This repository contains the overwhelming majority of the work done during my master's thesis.
Important
To clone, it is recommended to install git-lfs on your system.
This project explores the collaboration between an aerial drone and a terrestrial robot for navigating unstructured environments. The drone performs initial environmental mapping using sensors (e.g. GNSS, cameras...), generating a topological map with identified paths, intersections, and potential targets. The terrestrial robot receives high-level instructions (e.g., “reach a target”) and navigate using this topological map. It utilizes onboard sensors to match environmental features previously detected with the drone and associated to the topological map.
Techniques are proposed to handle the perspective differences between aerial and ground views, including bird's-eye view generation and sparse correspondence matching. Simulated real-world scenario demonstrates that the proposed system offers a promising foundation for real-world deployment. This project explores the collaboration between an aerial drone and a terrestrial robot for navigating unstructured environments. The drone performs initial environmental mapping using sensors (e.g. GNSS, cameras...), generating a topological map with identified paths, intersections, and potential targets. The terrestrial robot receives high-level instructions (e.g., “reach a target”) and navigate using this topological map. It utilizes onboard sensors to match environmental features previously detected with the drone and associated to the topological map.
- Datasets: An empty folder where the datasets should be stored. The list and links to the datasets used are specified in the dedicated README.md.
- Docker: Dockerfiles to simplify reproducibility.
- Latex: The latex source files for the project plan, report and defence.
- Notes: The notes of my research and intermediate results. See dedicated README.md.
- Scripts: Scripts to test/automate things. If some script become more than a test, it will be moved to a dedicated repository (the list will be made available here).
See the docker/README.md for more information about the dependencies and scripts/README.md for the scripts.
This is the source files of my Master's thesis for my M.Sc. Eng in Autonomous Systems at the Danmarks Tekniske Universitet. It took place at the U2IS lab of ENSTA Paris.
It was supervised by Søren HANSEN and co-supervised by Alexandre CHAPOUTOT and Thibault TORALBA.
Given that this repository contains multiple type of document, two licenses are used:
- The files in scripts and docker (mostly python and dockerfile) are under GNU GPL license.
- The files in latex and notes (mostly images, markdown, latex files), except for images in logo, are under CC BY-SA 4.0 license. The templates are largely based of DTU's templates.
- The images in logo are copyrighted, belong to their rightful owner and should be used only when permitted by law.
- No datasets will be stored in datasets, please refer to the license given by the author of the dataset.
If you have any doubt regarding the licensing of part of this repository, please consider submitting an issue.