DeepMind Control Suite
It provides a performance benchmark for reinforcement learning researchers, but is incremental as it standardizes existing tasks.
The paper introduces the DeepMind Control Suite, a standardized set of continuous control tasks with interpretable rewards designed to benchmark reinforcement learning agents, and includes benchmarks for several algorithms.
The DeepMind Control Suite is a set of continuous control tasks with a standardised structure and interpretable rewards, intended to serve as performance benchmarks for reinforcement learning agents. The tasks are written in Python and powered by the MuJoCo physics engine, making them easy to use and modify. We include benchmarks for several learning algorithms. The Control Suite is publicly available at https://www.github.com/deepmind/dm_control . A video summary of all tasks is available at http://youtu.be/rAai4QzcYbs .