As we discussed in the first chapter, OpenAI Gym is an attempt to standardize reinforcement learning research and development and to compare RL models to each other for the purposes of developing baseline research frameworks.
The following screenshot is a still from the Neon Race Car environment with OpenAI Gym and Universe:
As RL researchers, we want to be able to develop benchmarks and widely-used, well-known training and testing datasets like the ones available for supervised learning, such as ImageNet for image recognition, the familiar iris dataset from the UCI Machine Learning Repository, or the MNIST handwritten digit dataset.
The RL analogue for a widely-used labeled training dataset is a standardized set of environments such as the one that Gym provides. A standardized set of environments lets us compare the work of different researchers,...