GPIR

This is the project page of the paper "Real-Time Trajectory Planning for Autonomous Driving with Gaussian Process and Incremental Refinement", to appear in ICRA 2022. GPIR is developed aiming to provide an unified real-time trajectory planning framework in dynamic environment for autonomous driving.

About

Demo video: YouTube

Author: Jie Cheng, Yingbing Chen, Qingwen Zhang, Lu Gan and Ming Liu, RAM-LAB

If you find this repo helpful for your research, please cite our related paper.

@inproceedings{Jie2022gpir,
  title        = {Real-Time Trajectory Planning for Autonomous Driving with {G}aussian Process and Incremental Refinement},
  author       = {Cheng, Jie and Chen, Yingbing and Zhang, Qingwen and Gan, Lu and Liu, Chengju and Liu, Ming},
  booktitle    = {2022 IEEE International Conference on Robotics and Automation (ICRA)},
  year         = {2022},
  organization = {IEEE}
}

Prerequisites

Our software is developed and tested in Ubuntu20.04 with ROS noetic. Follow this link to install noetic.

Simulation environment

We use Carla-0.9.11 as the simulator (download via this link), and recommend to use a conda environment.

conda create -n carla_py37 python=3.7

To use the Carla PythonAPI, you need to export the path or install it

export PYTHONPATH=$PYTHONPATH:/carla_directory/PythonAPI/carla/dist/carla-0.9.11-py3.7-linux-x86_64.egg

or

pip install /carla_directory/PythonAPI/carla/dist/carla-0.9.11-py3.7-linux-x86_64.egg

Note that ROS melodic or below use Python2, and you may need to change python version above accordingly.

Build

Clone the repository

cd ${YOUR_WORKSPACE}/src
git clone --recurse-submodules https://github.com/jchengai/gpir.git

Setup dependencies

cd gpir 
conda activate carla_py37
pip install -r ./requirements.txt
bash ./setup.sh

You may check the pip path via which pip.

Build gpir

cd ../../ 
catkin_make -DCMAKE_BUILD_TYPE=RELEASE -j4
source ./devel/setup.sh

Run a demo

Start carla-0.9.11

cd to_your_carla_folder && ./carla.sh

Start carla ros bridge

roslaunch planning_core carla_setup.launch

Start gp_planner

roslaunch planning_core planning.launch

Start the control console

Since this repo does not contain the behavior planner, we provide a simple interface for manually controlling the behavior of the Ego vehicle.

roslaunch planning_core keyboard.launch

Acknowledgements

We would like to express sincere thanks to the authors of the following tools and packages:

• Opendrive map parser: ad_map
• QP solver: osqp
• iSAM solver: GTSAM
• Visualization: EPSILON
• Spline smoothing & some math utils: Apollo

Please let me know if your work is not properly cited or acknowledged, and thanks again for these authors who kindly open-sourcing their work to the community.

License

The source code is released under MIT license.

Disclaimer

This is research code, it is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of merchantability or fitness for a particular purpose.