This is my Undergraduate SLAM system which is actually based on ORB_SLAM2 but without loop-closing algorithm. For evaluation, I compared the results of my project with ORB_SLAM and PTAM.
I tried my project on TUM dataset.
If you want to run my code, please drag the 'Thirdparty' and 'Vocabulary' profile provided by ORB_SLAM2 to the project dir.
-
Download a sequence from http://vision.in.tum.de/data/datasets/rgbd-dataset/download and uncompress it.
-
Associate RGB images and depth images using the python script associate.py. We already provide associations for some of the sequences in Examples/RGB-D/associations/. You can generate your own associations file executing:
python associate.py rgbd_datatest/rgb.txt rgbd_datatest/depth.txt > associations.txt
python associate.py fr1_desk/rgb.txt fr1_desk/depth.txt > associations.txt
python associate.py rgbd_desk/rgb.txt rgbd_desk/depth.txt > associations.txt
python associate.py rgbd_dataset_slam/rgb.txt rgbd_dataset_slam/depth.txt > associations.txt
python associate.py fr1_xyz/rgb.txt fr1_xyz/depth.txt > associations.txt
- Execute the following command. Change
TUMX.yamlto TUM1.yaml,TUM2.yaml or TUM3.yaml for freiburg1, freiburg2 and freiburg3 sequences respectively. ChangePATH_TO_SEQUENCE_FOLDERto the uncompressed sequence folder. ChangeASSOCIATIONS_FILEto the path to the coresponding associations file.
./main_tum setting/TUM1.yaml Vocabulary/ORBvoc.txt rgbd_datatest associations.txt
./main_tum setting/TUM1.yaml Vocabulary/ORBvoc.txt fr1_desk associations.txt
./main_tum setting/TUM1.yaml Vocabulary/ORBvoc.txt rgbd_desk associations.txt
./main_tum setting/TUM1.yaml Vocabulary/ORBvoc.txt rgbd_dataset_slam associations.txt
./main_tum setting/TUM1.yaml Vocabulary/ORBvoc.txt fr1_xyz associations.txt
#######
If you want to run the code in qtcreator， just use the cmd starting from setting/... without main
fr1/desk2 for desk_test2
fr2/desk
Drawing trajactory
python draw_groundtruth.py