Installation Guide of Kinfu and CUDA

• 1. Preparation 

• OS : Ubuntu 14.04 (LTS)

• gcc version : 4.8

• g++ version : 4.8

• Nvidia GPU : Geforce GTX 970

• Recorder : Howard

• Because  kinfu library is not included in the normal pcl package, so you cannot  build the package from PPA or pcl-1.7.2 , you have to build from source  and the package is pcl-master that you could git from github . In this  tutorial, I will show you step by step to set up the packages. 

• 2. Before Start

1. Before  we start to build  pcl-trunk, I assume that you already build and  install pcl packages on your computer and could execute the program (ex.  pcl_write.cpp) successfully. 

2. First,  you have to install CUDA and your GPU driver. Please visit  nvidia  official website for more detail information. After finishing  install  CUDA tool-Kit,  you shall able to run the sample code (ex. nbody)  successfully. 

• For more information : 

• http://docs.nvidia.com/cuda/cuda-getting-started-guide-for-linux/#axzz3vapSvjIy

• 3. Main 

1. Download pcl-trunk

•  Go  to pcl github and git the lastest version of pcl-trunk. Take me as an  example, my pcl-trunk is called pcl-master, so I git pcl-master to  /home. You could use git clone command or download the zip file using  browser are all okay.

• Extract the file build a directory called "build"

•  cd pcl-master
   mkdir build
   cd build

• [NOTICE] Before start to cmake, you have to add your gcc and g++ to path. Otherwise, you will DEFINITELY build fail !!! 

whereis gcc & whereis g++
export CC=/usr/bin/gcc  //My gcc is located here, may not be the same as urs 

export CXX=/usr/bin/g++ //My g++ is located here, same above

• cmake and ccmake

• In your build directory, ccmake to enable gpu and cuda and add the DIR or LIB if ccmake cannot find the directory.

• ccmake ..

    

• sudo make
  sudo make install

• 4. Test and Verification

• After you make and install the package, all the execute file will be store in /bin

• Try to run the program to see whether it works successfully.

• Have fun.

• 5. Reference

1. Some tutorial blogs

1. Install Kinect Fusion on Ubuntu

• https://lvzhaoyang.wordpress.com/2013/06/18/install-kinect-fusion-on-ubuntu/

2. Configuring your PC to use your Nvidia GPU with PCL

• http://pointclouds.org/documentation/tutorials/gpu_install.php

3. PCL/OpenNI tutorial 1: Installing and testing

• http://robotica.unileon.es/mediawiki/index.php/PCL/OpenNI_tutorial_1:_Installing_and_testing

2. Errors and solutions blogs

1. Help! Error in building pcl-trunk #399

• PointCloudLibrary/pcl#399

2. PCL trunk compilation error

• http://www.pcl-users.org/PCL-trunk-compilation-error-td4034337.html

3. Building error with latest pcl-trunk: emmintrin.h (SSE2) is included for armv7l platform. #1271

• PointCloudLibrary/pcl#1271

3. Change gcc and g++ version

• a. How to switch GCC version using update-alternatives

• http://codeyarns.com/2015/02/26/how-to-switch-gcc-version-using-update-alternatives/

• b. How to change the default GCC compiler in Ubuntu?

• http://stackoverflow.com/questions/7832892/how-to-change-the-default-gcc-compiler-in-ubuntu

http://www.pcl-users.org/can-not-open-the-depth-source-solved-td4029696.html