UAV Path Planning in 3D Unstructured Environment

Author: Yogeshkarna Govindaraj

Introduction

This repository contains implementation of UAV 3D Path planning algorithms. The global path planning algorithm based on Simulated Annealing (SA) and local path planning algorithms I-PRM and A*. For Lazy-Theta* the recent implementation by [1] is used by modifying the dataset.

The original pointcloud files of datasets are too large for github repository, The .pcd files can be downloaded into dataset folder from below link.

Link to Pointcloud Datasets

https://hochschulebonnrheinsieg-my.sharepoint.com/:f:/g/personal/yogeshkarna_govindaraj_365h-brs_de/EuGETRMeFsNHuxAuSCPIrnUBsj4sKXes7GBqepAkbaC8Uw?e=f24C6K

Installation

The project is developed in anaconda environment. The requirements.yml file contans all the dependencies required to run all path planning programs contained in src folder.

conda env update -n my_env --file requirements.yml

The above command creates a new anaconda environment named my_env consisting of all the dependencies required to run the path planners.

Repository structure

dataset folder contains octomap files of all three datasets used for testing the local path planning algorithms. The src folder contains three .ipynb files where SA implements the global path planning solution, A-star and I-PRM files implements the respective local path planning algorithm.

All these programs are standalone containing program explanation within. For Lazy-Theta* an method proposed by [1] is used, which can be found in the link below.

Link to Lazy-Theta* Implementation

https://sites.google.com/view/lazythetastaronline/autonomous-exploration?authuser=0

References

1. Faria, M., Ferreira, A. S., Pérez-Leon, H., Maza, I., & Viguria, A. (2019). Autonomous 3D Exploration of Large Structures Using an UAV Equipped with a 2D LIDAR. Sensors, 19(22), 4849. DOI 10.3390/s19224849