ORB_SLAM_2_DOCKER

This is a repositroy that contains dockerfile and instructions for installing the working fork of ORB SLAM 2, specifically for the purpose of using it with Tonic.

Installing pre-requisite docker

In order to run this docker you need another one. I did not merge those together because the first one contains OpenCV, and can take A LOT of time to build. So more sensible way is to first build the first one, and build second one.

wget https://github.com/Valian/docker-python-opencv-ffmpeg/archive/master.zip -O docker-python-opencv-ffmpeg.zip
unzip -q docker-python-opencv-ffmpeg.zip
rm docker-python-opencv-ffmpeg.zip
cd docker-python-opencv-ffmpeg-master

Then build the image:

sudo docker image build -t cv3_cuda_contrib_py2 -f Dockerfile-py2-cuda .

So building this will take about 2-3 hours, go grab coffee, solve world hunger or smth. Go watch Geohot building a slam.

And then delete unused files:

cd ..
rm -rf docker-python-opencv-ffmpeg-master

Installing some dependencies, for dev purposes

For easier development, some dependencies are staying outside the docker, run these lines.

git clone https://github.com/mmajewsk/ORB_SLAM2

wget https://github.com/Algomorph/pyboostcvconverter/archive/master.zip -O pyboostcvconverter.zip
unzip -q pyboostcvconverter.zip
rm pyboostcvconverter.zip
mkdir ORB_SLAM2/include/pyboostcvconverter
mv pyboostcvconverter-master/include/pyboostcvconverter/pyboostcvconverter.hpp ORB_SLAM2/include/pyboostcvconverter/pyboostcvconverter.hpp
mv pyboostcvconverter-master/src/pyboost_cv3_converter.cpp ORB_SLAM2/src/pyboost_cv3_converter.cc
rm -rf pyboostcvconverter-master

Building image

Everything is contained in the dockerfile, so just build it.

docker image build -t orb_slam_py2 -f Dockerfile-orb-slam-py2 .

This should not take long.

If you need to limit the resources for your machine you can do it like this:

docker image build -t orb_slam_py2 -f Dockerfile-orb-slam-py2 . --cpu-shares=100 --memory=1024m

Running

In order to be able to see gui, you must start the x session (for example mobaxterm), and then you can just run:

docker run --rm -p 2207:2207 -e DISPLAY=192.168.1.195:0.0 -it orb_slam_py2 

You can also run bash.

docker run --rm -p 2207:2207 -e DISPLAY=192.168.1.195:0.0 -it orb_slam_py2 bash

Dependencies and other useful links

https://github.com/Valian/docker-python-opencv-ffmpeg

https://github.com/Algomorph/pyboostcvconverter

https://github.com/mmajewsk/ORB_SLAM2