Train a balance bot with reinforcement learning

Shawn Hymel wrote a guide in Make: Volume 99 on training a two-wheeled balance bot with reinforcement learning instead of a classical control loop. The agent is trained with proximal policy optimiz…