# ORB-SLAM3 for Linux Redmine issue: https://redmine.drivecast.tech/issues/335 ## Installation Please refer to the original repository on the details of the installation procedure: https://github.com/UZ-SLAMLab/ORB_SLAM3 You can use the script to install the Thirdparty libraries and ORB-SLAM3 core and the examples. But make sure you have installed all the required packages (You may need to change the paths to the packages in your CMakeLists.txt): ``` cd ORB-SLA3_Linux chmod +x build.sh ./build.sh ``` ## Running ### Mono-Kitti Examples 1. Скомпилируйте файл папке **Examples/Stereo/stereo-kitti.cc** и запустите бинарник с параметрами: путь к словарю, путь к файлу с настройками, путь к папке с изображениями. Пример: ``` stereo-kitti /home/ivan/ivan/git/work_drivecast2/SLAM/ORB_SLAM3-1.0-release/Vocabulary/ORBvoc.txt /home/ivan/ivan/git/work_drivecast2/SLAM/ORB_SLAM3-1.0-release/Examples/Stereo/KITTI04-12.yaml /home/ivan/ivan/git/work_drivecast2/SLAM/datasets/mono-kitti/sequences/06 ``` 2. Бинарник создаст файлы `CameraTrajectory.txt` и `RealtimeCameraTrajectory.txt`, они практически идентичные. Первый файл выдаёт чуть более точную траекторию, но она недоступна в режиме "онлайн", когда позиция камеры в моменте. 3. Используйте данные файлы далее в репозитории [ue_demo](https://git.drivecast.tech/pi/ue_demo) ### Other Examples Besides all the examples listed in ORB-SLAM3 original repository you can also run uzh-fpv and kaist urban examples: For stereo examples change the paths to your dataset files in the scripts (Examples/Stereo/run_stereo_uzh_fpv.sh). Added scrips: - Examples/Stereo/run_stereo_uzh-fpv.sh - Examples/Stereo/run_stereo_complex_urban.sh - Examples/Stereo/run_stereo_uzh-fpv_multiple.sh - Examples/Stereo-Inertial/run_stereo_inertial_uzh-fpv.sh - Examples/Stereo-Inertial/run_stereo_inertial_zurich_urban_mav.sh ### Datasets - Download uzh-fpv dataset from here: https://fpv.ifi.uzh.ch/ - Download complex-urban dataset from here: https://sites.google.com/view/complex-urban-dataset/system NOTE: The download of the complex-urban dataset may take a while because you need to send an email to them to ask the permission. You can use the downloaded sequence from `/mnt/disk-small/podmivan/datasets/complex-urban` folder in the DriveCast server. After all the datasets are downloaded and the paths to them are changed in the scrips you can run them. From the ORB-SLAM3_Linux directory run the example: ``` cd Examples/Stereo/ ./run_stereo_uzh_fpv.sh ``` The other examples you can run accordingly to the details provided in the [ORB-SLAM3](https://github.com/UZ-SLAMLab/ORB_SLAM3) repository.