Reinforcement learning simulations

Nobody tells the AI the answer: it fails, tries again thousands of times and works it out by itself. Peppo lasting 30 days on a desert island and Peppo learning to park like a human driver are up now.