Road Crossing

Overview

This package provides a ROS node to facilitate safe traverse of an autonomous robot across a road. It was done as a bachelor's thesis at FEE CTU.

The main algorithm is based on a behavior-tree structure. This work does not include the detection of incoming vehicles that are provided by a different node.

Installation

Clone this repository into your catkin workspace.

git clone https://github.com/vlk-jan/road_crossing.git

For this package to work you need to install additional packages:

• Compass - provides the robots azimuth.
• GPS to path - previously part of gps-navigation package.

The package also requires additional libraries:

• BehaviorTree.CPP - c++ library containing the behavior-tree implementation we use.
• GeographicLib - c++ library for geographical conversions.
• Numpy - python library for numerical calculations.
• Overpy - python library for accesing the OSM overpass API.
• Shapely - python library for work with geometrical objects.
• UTM - python library for bidirectional UTM <-> WGS84 conversion.

Installing c++ libraries:

sudo apt-get install libgeographic-dev ros-$ROS_DISTRO-behaviortree-cpp-v3

Installing python libraries

pip install -r requirements.txt

Simulation

This package can be simulated using the Gazebo simulator. The necessary information about the simulation environment, needed packages and more is available at the road_crossing_gazebo.

Documentation

This package does not currently have any documentation. However, the text of my bachelor thesis is here, where I try to explain the code a bit.

License

TODO

• Short-term
  • Create unit tests?
• Long-term
  • Service for obtaining better place - possible need to rewrite road cost
  • move_to_place weight map (point-cloud) creation & implementation
  • Vehicle obtaining (from ROS node) message type & implementation
  • Obtaining road info from real-world - using camera? - possibly different node, so just integration of the messages