger area. However, reconstructing a 2D or 3D map from multiple sources on a centralised system is non-trivial. Challenges include loop closing, network connectivity and communication between UAVs.
The aim of this topic is to 1) explore efficient SLAM frameworks (monocular or visual SLAM), 2) design an efficient system collaborative mapping system that includes multiple (at least two) UAVs and a central server.
From the perspective of Open Science and reproducible science, we ask that the student places their code on our public GitHub. Furthermore, we encourage the student to explain their work through a short video that can be used for dissemination purposes.
[1] P. Schmuck and M. Chli, "Multi-UAV collaborative monocular SLAM," 2017 IEEE International Conference on Robotics and Automation (ICRA), Singapore, 2017, pp. 3863-3870, doi: 10.1109/ICRA.2017.7989445.