X-lanniDocker image used to reproduce all results for KITTI, EuRoC and TUM datasets of the following paper (for MIT Stata Center dataset, see this page):
- M. Labbé and F. Michaud, “RTAB-Map as an Open-Source Lidar and Visual SLAM Library for Large-Scale and Long-Term Online Operation,” in Journal of Field Robotics, accepted, 2018. (pdf) (Wiley)
Pull image:
$ docker pull introlab3it/rtabmap:jfr2018
or create the image with the Dockerfile provided in this directory:
$ cd rtabmap/docker/jfr2018
$ docker build -t introlab3it/rtabmap:jfr2018 .
Make sure to extract datasets in a subfolder called datasets relative to scripts in this folder (you can copy the scripts outside rtabmap source directory for convenience). The testing folder tree should look like this:
datasets/kitti/devkit
datasets/kitti/dataset/sequences/00
datasets/kitti/dataset/sequences/01
datasets/kitti/dataset/sequences/...
datasets/euroc/MH_01_easy
datasets/euroc/V1_01_easy
datasets/euroc/...
datasets/tum/rgbd_dataset_freiburg1_desk
datasets/tum/rgbd_dataset_freiburg3_long_office_household
datasets/tum/...
run_all.sh
run_kitti_datasets.sh
run_euroc_datasets.sh
run_tum_datasets.sh
For TUM dataset, use this script associate.py to synchronize RGB and depth images before processing. Usage in a TUM dataset: python associate.py rgb.txt depth.txt, this will create rgb_sync and depth_sync folders.
Process all datasets, results will be written to subfolders results/euroc, results/kitti and results/tum:
./run_all.sh
WARNING: processing all datasets with all different odometry approaches can require more than 10 hours to process. You can comment some configurations in run_all.sh if you are interested in just one odometry approach or one dataset. A single sequence can be tested like this too:
# Processing only sequence 07 of the kitti dataset with F2M odometry approach
./run_kitti_datasets.sh f2m 0 0 07
Show all results:
$ rtabmap-report --scale results