https://medium.com/@monocosmo77/how-inverse-reinforcement-learning-works-part2-artificial-intelligence-aacc5a8c1f0c