PyTorch 1.x Reinforcement Learning Cookbook

How to do it...

We'll develop the Flappy Bird game environment using Pygame as follows:

1. We start by developing a utility function that loads images and transforms them into the right format:

>>> from pygame.image import load >>> from pygame.surfarray import pixels_alpha >>> from pygame.transform import rotate >>> def load_images(sprites_path): ...     base_image = load(sprites_path +                              'base.png').convert_alpha() ...     background_image = load(sprites_path +                              'background-black.png').convert() ...     pipe_images = [rotate(load(sprites_path +                         'pipe-green.png').convert_alpha(), 180), ...                    load(sprites_path +                              'pipe-green.png').convert_alpha()] ...     bird_images = [load(sprites_path +                            'redbird-upflap.png').convert_alpha(), ...                    load(sprites_path +  'redbird-midflap.png').convert_alpha(), ...