No description, website, or topics provided.
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
config
launch
CMakeLists.txt
README.md
cartographer.rviz
husky_cartographer.png
husky_cartographer_install.sh
package.xml

README.md

Husky Cartographer Demo

This tutorial shows you how to use move_base with Google Cartographer to perform autonomous planning and movement with simultaneous localization and mapping (SLAM), on a simulated Husky, or a factory-standard Husky with a laser scanner publishing on the /scan topic.

Developed by Google, Cartographer is an open-source library used for real-time simultaneous localization and mapping (SLAM). Using LIDAR data, it generates submaps which are later optimized and scan-matched to provide real-time loop closure. As a result, Cartographer is able to successfully close loops and generate a consistent map each time whereas Gmapping can occasionally fail to do so, resulting in unique maps generated after every iteration [1]. A variety of sensor configurations are supported allowing Cartographer to be used for a broad range of applications. In this demo, Cartographer uses IMU and LIDAR data from the Husky to perform real-time SLAM.

To adapt this demo to your own Husky, you may need to clone the husky_cartographer_navigation repository, and modify the relevant parameters. To learn about move_base and the navigation stack, see the Navigation Tutorials. To learn more about Google Cartographer for ROS, see the Cartographer ROS documentation.

Instructions

  1. To get started with 2-D SLAM using Google Cartographer, clone this repository into your working directory:

    git clone http://github.com/husky/husky_cartographer_navigation.git

  2. Run the following script to create a workspace and install proto3. This script will also install the packages required to use Cartographer as well as the husky, husky_desktop, and husky_simulator packages:

    source $(pwd)/husky_cartographer_navigation/husky_cartographer_install.sh

  3. Open two new terminal/tabs, source the workspace for each terminal/tab:

    source install_isolated/setup.bash

    1. Launch the Gazebo simulation:

      roslaunch husky_gazebo husky_playpen.launch

    2. Launch the Cartographer node to begin SLAM (NOTE: This also launches RViz to visualize the robot):

      roslaunch husky_cartographer_navigation cartographer_demo.launch

  4. In the Rviz visualizer, make sure the visualizers in the Navigation group are enabled.

  5. Use the 2D Nav Goal tool in the top toolbar to select a movement goal in the visualizer. Make sure to select an unoccupied (dark grey) or unexplored (light grey) location.

  6. As the robot moves, you should see the grey static map (map topic) grow. There might be discrete jumps in the map as the Cartographer algorithm attempts to localize the robot.

Husky World Map

  1. To save the generated map, you can run the map_saver utility:

    rosrun map_server map_saver -f <filename>

Tuning Cartographer

To tune Cartographer for low latency SLAM, edit the husky.lua configuration file found in the husky_cartographer_navigation/config directory.

For more information on tuning, click here

[1] T. Coroiu and O. Hinton, "A Platform for Indoor Localisation, Mapping, and Data Collection using an Autonomous Vehicle," M.S. thesis, Lund Univ., Lund, Sweden, 2017.