search

reiniscimurs / DRL-robot-navigation

Deep Reinforcement Learning for mobile robot navigation in ROS Gazebo simulator. Using Twin Delayed Deep Deterministic Policy Gradient (TD3) neural network, a robot learns to navigate to a random goal point in a simulated environment while avoiding obstacles.
MIT License
486 stars 97 forks source link

Problems in training #143

Open Sunsundayz1 opened 1 month ago

Sunsundayz1 commented 1 month ago

Hi, I'm training Gazebo environment. but I don't know where his target point is. Can you tell me please?

reiniscimurs commented 1 month ago

Please describe the question in more detail.

Sunsundayz1 commented 1 month ago

a9e994295aad1edba696f890a02f21bdHi, I'm not finding the target point in this training environment. d978e25e251f2f7c939a2641b7d70f74 I don't know how to determine if target how to reach

reiniscimurs commented 1 month ago

Target is not displayed in Gazebo environment. You can see it in Rviz though.

Setting and reaching target criteria are explained in the tutorial: https://medium.com/@reinis_86651/deep-reinforcement-learning-in-mobile-robot-navigation-tutorial-part4-environment-7e4bc672f590

Sunsundayz1 commented 1 month ago

I am a newbie and I don't know where I can see the target points and then please see my picture which shows no point cloud data. Please can you tell me what I should do. Secondly please do you have a pre-trained model? Thanks again. ![Uploading 42d9600c59389aa87cfdad6ca178c6b6.JPG…]()

reiniscimurs commented 1 month ago

I suggest following some tutorials regarding ROS and familiarizing yourself with ROS as that will help you understand the implementation. It will be very difficult to explain any issues you encounter without some basic knowledge there and without knowing the differences between Gazebo and Rviz. This kind of data you are looking for is displayed in Rviz and not in Gazebo GUI.

Sunsundayz1 commented 1 month ago

42d9600c59389aa87cfdad6ca178c6b6 I know that this is the gui for my rviz. but when I look at pointclound2 I get an error. Thank you very much for your patience!

reiniscimurs commented 1 month ago

You are specifying the wrong topic. We are using velodyne lidar for pointcloud, not camera. Please familiarize yourself with the tutorial. This can help you clear up confusion you might have about the implementation.

Sunsundayz1 commented 1 month ago

Thank you very much, I am now able to visualize the training properly, but I have one more request. Can you give me a pre-trained model. Thanks again for your generous sharing! 360298f7af93c971ce4d19b1a3c7bb4d

reiniscimurs commented 1 month ago

There are no pre-trained model weights available for this implementation.

Sunsundayz1 commented 1 month ago

Hi author, I didn't find the code for selecting the local target point in the code of this article, which is the core innovation in the paper, can you please answer the question,thanks again

reiniscimurs commented 1 month ago

This paper does not implement the whole functionality of the paper, just the DRL navigation training part.

Parts of the code in the paper are in: https://github.com/reiniscimurs/GDAE