Ubuntu下安装LSD-SLAM

Posted on In , Views: Disqus:

安装

官方编译方法1

1
2
3
4
5
6
7
8
9
10
11
12
sudo apt-get install python-rosinstall
sudo apt-get install ros-indigo-libg2o ros-indigo-cv-bridge liblapack-dev libblas-dev freeglut3-dev libqglviewer-dev libsuitesparse-dev libx11-dev
mkdir ~/SLAM/Code/rosbuild_ws
cd ~/SLAM/Code/rosbuild_ws
roses init . /opt/ros/indigo
mkdir package_dir
roses set ~/SLAM/Code/rosbuild_ws/package_dir -t .
echo "source ~/SLAM/Code/rosbuild_ws/setup.bash" >> ~/.bashrc
bash
cd package_dir
git clone https://github.com/tum-vision/lsd_slam.git lsd_slam
rosmake lsd_slam

以上是官方的方法，但是我出现了编译错误：

1
2
3
4
5
opt/ros/indigo/include/sophus/sim3.hpp:339:5: error: passing ‘const RxSO3Type {aka const Sophus::RxSO3Group}’ as ‘this’ argument of ‘void Sophus::RxSO3GroupBase::setScale(const Scalar&) [with Derived = Sophus::RxSO3Group; Sophus::RxSO3GroupBase::Scalar = double]’ discards qualifiers [-fpermissive]
rxso3().setScale(scale);
^
make[3]: *** [CMakeFiles/lsdslam.dir/src/DepthEstimation/DepthMap.cpp.o] Error 1
make[3]: *** [CMakeFiles/lsdslam.dir/src/SlamSystem.cpp.o] Error 1

于是尝试使用Ros的catkin对LSD-SLAM进行编译

使用catkin对LSD-SLAM进行编译23

1
2
3
4
mkdir -p ~/catkin_ws/src
git clone https://github.com/tum-vision/lsd_slam.git
cd lsd_slam
git checkout catkin

change to the Catkin "branch" of the LSD-SLAM project, then:

1
sudo apt-get install ros-indigo-libg2o ros-indigo-cv-bridge liblapack-dev libblas-dev freeglut3-dev libqglviewer-dev libsuitesparse-dev libx11-dev

对lsd_slam/lsd_slam_viewer和lsd_slam/lsd_slam_core文件夹下的package.xml中添加：

1
2
<build_depend>cmake_modules</build_depend>
<run_depend>cmake_modules</run_depend>

对lsd_slam/lsd_slam_viewer和lsd_slam/lsd_slam_core文件夹下的CMakeFiles.txt中添加：

1
find_package(cmake_modules REQUIRED)

并且在所有的target_link_libraries中添加X11，如：

1
target_link_libraries(lsdslam ${FABMAP_LIB} ${G2O_LIBRARIES} ${catkin_LIBRARIES} sparse cxsparse X11)

然后开始编译：

1
2
cd ~/catkin_ws/
catkin_make

运行离线测试程序4

Download the Room Example Sequence and extract it.

打开一个终端:

1
roscoe

打开另外一个终端：

1
2
3
cd ~/catkin_ws/
source devel/setup.sh
rosrun lsd_slam_viewer viewer

打开另外一个终端：

1
2
3
cd ~/catkin_ws/
source devel/setup.sh
rosrun lsd_slam_core live_slam image:=/image_raw camera_info:=/camera_info

打开另外一个终端：

1
2
cd ~/catkin_ws/
rosbag play ~/LSD_room.bag

使用摄像头运行LSD_SLAM

安装驱动5：

1
2
3
4
5
6
7
8
cd ~/catkin_ws/
source devel/setup.sh
cd ~/catkin_ws/src
git clone https://github.com/ktossell/camera_umd.git
cd ..
catkin_make
roscd uvc_camera/launch/
roslaunch ./camera_node.launch

