Converting an autonomous vehicle controller in Webots into ROS
This project is an implementation of Carnegie Mellon’s 24-677 (Modern Control Theory) from Webots simulation to ROS Noetic and Gazebo. An autonomous vehicle navigates a racetrack using waypoints from a .csv file. Not only was this a fun project to finish over winter break, but this also serves as a great introduction to ROS for interested students.
Since this was a class project, I have removed potential answers. If you would like the full code, please send me an email ([email protected]).
For this project, I used Oracle VMWare to simulate a Linux environment instead of dual booting my computer. You can find a tutorial about installing ROS using VMWare here: https://www.instructables.com/How-to-Install-ROS/
NB: I had an issue with using the sudo command when installing Linux in this way, and I solved the problem with this tutorial: https://www.baeldung.com/linux/username-not-in-sudoers-file
Here's a great ROS introduction from Udemy (make sure to get the 7 day free trial!): https://www.udemy.com/course/ros-for-beginners/
I used the Clearpath Jackal as my model for the Gazebo simulation; install directions can be found here: https://www.clearpathrobotics.com/assets/guides/noetic/jackal/simulation.html
NB: The starter world has too many obstacles, so you can use the empty_world.launch file instead: https://www.clearpathrobotics.com/assets/guides/kinetic/jackal/additional_sim_worlds.html
CMU_PID.py is the test file for a PID controller operating in Gazebo. For future use, any controller should work in this context, as long as the data collection from the simulation matches with the controller.
The .launch file called check_odometry runs the experiment; there is a odometry checking Python file also listed (check_odom.py) that I used to ensure I was calling the correct topics. The file CMU_test.py is a test file for taking in the correct data about the simulation and has no controller.