一、前言
RTAB-Map (Real-Time Appearance-Based Mapping)是一种基于全局贝叶斯闭环检测的RGB-D Graph SLAM方法。它可以用kinect的深度信息结合kinect变
换得到的激光数据进行即时定位与建图（gmapping算法志只用到了kinect转换得到的激光数据，而把深度信息丢弃了）。
官网地址：http://introlab.github.io/rtabmap/
ROS下面安装请看：https://github.com/introlab/rtabmap_ros#rtabmap_ros
教程请看：https://github.com/introlab/rtabmap/wiki/Tutorials
在机器人上使用RTAB-Map：Setup RTAB-Map on Your Robot!
二、deb方式安装
嫌麻烦采用deb方式安装
Jade: $ sudo apt-get install ros-jade-rtabmap-ros Indigo: $ sudo apt-get install ros-indigo-rtabmap-ros Hydro: $ sudo apt-get install ros-hydro-rtabmap-ros
三、源码安装
请看github上的安装说明：https://github.com/introlab/rtabmap_ros#rtabmap_ros
简单的安装方法是：
indigo:
$ sudo apt-get install libsqlite3-dev libpcl-1.7-all libfreenect-dev libopencv-dev
hydro:
$ sudo apt-get install libsqlite3-dev libpcl-1.7-all ros-hydro-libfreenect ros-hydro-opencv2
然后下载RTAB-Map源码编译安装：
$ git clone https://github.com/introlab/rtabmap.git rtabmap $ cd rtabmap/build $ make -j4 $ make install
安装完之后插上Kinect在终端运行命令：
$ rtabmap
出现GUI，然后选择“File”——“New database”，再点击开始按键就可以出现图像了
在catkin工作空间中安装RTAB-Map ros-pkg
$ cd ~/catkin_ws $ git clone https://github.com/introlab/rtabmap_ros.git src/rtabmap_ros $ catkin_make
1、安装可选的依赖项
（1）如果你需要用到SURF/SIFT那就装一下OpenCV吧
请看：http://blog.csdn.net/u013453604/article/details/49781771
由于cv-bridge依赖的是opencv2，所以装opencv3的话rtabmap_ros里面有些库会冲突，这里建议装opencv 2.4.11。
（2）安装g2o，这是在闭环检测中实现图优化的包
先安装依赖项
$ sudo apt-get install cmake libeigen3-dev libsuitesparse-dev
下载源码编译安装
$ cd $ git clone https://github.com/RainerKuemmerle/g2o.git $ cd g2o/ $ mkdir build $ cd build/ $ cmake ../ $ make
安装完之后库文件在/usr/local/lib，头文件在/usr/local/include/g2o
（3）安装其他依赖项Qt, PCL, dc1394, OpenNI, OpenNI2, Freenect, g2o, Costmap2d, Rviz, Octomap, CvBridge
$ sudo apt-get install libqt4-dev libpcl-1.7-all-dev libdc1394-22-dev ros-indigo-openni-launch ros-indigo-openni2-launch ros-indigo-freenect-launch ros-indigo-costmap-2d ros-indigo-octomap-ros ros-indigo-rviz ros-indigo-cv-bridge
如果按照github上的命令安装依赖项不成功就运行上述命令吧，其中opencv和g2o我是源码安装的，所以相对于github上的命令我这里把ros-indigo-g2o去掉了，把libdc1394-dev改成了libdc1394-22-dev
（4）安装GTSAM
GTSAM is a library of C++ classes that implement smoothing and mapping (SAM) in robotics and vision, using factor graphs and Bayes networks as the underlying computing paradigm rather than sparse matrices.把gtsam包下载到某处进行编译安装:
$ cd software/ $ git clone https://bitbucket.org/gtborg/gtsam.git $ cd gtsam $ mkdir build $ cd build $ cmake .. $ make check $ make install
（5）安装cvsba
cvsba is an OpenCV wrapper for the well-known Sparse Bundle Adjustment library (sba) by M. Lourakis.
安装依赖项
$ sudo apt-get install liblapack-dev libf2c2-dev
下载cvsba：http://sourceforge.net/projects/cvsba/files/latest/download?source=typ_redirect
解压得到cvsba-1.0.0文件夹，我把它放到～/software/文件夹下
$ cd ~/software/cvsba-1.0.0 $ mkdir build $ cd build $ cmake .. $ make $ sudo make install
安装完之后还要执行以下命令才可以让rtabmap_ros找到它
$ sudo mkdir /usr/local/lib/cmake/cvsba $ sudo mv /usr/local/lib/cmake/Findcvsba.cmake /usr/local/lib/cmake/cvsba/cvsbaConfig.cmake
（6）安装Freenect2
请看这里：https://github.com/OpenKinect/libfreenect2#debianubuntu-1404-perhaps-earlier
Driver for Kinect for Windows v2 (K4W2) devices (release and developer preview). Note: libfreenect2 does not do anything for either Kinect for Windows v1 or Kinect for Xbox 360 sensors. Use libfreenect1 for those sensors. This driver supports: RGB image transfer IR and depth image transfer registration of RGB and depth images
先下载Freenect2源代码
$ cd ~/software $ git clone https://github.com/OpenKinect/libfreenect2.git
安装依赖项
$ sudo apt-get install build-essential cmake pkg-config libturbojpeg libjpeg-turbo8-dev mesa-common-dev freeglut3-dev libxrandr-dev libxi-dev
安装libusb
$ sudo apt-get install libusb-1.0-0-dev
注意libusb版本要大于1.0.20，否则编译freenect2的时候会报错：
CMake Error: The following variables are used in this project, but they are set to NOTFOUND. Please set them or make sure they are set and tested correctly in the CMake files: LibUSB_LIBRARY linked by target "freenect2" in directory /home/siat/libfreenect2
如果libusb版本太低可以先卸载libusb-1.0-0-dev更新软件源重装libusb-1.0-0-dev，运行如下命令
$ sudo apt-get remove libusb-1.0-0-dev $ sudo apt-add-repository ppa:floe/libusb $ sudo apt-get update $ sudo apt-get upgrade $ sudo apt-get install libusb-1.0-0-dev
安装 GLFW3
$ sudo apt-get install libglfw3-dev
如果上面命令无效请使用下面命令：
$ cd libfreenect2/depends $ sh install_ubuntu.sh $ sudo dpkg -i libglfw3*_3.0.4-1_*.deb
安装OPENCL依赖
对于AMD GPU：$ apt-get install opencl-headers
对于Nvidia GPU：（前提是你已经安装Nvidia驱动）$ apt-get install opencl-headers
对于Intel GPU：
$ apt-get install beignet-dev
不行的话运行：$ sudo apt-add-repository ppa:pmjdebruijn/beignet-testing
编译Freenect2
$ mkdir build && cd build $ cmake .. $ make $ sudo make install # without sudo if cmake -DCMAKE_INSTALL_PREFIX=$HOME/...
