search

SaidGuerazem / A-Deep-Reinforcement-Learning-Approach-for-Active-Slam-Exploration

1 stars 0 forks source link

How to implement your job? #1

Open nsa05605 opened 4 weeks ago

nsa05605 commented 4 weeks ago

Hi,

I want to implement your code on my ubuntu 20.04 ROS noetic. Can you explain how to implement?

Thank you.

SaidGuerazem commented 4 weeks ago

Hey!

We've done this work using Ubuntu 18.04 with the ROS Kinetic distribution. The original package we used was outdated and archived, so we had to make several changes to implement it.

Here is the link to the original package: gym-gazebo.

If you want to switch to Ubuntu 20.04, you can start by setting up Gazebo as an environment in your Gym/OpenAI Gym library. This will allow you to create a wrapper and modify the simulation file you are going to use, set up the ROS nodes, and establish communication between your ROS Master and Gazebo Master. This will involve tasks like deploying TurtleBot in the world, etc.

The next step would be to create a wrapper to modify the reward system, the action set, and other parameters. This step is largely dependent on your familiarity with Gym. The final step would be to develop the algorithm (in our case, it was D3QN+) and launch the training directly.

This is a general overview, and I hope it will help you in making the transition. My team and I may try to launch our own version as well.

If you have any other questions, I would be happy to answer them.

Kind regards, Said