Using Reinforcement Learning to Herd a Robotic Swarm to a Target Distribution