Deep Reinforcement Learning - Hit a Moving Target

My Deep Reinforcement Learning Example inspired by the Google DeepMind "Playing Atari with Deep Reinforcement Learning" paper. https://www.cs.toronto.edu/~vmnih/docs/dqn.pdf https://deepmind.com/research/publications/playing-atari-deep-reinforcement-learning/ The machine learns by itself to hit a moving target. The target moves horizontally at constant speed, but its starting point is chosen randomly at each episode. The color red means missed, green hit. Perfect score is reached after 1000 epochs of training.