Racing-LMPC-ROS2

C++ ROS2 packages that implement learning model predictive control for real-world autonomous race cars.

Paper: Learning Model Predictive Control with Error Dynamics Regression for Autonomous Racing

icra_2024_lmpc_video.-.Made.with.Clipchamp.7.mp4

Install

Dependencies

• Ubuntu 22.04
• ROS2 Humble
• CasADi >= 3.6.3. Must be built from source, since the pre-compiled binary is C++ 11 and therefore not ABI compatible with C++ 17.
  • See here for official build instructions.
  • Refer to the "Build Casadi" section in CI file here for a working build instruction. Note that we installed llvm for JIT compilation, and ipopt for nonlinear optimization. Also note that we appended /usr/local/lib to LD_LIBRARY_PATH in order to find the shared library libcasadi.so at runtime.
• Foxglove Studio (optional). Rviz2 is fine too. See here for installation instructions.

Build

# Clone the repo
cd Racing-LMPC-ROS2
git checkout humble-release
# Install rosdep dependencies
rosdep update
rosdep install --from-paths src --ignore-src -r -y
# Build
colcon build --packages-up-to racing_lmpc_launch --cmake-args -DCMAKE_BUILD_TYPE=Release

Run LMPC

Open Foxglove Studio. Load lmpc.foxglove.json layout file from the root of this repo. Open a Foxglove Bridge connection with the default port and IP setting.

In terminal 1, run the following command to launch the simulator:

source install/setup.bash
ros2 launch racing_lmpc_launch sim_barc_lmpc.launch.py

In terminal 2, run the following command to launch the foxglove bridge:

source install/setup.bash
ros2 launch foxglove_bridge foxglove_bridge_launch.xml

You can now use Foxglove Studio to visualize the simulation.

Run Tracking MPC

Same as above, except run the following command in terminal 1:

source install/setup.bash
ros2 launch racing_lmpc_launch sim_barc_tracking_mpc.launch.py

For IAC Putnam full course, run the following command in terminal 1:

source install/setup.bash
ros2 launch racing_lmpc_launch sim_putnam_config_a_tracking_mpc.launch.py

Remember to change the line scales of the 3D pannel accordingly to view them easily. Change display frame to base_link because the track is huge.

Cite As

@misc{xue2023lmpc,
      title={Learning Model Predictive Control with Error Dynamics Regression for Autonomous Racing}, 
      author={Haoru Xue and Edward L. Zhu and Francesco Borrelli},
      year={2023},
      eprint={2309.10716},
      archivePrefix={arXiv},
      primaryClass={cs.RO}
}