会显示一些摄像头的信息，如：

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
opening /dev/video1
pixfmt 0 = 'YUYV' desc = 'YUYV 4:2:2'
  discrete: 640x480:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 160x120:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 176x144:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 320x180:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 320x240:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 352x288:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 424x240:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 480x270:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 640x360:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 800x448:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 800x600:   1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 848x480:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 960x540:   1/15 1/10 2/15 1/5 
  discrete: 1024x576:   1/15 1/10 2/15 1/5 
  discrete: 1280x720:   1/10 2/15 1/5 
  discrete: 1600x896:   2/15 1/5 
  discrete: 1920x1080:   1/5 
  discrete: 2304x1296:   1/2 
  discrete: 2304x1536:   1/2 
pixfmt 1 = 'MJPG' desc = 'Motion-JPEG'
  discrete: 640x480:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 160x120:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 176x144:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 320x180:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 320x240:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 352x288:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 424x240:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 480x270:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 640x360:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 800x448:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 800x600:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 848x480:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 960x540:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 1024x576:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 1280x720:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 1600x896:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  discrete: 1920x1080:   1/30 1/24 1/20 1/15 1/10 2/15 1/5 
  int (Brightness, 0, id = 980900): 0 to 255 (1)
  int (Contrast, 0, id = 980901): 0 to 255 (1)
  int (Saturation, 0, id = 980902): 0 to 255 (1)
  bool (White Balance Temperature, Auto, 0, id = 98090c): 0 to 1 (1)
  int (Gain, 0, id = 980913): 0 to 255 (1)
  menu (Power Line Frequency, 0, id = 980918): 0 to 2 (1)
    0: Disabled
    1: 50 Hz
    2: 60 Hz
  int (White Balance Temperature, 16, id = 98091a): 2000 to 7500 (1)
  int (Sharpness, 0, id = 98091b): 0 to 255 (1)
  int (Backlight Compensation, 0, id = 98091c): 0 to 1 (1)
  menu (Exposure, Auto, 0, id = 9a0901): 0 to 3 (1)
  int (Exposure (Absolute), 16, id = 9a0902): 3 to 2047 (1)
  bool (Exposure, Auto Priority, 0, id = 9a0903): 0 to 1 (1)
  int (Pan (Absolute), 0, id = 9a0908): -36000 to 36000 (3600)
  int (Tilt (Absolute), 0, id = 9a0909): -36000 to 36000 (3600)
  int (Focus (absolute), 0, id = 9a090a): 0 to 255 (5)
  bool (Focus, Auto, 0, id = 9a090c): 0 to 1 (1)

配置camera_node.launch文件6，如：

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
<launch>
  <node pkg="uvc_camera" type="uvc_camera_node" name="uvc_camera" output="screen">
    <param name="width" type="int" value="640" />
    <param name="height" type="int" value="480" />
    <param name="fps" type="int" value="30" />
    <param name="frame" type="string" value="wide_stereo" />

    <param name="auto_focus" type="bool" value="False" />
    <param name="focus_absolute" type="int" value="0" />
    <!-- other supported params: auto_exposure, exposure_absolute, brightness, power_line_frequency -->

    <param name="device" type="string" value="/dev/video1" />
    <param name="camera_info_url" type="string" value="file://$(find uvc_camera)/example.yaml" />
  </node>
</launch>

注意官方程序默认分辨率为640*480。

打开一个窗口运行roscore； 打开另外一个窗口：

1
2
3
cd ~/catkin_ws/
source devel/setup.sh
rosrun lsd_slam_viewer viewer

再打开另外一个窗口：

1
2
3
cd ~/catkin_ws/
source devel/setup.sh
roslaunch uvc_camera camera_node.launch

再打开另外一个窗口：

1
rosrun lsd_slam_core live_slam /image:=image_raw _calib:=<calibration_file>

校正文件calibration_file可参考lsd_catkin_ws/src/lsd_slam/lsd_slam_core/calib中的cfg文件。

  1. https://github.com/tum-vision/lsd_slam↩︎

  2. http://visbot.blogspot.jp/2014/11/tutorial-building-of-lsd-slam-on-ros.html↩︎

  3. https://github.com/tum-vision/lsd_slam/issues/206↩︎

  4. http://visbot.blogspot.jp/2014/11/tutorial-building-of-lsd-slam-on-ros.html↩︎

  5. https://defendtheplanet.net/2014/11/05/using-ros-indigo-webcam-by-the-uvc_camera-usb-video-class-package↩︎

  6. http://www.cnblogs.com/xtl9/p/4694881.html↩︎