pjpo-curvature

This is the project page of the paper "Path Planning for Autonomous Driving with Curvature-considered Quadratic Optimization", to appear in IV 2023.

About

The proposed planner generates the curvature-concerned paths in structure scenarios.

Prerequisites

1. Build HSL (recommended for better performance)

   git clone https://github.com/coin-or-tools/ThirdParty-HSL.git --depth=1
   
   # Obtain a tarball with HSL source code from http://www.hsl.rl.ac.uk/ipopt/ and unpack this tarball
   gunzip coinhsl-x.y.z.tar.gz
   tar xf coinhsl-x.y.z.tar
   
   # Rename the directory `coinhsl-x.y.z` to `coinhsl`, or set a symbolic link:
   ln -s coinhsl-x.y.z coinhsl
   
   ./configure
   make
   sudo make install

2. Build OSQP-eigen

Build

Clone repository to any catkin workspace and compile workspace

  cd ~/catkin_ws/src
  git clone https://github.com/badaolaziang/pjpo-curvature
  cd .. && catkin_make
  source devel/setup.bash

Evaluate the Taylor estimation

  // curtvature estimation
  python3 src/scripts/dynamic_kappa_warmstart_estimation.py
  // squared-curvature estimation
  python3 src/scripts/dynamic_kappa_square_warmstart_estimation.py

Run a demo

1. launch the code:

  roslaunch pjpo_curvature combine.launch

2. run the planner:

   set a random 2D goal in the rviz to start the calculation, and get the path(s) in the rviz window.

3. change the parameters of src/pjpo_curvature/launch/combine.launch and src/pjpo_curvature/script/reference_publisher.py to change the state of the starting pose and the shape of reference path.

Acknowledgement

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

Guide line and framework:Cartesian_planner

QP solver: OSQP&OSQP-eigen