mbendjilali / PIE

Python version of the RRT-Star algorithm in a 3D environment

PIE Autonomous UAV / Path Planning

Python version of the RRT-Star algorithm in a 3D mesh environment. This code is meant to be run in a ROS environment.

Description

This code is an attempt at implementing a 3D version of the previously coded RRT-start by Fanjin Zeng using a polygonal representation of the obstacles.

It is parted in 5 different Python files :

• ClassDefinition.py -> Defines the most basic structures needed to implement an RRT algorithm, i.e. what a graph, a line and a window are.
• Geometry.py -> Sets up collision criteria for objects in a 3D space (spheres, cubes, and triangles).
• PathPlanning.py -> Implements the RRT-Star and RRT-Connect (currently not working) algorithms. It also retrieves the correct path_list from start to goal within a successful tree.
• meshRecon.py -> Uses the Alpha Shape algorithm to generate a triangle set from a point cloud.
• main.py -> Sets up every parameter, loads a point cloud, derives the RRT algorithm and plots the result.

Getting Started

Dependencies

• Pyhon 3.6 and latest releases
• numpy, random, time, open3d, matplotlib.pyplot, mpl_toolkits.mplot3d.art3d

Installing

• You can directly download the files from the main branch of this git.
• Some obsolete functions are still implemented, but not called, within the files, they are marked as such.

Executing program

• You can start off by running main.py, but you can also add these files to your current projects:
• Comment the following line and provide your own point cloud (pcd).

    pcd = list2PointCloud(txt2list('obstacles.txt'))
    mesh = pointCloud2Mesh(pcd)

Help

This code has been written using the latest Python release at the time, I do not guarantee that it will work with previous ones. Also, almost every function is coded with an added docstring to it,so help yourself and call help():

help(RRT_star)
Help on function RRT_star in module PathPlanning:
RRT_star(startpos, endpos, obstacles, window, n_iter, stepSize, radius=None)
    Implements the RRT Star algorithm
    :param startpos: starting position -> tuple of size 3
    :param endpos: goal position -> tuple of size 3
    :param obstacles: sets of triangles -> tuple of 2 arrays containing edges positions and idx
    :param window: search window
    :param n_iter: maximum number of iterations
    :param stepSize: minimum increment in the tree
    :param radius: obsolete parameter for polygonal representation of the obstacles
    :return: graph

Authors

Contributors names and contact info

ex. Moussa Bendjilali at [e-mail]

Version History

• 2.0
  • Various bug fixes and optimizations
• 1.0
  • Initial Release

Acknowledgments

Inspiration, code snippets, etc.

• Fanjin Zeng