Building SLAMBench on Windows - nikolaradulov/SLAMFuse GitHub Wiki

To install SLAMBench on Windows

  • Set up WSL

    1. Now WSL supports Linux GUI so update it, otherwise a lot of stuff will not work: wsl --update

    2. Get a fresh Ubuntu 20.04 installation: wsl --install -d Ubuntu-20.04

    3. Might want to set a personalised configuration to make sure it doesn't eat your pc : * Make sure wsl is dead: wsl --shutdown * Wait like 10s after the command for everything to shutdown * Create config file: notepad "$env:USERPROFILE.wslconfig" * On my machine I use (if you have less than 16 GB you might want to use less memory) : ` # Settings apply across all Linux distros running on WSL 2 [wsl2]

       	# Limits VM memory to use no more than 4 GB, this can be set as whole numbers using GB or MB
 	memory=6GB 

 	# Sets the VM to use two virtual processors
 	processors=8`

    4. Start your fresh installation: wsl -d Ubuntu-20.04

    5. Update it: _sudo apt-get update _

    6. Dependency installation: sudo apt-get -y install libvtk6.3 libvtk6-dev unzip libflann-dev wget mercurial git gcc g++ cmake python-numpy freeglut3 freeglut3-dev libglew-dev libglu1-mesa libglu1-mesa-dev libgl1-mesa-glx libgl1-mesa-dev libxmu-dev libxi-dev libboost-all-dev cvs libgoogle-glog-dev libatlas-base-dev gfortran gtk2.0 libgtk2.0-dev libyaml-dev build-essential libyaml-cpp-dev

    7. If you plan on using cuda algorithms run the following make sure that you have the nvidia drivers installed for windows. WSL should automatically be able to detect them and take them from windows. Check by running nvidia-smi. You should see your GPUs.

    8. Install clinfo: sudo apt-get install clinfo

  • Clone the repository.

  • Build SLAMBench. Use instructions from: https://github.com/pamela-project/slambench

    1. Install python
    2. Do: make deps (takes a long while, >2h)
    3. Build slambench: make slambench

  • To test the functionlity of the SLAMBench installation:

    • build lsdslam as an algorithm (or prefered)
    • install prefered dataset : eg make datasets/ICL_NUIM/living_room_traj2_loop.slam
    • run: eg ./build/bin/slambench -i datasets/ICL_NUIM/living_room_traj2_loop.slam -load build/lib/liblsdslam-cpp-library.so -o ICL_lsd --gui true