Two hours of Reinforcement Learning training of a Cartpole

Video shows a Gazebo simulation of a Cartpole robot controlled with ROS Control, being trained with a Reinforcement Learning algorithm, using the openai_ros package.

Data:
▸ We used ROS velocity controllers
▸ We used the cartpole simulation that you can find here:
▸ We used the openai_packages for the cartpole, the TaskEnvironment and the RobotEnvironment, that you can find here: http://wiki.ros.org/openai_ros
▸ The whole development and training process was done online in the ROS Development Studio: http://rosds.online
▸ It took almost two hours of training.