Here a ROS software engineer,
Well, I have experience not only with Nano board, but also with TX1/2, so, I will help you with the ROS setup.
About the DLRL, I've been used Gym(a OpenAI package) to perform DLRL on my projects, I've also developed a RL state-of-the-art algorithm for tracking with a Husky Robot.
Contact me for further details.