Millipede-Control-with-Reinforcement-Learning-and-VAEs Low Dimensional Embeddings for High-Dimensional Control of a Simulated Millipede