Artificial Intelligence Wiki

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
====== Creating an occupancy (ground) map with PR2 robot ======

In general, it should take you less than one hour to create a new map, usually about half an hour or less.

  * Step 0: position your robot to look in the direction of the X axis of your desired map frame. The resulting map frame will be very close to your odometry frame, so after rotating the robot turn the odometry off and on again (killing the ''robot.launch'').
  * Step 1: start the robot and the joystick.
  * Step 2: on your computer start Rviz.
  * Step 3: to create the map you need: (1) TF data of the odometry of your robot, (2) laser scanner data. Visualize them in Rviz.
  * Step 4: for the mapping you'll need the GMapping algorithms from the ''slam_gmapping'' package. Here's the launch file (tweaked from https://github.com/PR2/pr2_navigation/blob/hydro-devel/pr2_navigation_slam/slam_gmapping.xml):
<code>
<launch>
    <node pkg="gmapping" type="slam_gmapping" name="slam_gmapping" output="screen">
      <remap from="scan" to="base_scan"/>
      <param name="odom_frame" value="odom_combined"/>
      <param name="map_update_interval" value="30.0"/>
      <param name="maxUrange" value="16.0"/>
      <param name="sigma" value="0.05"/>
      <param name="kernelSize" value="1"/>
      <param name="lstep" value="0.05"/>
      <param name="astep" value="0.05"/>
      <param name="iterations" value="5"/>
      <param name="lsigma" value="0.075"/>
      <param name="ogain" value="3.0"/>
      <param name="lskip" value="0"/>
      <param name="srr" value="0.01"/>
      <param name="srt" value="0.02"/>
      <param name="str" value="0.01"/>
      <param name="stt" value="0.02"/>
      <param name="linearUpdate" value="0.5"/>
      <param name="angularUpdate" value="0.436"/>
      <param name="temporalUpdate" value="-1.0"/>
      <param name="resampleThreshold" value="0.5"/>
      <param name="particles" value="80"/>
      <param name="xmin" value="-50.0"/>
      <param name="ymin" value="-50.0"/>
      <param name="xmax" value="50.0"/>
      <param name="ymax" value="50.0"/>
      <param name="delta" value="0.0125"/>
      <param name="llsamplerange" value="0.01"/>
      <param name="llsamplestep" value="0.01"/>
      <param name="lasamplerange" value="0.005"/>
      <param name="lasamplestep" value="0.005"/>
      <param name="occ_thresh" value="0.65"/>
    </node>
</launch>
</code>
The laser topic should be ''scan'', so we remap from PR2's ''base_scan'', the odometry frame is specified as a parameter (''odom_combined'' for PR2). Another important parameter is ''delta'', that's the resolution of your map, we use 0.0125. ''occ_thresh'' is the occupied threshold from the yaml file of the generated map, we use 0.65 (it's the threshold from when a pixel is considered occupied).
  * Step 5: ''roslaunch'' the launch file, visualize the map in Rviz, drive the robot around carefully and slowly
  * Step 6: save the map: ''rosrun map_server map_saver''. This will create ''map.pgm'' and ''map.yaml'' in ''pwd'' of the terminal. The position of the origin of the resulting map in the image is specified in the ''map.yaml''. This is an example of the ''yaml'' file:
<code>
image: map.pgm
resolution: 0.012500
origin: [-4.9, -1.9, 0.000000]
negate: 0
occupied_thresh: 0.65
free_thresh: 0.196
</code>
So to find out where the origin of the map is in image (.pgm file), find the coordinates in pixels:
  4.9/0.0125 = 392  (so the origin of the map will be 392pixels from the left border of the image).
  1.9/0.0125 = 152  (the origin is 152pixels from the bottom border)
The X axis of the rotation of the origin is the X axis of the picture, from left to right (→). The Y axis is from down to up (↑). The counting starts from top right corner. If you want to change the rotation of the origin, rotate the picture. It is **recommended to upscale the image** to a much higher resolution (eg 2000x2000), then rotate and align, and then scale back to the original resolution, otherwise, your resulting map will have a bad resolution.
  * Step 7: test the map by localizing your robot. For that you'll use the Adaptive Monte Carlo Localization algorithm from the ''amcl'' package. Kill the ''gmapping'' and start publishing the map and the ''amcl'' node. Here's a launch file:
<code>
<launch>
  <include file="$(find pr2_machine)/$(env ROBOT).machine" />
  <node name="map_server" pkg="map_server" type="map_server" args="$(env PWD)/map.yaml" />
  <include file="$(find pr2_navigation_global)/amcl_node.xml" />
</launch>
</code>
  * Step 8: once you're happy with your new map, rename it to add a date suffix to the filename, e.g.: ''map.pgm'' -> ''map-2016-06-17.pgm''. When renaming make sure you also rename the ''yaml'' file and don't forget the map name **inside** the ''yaml'' file.
  * Step 9: upload your new map to the ''iai_maps'' repository, ''iai_maps/iai_maps/maps'' directory
  * Step 10: update the ''iai_maps/iai_maps/launch/map.launch'' file to point to your new awesome map.

**Important**: if you change the origin of the map from what it used to be previously, make sure you also adapt the transform between the origin of the furniture of the room and the map. That's usually specified in the static transform publisher of the map URDF launch file, e.g., ''iai_kitchen/launch/upload_kitchen.launch'' or ''upload_kitchen_obj.launch'' or similar. In addition to editing this launch file, also make sure that you tell everybody, who uses the environment models (pretty much everyone who has anything to do with the robot) that you have changed the origin.