/DRL-based_UAV_Motion_Planning

code for `A Hybrid Human-in-the-Loop Deep Reinforcement Learning Method for UAV Motion Planning'

Primary LanguagePython

DRL-based_UAV_Motion_Planning

code for `A Hybrid Human-in-the-Loop Deep Reinforcement Learning Method for UAV Motion Planning' Google Scholar

Dependencies

Ubuntu 20.04 LTS
ROS Noetic
gazebo 11
Python 3.8

Installation

This project is based on the project Prometheus. To install it, please follow the instructions in Prometheus wiki.

Notice:
the guide Installation and Compilation here recommends installing Ubuntu 18.04 and ROS Melodic. However, Ubuntu 20.04 and ROS Noetic are also compatible. To use them, you may need to make modifications in the following places.

During the DRL training process, resetting the drone's position at the beginning of each episode can cause issues with the onboard sensors. Modifying the following two places (in Prometheus/Modules/control/src/px4_sender.cpp) might be helpful:

I have made revisions to several files (such as sitl.launch, models, worlds, etc.) in gazebo_simulator to accommodate DRL training. gazebo_simulator that I used has been uploaded here for your reference.

How to run

  • Run the command in the terminal roslaunch prometheus_gazebo sitl.launch
  • Run train.py