Summary of Solution

This repository contains my implementation of a pothole detector using the provided simple world map. The solution makes use of OpenCV to evaluate the pothole colours and thus their contours, as well as using the Depth sensing camera to determine to the position of a given pothole relative to the world frame. The robot navigates around the map using a set of waypoints and detects the potholes positions and size. Once the run is complete, a summary report will be produced, which can be found in the Pothole-Reporting-Robot/ros2_ws/summary_reports to be evaluated.

Guideline to run the code

This assumes that ros2 humble is already installed onto the system this will run on. This has also only been configured and tested on a native enviroments of ROS2.

Additional python packages to include

pip install reportlab
pip install matplotlib

Configuration

From a desired directory you will need to clone the limo_ros2 repository, as in the native instructions in https://github.com/LCAS/CMP9767_LIMO/wiki/Simulator-Setup. In a folder on the same level as the limo_ros2 installation, clone this Pothole-Reporting-Robot repository

git clone https://github.com/Olseda20/Pothole-Reporting-Robot.git

Your file structure should look like this. This is important due to how some of the automation of the pothole detection is being run.

.
├── limo_ros2
│   ├── src
│   ├── install
│   ├── build
│   └── ... 
└── Pothole-Reporting-Robot
    ├── README.md
    ├── ros2_ws
    └── ... 

Once both directories are available change into the Pothole-Reporting-Robot directory.

cd Pothole-Reporting-Robot

Ideally if the rviz2 config file could be loaded into rviz by default, this would be useful in seeing the markers of the pothole in realtime. However this information will be produced in the summary report.

From here all that needs to be run is the the shell script. This will start up

1. Gazbo
2. Rviz2
3. Launch the my_robot_bringup launch file
4. Begin running follow_waypoint to begin recording the pothole position

If zsh is installed on your system, this can be started by making sure it is executable and simply running the script

chmod +x run_pothole_detector.sh
./run_pothole_detector.sh

Otherwise use the bash variant

chmod +x run_pothole_detector_bash.sh
./run_pothole_detector_bash.sh

Now enjoy as the potholes are being detected.
Once this is complete, navigate to the directory Pothole-Reporting-Robot/ros2_ws/summary_reports to see the output of the detection.

A demo run can be seen here: https://www.youtube.com/watch?v=fvODO8_kK-I

---

To manually deploy the parts

If you would like to run the different sections independently, navigate to the Pothole-Reporting-Robot/ros2_ws directory and run: For every new terminal, please made sure all of the source setups files are being run. Note: please use the correct setup for your interpreter

source /opt/ros/humble/setup.zsh
source /usr/share/colcon_argcomplete/hook/colcon-argcomplete.zsh
colcon build
source ../../limo_ros2/install/setup.zsh
source install/setup.zsh

To start Gazebo:

ros2 launch limo_gazebosim limo_gazebo_diff.launch.py world:='../worlds/potholes_simple.world'

In a new terminal to start RViz:

ros2 launch limo_navigation limo_navigation.launch.py map:=../maps/simple_map.yaml params_file:=/home/krono/dev/RobotProgramming/Pothole-Reporting-Robot/params/nav2_params.yaml use_sim_time:=true;

In a new terminal to start the node for pothole detection

ros2 run my_robot_controller simple_pothole_detector

In a new terminal to start the node to publishing the pothole positions as markers in the '/odom' frame

ros2 run my_robot_controller pothole_mapper

In a new terminal to begin the waypoint following and report generation

ros2 run my_robot_controller pothole_reporter