ROS进阶学习笔记（11）- Turtlebot Navigation and SLAM

(写在前面： 这里参考rbx书中第八章和ROS社区教程进行学习，先看社区教程)

  ===  Doing the Turtlebot Navigation   ===

ref ros wiki: http://wiki.ros.org/turtlebot_navigation/Tutorials

1. Create the Data under remote control

Referring the RBX book(8.4.2 Collecting and Recording Scan Data),
then log into the TurtleBot's laptop and run:
  $ roslaunch rbx1_bringup turtlebot_minimal_create.launch
Replace the above command with the appropriate launch file for your own robot if you have one.
Next, log into the TurtleBot using another terminal window and run the command:
  $ roslaunch rbx1_bringup fake_laser.launch
(If you have a real laser scanner, you would run its launch file here instead of the fake
laser launch file.)
Next, launch the gmapping_demo.launch launch file. You can launch this file on your desktop workstation or the robot's laptop:
  $ roslaunch rbx1_nav gmapping_demo.launch
  or follow ros wiki: exbot@my_robot1:~$ roslaunch rbx1_nav gmapping_demo.launch

Then bring up RViz with the included gmapping configuration file On_Desktop:
  $ rosrun rviz rviz -d `rospack find rbx1_nav`/gmapping.rviz
  or follow ros wiki: exbot@my_desktop:~$ roslaunch turtlebot_rviz_launchers view_navigation.launch  # I used this, cuz the above causes problem.

Next, launch a teleop node On_Desktop for either the keyboard or joystick depending on your hardware:
  $ roslaunch rbx1_nav keyboard_teleop.launch

The final step is to start recording the data to a bag file. You can create the file
anywhere you like, but there is a folder called bag_files in the rbx1_nav package for
this purpose if you want to use it:
  $ roscd rbx1_nav/bag_files

Now start the recording process:
  $ rosbag record -O my_scan_data /scan /tf

where my_scan_data can be any filename you like. The only data we need to record is
the laser scan data and the tf transforms. (Thetf transform tree includes the
transformation from the /odom frame to the/base_link or/base_footprint frame
which gives us the needed odometry data.)

You are now ready to drive the robot around the area you'd like to map. Be sure to
move the robot slowly, especially when rotating. Stay relatively close to walls and
furniture so that there is always something within range of the scanner. Finally, plan to
drive a closed loop and continue past the starting point for several meters to ensure a
good overlap between the beginning and ending scan data.

Use the Steps above can let your turtlebot move under your control and create the map of the environment.

2. Create the map simutanlously.

There are at least two ways to create the map: Recording or rePlaying the bag data file. Let's see the first here and second next section.

When you are finished driving the robot, type Ctrl-C in the rosbag terminal window
to stop the recording process. Then save the current map as follows:
  $ roscd rbx1_nav/maps
  $ rosrun map_server map_saver -f my_map
where " my_map" can be any name you like. This will save the generated map into the
current directory under the name you specified on the command line. If you look at the
contents of the rbx1_nav/maps directory, you will see two new files: my_map.pgm
which is the map image and my_map.yaml that describes the dimensions of the map. It
is this latter file that you will point to in subsequent launch files when you want to usethe map for navigation.

To view the new map, you can use any image viewer program to bring up the .pgm file
created above. For example, to use the Ubuntu eog viewer ("eye of Gnome") run the

command:
  $ roscd rbx1_nav/maps
  $ eog my_map.pgm

You can zoom the map using your scroll wheel or the +/- buttons.

Here is a video demonstrating the gmapping process using Pi Robot and a Hokuyo laser scanner: http://youtu.be/7iIDdvCXIFM

3. Create the map by bag data file.

You can also create the map from the bag data you stored during the scanning phase
above. This is a useful technique since you can try out different gmapping parameters
on the same scan data without having to drive the robot around again.
当你想调试gmapping 参数的时候，不用每次都重新跑一遍机器人。

把所有的Terminal中运行的node,launch关掉；Next, turn on simulated time by setting the use_sim_time parameter to true:
  $ rosparam set use_sim_time true

Then clear the move_base parameters and re-launch the gmapping_demo.launch file
again:
  $ rosparam delete /move_base
  $ roslaunch rbx1_nav gmapping_demo.launch
You can monitor the process in RViz using the gmapping configuration file:
  $ rosrun rviz rviz -d `rospack find rbx1_nav`/gmapping.rviz
Finally, play back your recorded data:
  $ roscd rbx1_nav/bag_files
  $ rosbag play my_scan_data.bag
You will probably have to zoom and/or pan the display to keep the entire scan area in
view.
When the rosbag file has played all the way through, you save the generated map the
same way we did with the live data:
  $ roscd rbx1_nav/maps
  $ rosrun map_server map_saver -f my_map
where "my_map" can be any name you like.

This will save the generated map into the current directory under the name you specified on the command line. If you look at the
contents of the rbx1_nav/maps directory, you will see two files: my_map.pgm which is the map image and my_map.yamlthat describes the dimensions of the map. It is this latter file that you will point to in subsequent launch files when you want to use the map
for navigation.

To view the map created, you can use any image viewer program to bring up the .pgm
file created above. For example, to use the Ubuntu eog viewer ("eye of Gnome") run
the command:
  $ roscd rbx1_nav/maps
  $ eog my_map.pgm
You can zoom the map using your scroll wheel or the +/- buttons.
NOTE: Don't forget to reset the use_sim_time parameter after you are finished map
building. Use the command:
  $ rosparam set use_sim_time false

4. Conclusion

Now that your map is saved we will learn how to use it for localization in the next section.
For additional details about gmapping, take a look at the gmapping_demo.launch file
in the rbx1_nav/launch directory.  There you will see many parameters that can be
tweaked if needed.  This particular launch file is a copied from the
turtlebot_navigation package and the folks at OSRG have already dialed in the
settings that should work for you.  To learn more about each parameter, you can check
out the gmapping Wiki page.

视频: 结构化环境中机器人导航 - Navigation with Structured Env.SLA  https://youtu.be/EwNl1cfNjt4