Repo to run a Turtlebot3 Reinforcement Learning Office Environment.
This is built on the amazing work of the ROBOTIS Open Source Team. It is easier to run since it doesn't require a local ROS installation as we use Docker.
Enable ports for GUI to run properly:
xhost +local:root
Go to the docker_setup folder and build the image:
sudo docker build . -t ros_ml
Start docker container:
. docker_start.sh
In the container start simulation and agent:
. /my_scripts/start_ros.sh <$STAGE_PARAM>
where <$STAGE_PARAM>
defaults to 1 and can be replaced with a number from 1 to 4
depending on the stage you want to start.
This can only be set once, if you want to change it afterwards you need to restart the
Docker container.
If you didn't map a volume in the docker_start.sh
script you need to set
save_model_to_disk
and save_to_db
in turtlebot3_dqn.py
to False
.
If no simulation is starting on your screen make sure $DISPLAY
is set correctly.
In the Gazebo Gui on the left side on the world tab click Physics and set real time update to 0. This will make Gazebo run as fast as possible.
To increase FPS change view angle to orthographic.