运行测试
$ ./bin/Protonect
2、下载安装RTAB-Map
$ cd ~ $ git clone https://github.com/introlab/rtabmap.git rtabmap $ cd rtabmap/build $ cmake -DCMAKE_INSTALL_PREFIX=~/catkin_ws/devel .. $ make -j4 $ make install
3、在catkin工作空间中安装RTAB-Map ros-pkg
$ cd ~/catkin_ws $ git clone https://github.com/introlab/rtabmap_ros.git src/rtabmap_ros $ catkin_make
4、更新版本
$ cd rtabmap $ git pull origin master $ cd build $ make $ make install $ roscd rtabmap_ros $ git pull origin master $ cd ~/catkin_ws $ catkin_make
四、遇到的报错
编译时遇到如下报错，最后发现是关于openni的库有问题，因为我用源码装过openni驱动，有可能是版本冲突了，其实根本没必要从源码安装， 可以从其他电脑上把usr/include下面的ni文件夹拷过来，我是重装了Ubuntu和ros indigo，然后按照前面简单的安装方法安装的，过程中自动安装了openni
[email protected]:~/rtabmap/build$ make -j4
[ 7%] Built target rtabmap_utilite
[ 7%] Built target uresourcegenerator
[ 8%] Built target imagesJoiner
[ 9%] Built target extractObject
[ 10%] Building CXX object corelib/src/CMakeFiles/rtabmap_core.dir/CameraThread.cpp.o
[ 10%] [ 11%] [ 12%] Building CXX object corelib/src/CMakeFiles/rtabmap_core.dir/CameraRGBD.cpp.o
Building CXX object corelib/src/CMakeFiles/rtabmap_core.dir/VisualWord.cpp.o
Building CXX object corelib/src/CMakeFiles/rtabmap_core.dir/EpipolarGeometry.cpp.o
In file included from /usr/include/ni/XnOS.h:27:0,
from /usr/include/pcl-1.7/pcl/io/openni_camera/openni.h:47,
from /usr/include/pcl-1.7/pcl/io/openni_camera/openni_depth_image.h:44,
from /home/siat/rtabmap/corelib/src/../include/rtabmap/core/CameraRGBD.h:37,
from /home/siat/rtabmap/corelib/src/CameraRGBD.cpp:28:
/usr/include/ni/XnPlatform.h:73:3: error: #error OpenNI Platform Abstraction Layer - Unsupported Platform!
#error OpenNI Platform Abstraction Layer - Unsupported Platform!
^
.
.
.
In file included from /usr/include/pcl-1.7/pcl/io/openni_camera/openni.h:50:0,
from /usr/include/pcl-1.7/pcl/io/openni_camera/openni_depth_image.h:44,
from /home/siat/rtabmap/corelib/src/../include/rtabmap/core/CameraRGBD.h:37,
from /home/siat/rtabmap/corelib/src/CameraThread.cpp:31:
/usr/include/ni/XnCppWrapper.h:10045:76: error: macro "XN_VALIDATE_NEW" passed 4 arguments, but takes just 2
XN_VALIDATE_NEW(pTrans, StateChangedCallbackTranslator, handler, pCookie);
^
/usr/include/ni/XnCppWrapper.h:10104:75: error: macro "XN_VALIDATE_NEW" passed 4 arguments, but takes just 2
XN_VALIDATE_NEW(pTrans, StateChangedCallbackTranslator, handler, pCookie);
^
[ 13%] Building CXX object corelib/src/CMakeFiles/rtabmap_core.dir/VWDictionary.cpp.o
In file included from /usr/include/ni/XnStatusCodes.h:27:0,
from /usr/include/ni/XnMacros.h:27,
from /usr/include/ni/XnOS.h:28,
from /usr/include/pcl-1.7/pcl/io/openni_camera/openni.h:47,
from /usr/include/pcl-1.7/pcl/io/openni_camera/openni_depth_image.h:44,
from /home/siat/rtabmap/corelib/src/../include/rtabmap/core/CameraRGBD.h:37,
from /home/siat/rtabmap/corelib/src/CameraRGBD.cpp:28:
/usr/include/ni/XnStatus.h:33:9: error: ‘XnUInt32’ does not name a type
typedef XnUInt32 XnStatus;
^