Using SLAM for Map Building

Create custom maps with your car

By: Madison Doerr and Markus Schiffer

Introduction

The MuSHR car uses a map of its environment to localize itself. We can create custom 2D maps for our cars by using gmapping and our car’s laser scanner to survey the area to be mapped.

Goal

Create a custom map of an area by driving the car to survey the surroundings.

Requirements

  • Complete the quickstart tutorial
  • Complete the first steps tutorial
  • A desktop/laptop that can ssh into the car

Ssh into the car

Boot up the car and connect the controller.

Ssh into the car on your computer, using the following command while connected to the ROBOT_AP network.

$ ssh robot@10.42.0.1 -X

Or, if you set up Wifi connect in the first steps tutorial, use the following command while connected to the same wifi network as your robot. Replace the IP listed below with the IP of your robot.

$ ssh robot@172.16.77.37 -X

Update/Install gmapping

In a terminal sshed into the car using the above steps, make sure gmapping is installed.

If you installed ros-kinetic in the quickstart tutorial:

$ sudo apt install ros-kinetic-gmapping

If you installed ros-melodic in the quickstart tutorial:

$ sudo apt install ros-melodic-gmapping

Start teleop and gmapping

Once you have sshed into the car, start teleop to start the laser scanner and control the car.

$ roslaunch mushr_base teleop.launch

Verify that the controller is working and you can move the car. If not, try reconnecting the controller and/or restarting teleop.

In a separate terminal sshed into the robot, using the above process, set ROS_IP to the IP of your robot, then start slam. Slam will take the data gathered from the laser scanner started by teleop and transform that into a 2D map.

$ export ROS_IP=YOUR-IP
$ roslaunch mushr_base slam.launch

Start rviz to visualize the map building

In a terminal on your computer (not ssh-ed into the robot), set ROS_IP to your IP and ROS_MASTER_URI to the IP of the car, and then start rviz.

$ export ROS_IP=YOUR-IP
$ export ROS_MASTER_URI=http://ROBOT_IP:11311
$ rviz

A screen should appear of the robot and the map it is creating. If you don’t see the map, make sure the /car/map topic is added to the left sidebar (different from the /map/ topic!). You may also need to set the Fixed Frame value in the Displays > Global Options menu in rviz to /car/map/ instead of /map/. The map will dynamically update and refine as the laser scanner gathers new data.


Survey the surroundings

Now that we have started mapping, we need to survey the area! Use the controller to drive the car very slowly around the area you want to survey, pausing periodically to allow the laser scanner to gather data. You should be able to see the map in rviz updating as you drive.

Save the map

When you’re satisfied with the map in rviz, you can save it to your computer using the following steps.

In a local terminal, ensure ROS_IP is set to your IP and ROS_MASTER_URI is set to the IP of the car. Then, in that terminal, run map_saver using the following command, replacing map_name with the desired name of the map.

$ export ROS_IP=YOUR-IP
$ export ROS_MASTER_URI=http://ROBOT_IP:11311
$ rosrun map_server map_saver -f map_name map:=/car/map

This should create a .pgm file with the map and a .yaml file with the map metadata.

Using the map

On the sim

To load your new map into the simulator, move both files to ~/catkin_ws/src/mushr/mushr_base/mushr_base/mushr_base/maps/, replacing map_name with the name of your map.

$ mv map_name.pgm ~/catkin_ws/src/mushr/mushr_base/mushr_base/mushr_base/maps/
$ mv map_name.yaml ~/catkin_ws/src/mushr/mushr_base/mushr_base/mushr_base/maps/

Then, edit ~/catkin_ws/src/mushr/mushr_base/mushr_base/mushr_base/launch/includes/map_server.launch to set the map to the name of the .yaml file.

$ nano ~/catkin_ws/src/mushr/mushr_base/mushr_base/mushr_base/launch/includes/map_server.launch

It should look like the following, replacing map_name with the name of your map.

1
2
3
4
<launch>
    <arg name="map" default="$(find mushr_base)/maps/map_name.yaml" />
    <node pkg="map_server" name="map_server" type="map_server" args="$(arg map)" />
</launch>

Now, launch the sim to use your new map!

$ roslaunch mushr_sim teleop.launch

In a new terminal, run rviz.

$ rviz

Your new map and the car should appear in the sim! If you don’t see the map, make sure you are subscribed to the /map topic in the left sidebar. The map might not be in the center of the grid, and you may have to zoom out to see it.

On the car

To get the map from your computer to the robot, we need to use scp, replacing map_name with the name of your map and ROBOT_IP with the IP of your robot. From your computer, run the following from where you saved your maps:

$ scp map_name.pgm robot@ROBOT_IP:~/catkin_ws/src/mushr/mushr_base/mushr_base/mushr_base/maps/
$ scp map_name.yaml robot@ROBOT_IP:~/catkin_ws/src/mushr/mushr_base/mushr_base/mushr_base/maps/

Then, edit ~/catkin_ws/src/mushr/mushr_base/mushr_base/mushr_base/launch/includes/map_server.launch on the robot to set the map to the name of the .yaml file.

$ nano ~/catkin_ws/src/mushr/mushr_base/mushr_base/mushr_base/launch/includes/map_server.launch

It should look like the following, replacing map_name with the name of your map.

1
2
3
4
<launch>
    <arg name="map" default="$(find mushr_base)/maps/map_name.yaml" />
    <node pkg="map_server" name="map_server" type="map_server" args="$(arg map)" />
</launch>

Now, start teleop on the robot:

$ roslaunch mushr_base teleop.launch

And on your computer, set the IPs and start rviz to see the map!

$ export ROS_IP=YOUR-IP
$ export ROS_MASTER_URI=http://ROBOT_IP:11311
$ rviz

The map should appear, and you can move your robot into the map using Set 2D Pos Estimate.

Touching up the map

The map may have some stray pixels or jagged lines. To touch this up, we recommend using gimp to edit the .pgm file.

When doing this, use solid black (hex: #000000) pixels for edges, solid white (hex: #FEFEFE) areas for areas valid for the car, and solid gray areas (hex: #CDCDCD) for invalid space for the car.

Citation

If you plan to use any part of the the MuSHR platform (including tutorials, codebase, or hardware instructions) for a project or paper, please cite MuSHR: A Low-Cost, Open-Source Robotic Racecar for Education and Research.

@article{srinivasa2019mushr,
 title={{MuSHR}: A Low-Cost, Open-Source Robotic Racecar for Education and Research},
 author={Srinivasa, Siddhartha S. and Lancaster, Patrick and Michalove, Johan and Schmittle, Matt and Summers, Colin and Rockett, Matthew and Smith, Joshua R. and Chouhury, Sanjiban and Mavrogiannis, Christoforos and Sadeghi, Fereshteh},
 journal={CoRR},
 volume={abs/1908.08031},
 year={2019}
}