PGM Map Creator is added to the project as Git Submodule, but you can download it as an individual package:

git clone https://github.com/udacity/pgm_map_creator.git

If you prefer to control your robot to help it localize itself, you can install the following teleop package:

sudo apt install ros-melodic-teleop-twist-keyboard

1. We need libignition-math2-dev and protobuf-compiler to compile the map creator: sudo apt-get install libignition-math2-dev protobuf-compiler
2. Clone this repo inside the src folder of a catkin workspace with the PGM Map Creator submodule: git clone --recurse-submodules https://github.com/dudasdavid/Udacity-Robotics-RobotLocalization
3. Build workspace: catkin_make
4. Source environment: source devel/setup.bash
5. Start the Gazebo simulation and RViz: roslaunch my_robot world.launch
6. Start map server, localization and navigation: roslaunch my_robot amcl.launch
7. Load the RViz confuguration from: ~/catkin_ws/src/Udacity-Robotics-RobotLocalization/my_robot/rviz/localization.rviz
8. Optionally start the Teleop package: rosrun teleop_twist_keyboard teleop_twist_keyboard.py

This project is based on the my_robot package of Project 2: https://github.com/dudasdavid/Udacity-Robotics-ChaseTheBall.

The static map of the Gazebo environment was saved into the map folder using the PGM Map Creator package.

1. Successfully build PGM Map Creator.
2. Copy the MyWorld.world Gazebo world file into src/pgm_map_creator/world/ folder.
3. Add <plugin filename="libcollision_map_creator.so" name="collision_map_creator"/> before the </world> tag.
4. Run Gazebo server with: gzserver src/pgm_map_creator/world/MyWorld.world.
5. Save the map with PGM Map Creator: roslaunch pgm_map_creator request_publisher.launch.
6. Copy the saved map.pgm to maps folder of the my_robot package and create a map.yaml with default values (see my_robot package).

With the map.pgm and map.yaml we can start the map server, AMCL localization and the move_base navigation nodes. See my_robot/launch/amcl.launch.

After successfully start the simulation and localization you should see the following graph after executing rqt_graph: And the following TF tree after executing rqt_tf_tree:

1. At init the robot's pose is unknown on the map, we can see that even the yaw angle isn't correct:
2. After manually driving a few meters the robot pose is succesfully identified by the AMCL localization:
3. After the successful localization we can set up a navigation goal in RViz:
4. And the robot will follow the trajectory calculated by the move_base navigation stack:
5. And finally robot can successfully reach the desired navigation goal:

tree

.Udacity-Robotics-RobotLocalization             # Where am I? Project
├── README.md                                   # Project documentation
├── documentation
│   ├── 01_init.png
│   ├── 02_localization.png
│   ├── 03_navigation.png
│   ├── 04_navigation.png
│   ├── 05_navigation.png
│   ├── frames.png
│   ├── gazebo_sim.png
│   ├── map.png
│   ├── rosgraph.png
│   └── video.png
├── my_robot                                    # my_robot node
│   ├── CMakeLists.txt                          # Link libraries
│   ├── config                                  # move_base config files
│   │   ├── base_local_planner_params.yaml
│   │   ├── costmap_common_params.yaml
│   │   ├── global_costmap_params.yaml
│   │   └── local_costmap_params.yaml
│   ├── launch
│   │   ├── amcl.launch                         # Starts the map_server, amcl and move_base
│   │   ├── robot_description.launch
│   │   └── world.launch                        # Initialize robot in Gazebo environment
│   ├── maps                                    # Saved PGM map of Gazebo environment
│   │   ├── map.pgm
│   │   └── map.yaml
│   ├── meshes                                  # Custom robot meshes
│   │   ├── chassis.dae
│   │   ├── chassis.SLDPRT
│   │   ├── chassis.STEP
│   │   ├── hokuyo.dae
│   │   ├── wheel.dae
│   │   ├── wheel.SLDPRT
│   │   └── wheel.STEP
│   ├── package.xml
│   ├── rviz
│   │   ├── localization.rviz                   # RViz config file for this project
│   │   └── my_robot.rviz
│   ├── urdf                                    # Robot URDF description
│   │   ├── my_robot.gazebo
│   │   └── my_robot.xacro
│   └── worlds                                  # Simulated world in Gazebo
│       ├── empty.world
│       └── MyWorld.world
└──  pgm_map_creator                            # PGM Map Creator Git Submodule
    ├── CMakeLists.txt
    ├── launch
    │   └── request_publisher.launch
    ├── LICENSE
    ├── maps                                    # Saved PGM maps
    │   ├── example_map.pgm
    │   └── map.pgm
    ├── msgs
    │   ├── CMakeLists.txt
    │   └── collision_map_request.proto
    ├── package.xml
    ├── README.md
    ├── src
    │   ├── collision_map_creator.cc
    │   └── request_publisher.cc
    └── world                                   # Gazebo worlds
        ├── MyWorld.world
        └── udacity_mtv

• Project was built on Ubuntu 18.04 with ROS Melodic
• PGM Map Creator had several build issues with ROS Melodic, these are fixed in the Git Submodule
• Lidar range is set to 360 degree