MPC_PID_ROS2_Control

Requirements

• Ubuntu 20.04
• ROS2 Foxy (or newer ROS2 is also proficient) with Turtlebot3 package installed
• Ceres Solver

Note : You can also clone and build the Turtlebot3 from source.

Install

• mkdir -p ~/ros2_ws/src
• cd ~/ros2_ws/src
• git clone https://github.com/lenhatquang2512/MPC_PID_ROS2_Control.git
• cd ..
• colcon build --symlink-install
• source install/setup.bash

Usage

• Run turtlebot3 gazebo emptyworld simulator:

  ros2 launch turtlebot3_gazebo empty_world.launch.py

• Add some obstacles by inserting some pillars

• The goal is set (by default) at x = 5.0, y = -3.0, you can easily change the target goal later in src folder and then build the package again.

• Open another Terminal, then give this command:

  source install/setup.bash

  ros2 run hello mpc_obstacle_avoid

• Enjoy watching the robot navigates and reached the goal and avoid all obstacles !

Warning

• Since this is Naive MPC planner and control, setting too much obstacles are not recommended. It is possible, though, by tuning all the parameters. This is trial and error method, but the purpose of this package is to implement PID and MPC from scratch without using Nav2 plugin controller/planner.
• Ceres solver is a good tool to solve optimization problem, personally I think it is better than IpOpt.

UPDATE (September 2023)

• Currently the master branch can be built and run for any version of ROS2 including Dashing and Eloquent for Ubuntu 18.04 (running succesfully on Jetson Nano confirmed), and also worked with Foxy installed in VMWare. My package can also be used for almost any kind of mobile robots as long as they have 3 basic topics : /odom , /cmd_vel and /scan. So the requirements for Ubuntu 20.04 with Foxy is not that strictly necessary anymore. I will make more test cases to fully confirm and update the README soon.

• I am also still developing a new branch named mpc-pid-dev which I implemented Adaptive Fine-tuned PID Path Following for Turtlebot3 also immediately outputs all the plots after running on Gazebo using Gnuplot. I also configured the LiDAR scan in mpc_obstacle_avoid.cpp (currently under developing) so that it can converge to the optimal solution faster with more adding obstacles. Hopefully I can merge this branch to the master branch soon. Please check mpc-pid-dev branch for latest update.

• Also, I’m still in the 3rd semester of my 2nd year master’s at UMich so I will update this repo slowly.