This application builds DynaSLAM algorithm in a docker container.
It is required to have nvidia-docker installed.
DynaSLAM paper: https://arxiv.org/pdf/1806.05620.pdf
DynaSLAM git repository: https://github.com/BertaBescos/DynaSLAM
Run the following command to build the docker. Building the docker can take a while, and at least 15GB from disk space.
mkdir Dataset
sudo docker build -t dyna .
The following command will open a docker terminal
./Run_docker.sh
To move files from/to the docker just move to the Dataset folder (for example, the SLAM sequence dataset)
To run DynaSLAM a important note is to execute the terminal command at /root/DynaSLAM. Executions outside this folder will probably return Segmentation fault message.
./Examples/RGB-D/rgbd_tum Vocabulary/ORBvoc.txt Examples/RGB-D/TUM3.yaml /home/Dataset/rgbd_dataset_freiburg3_walking_xyz/ Examples/RGB-D/associations/fr3_walking_xyz.txt src/python/mask_rcnn_coco.h5
Inside the docker just type the following command, but the docker will still be activate when you leave:
exit
If you want to return to the docker process:
./Attach_docker.sh
Outside the docker, type the following command to show all docker process running
sudo docker ps -a
To kill a docker:
sudo docker rm <CONTAINER ID>
Outside the docker, type the following command to show all docker images running
sudo docker images
To erase a docker image:
sudo docker rmi <IMAGE ID>