DeepMind Control Suite

By Yuval Tassa, Yotam Doron, Alistair Muldal, Tom Erez, Yazhe Li, Diego de Las Casas, David M. Budden, Abbas Abdolmaleki, Josh Merel, Andrew Lefrancq and 2 more. 2018

1,440 citations recorded by the source corpus.

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 this https URL A video summary of all tasks is available at this http URL .

Subfields: Reinforcement Learning, Evaluation & Benchmarks

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Reinforcement Learning

In Evaluation & Benchmarks

Citations in the corpus

Learning Latent Dynamics for Planning from Pixels (2018, methodology): “For our evaluation, we consider six image-based continuous control tasks of the DeepMind control suite (Tassa et al., 2018), shown in Figure 1.”
Distributed Prioritized Experience Replay (2018, methodology): “…gradient system based on DDPG (Lillicrap et al., 2016), an implementation of deterministic policy gradients Silver et al. (2014) also similar to older methods (Werbos, 1990; Prokhorov & Wunsch, 1997), and tested it on continuous control tasks from the DeepMind Control Suite (Tassa et al., 2018).”
CausalWorld: A Robotic Manipulation Benchmark for Causal Structure and Transfer Learning (2020, methodology): “Previous benchmarks proposed for RL mostly focused on the single task learning setting such as OpenAI Gym and DM control suite (Tassa et al., 2018; Brockman et al., 2016).”
Unsupervised Learning of Object Keypoints for Perception and Control (2019, methodology): “We report precision and recall for trajectories of varying lengths (lengths =1 – 200 frames; each frame corresponds to 4 action repeats) against ground-truth keypoints on Atari ALE [1] and Manipulator [37] domains.”
Keep Doing What Worked: Behavioral Modelling Priors for Offline Reinforcement Learning (2020, methodology): “In a first set of experiments we compare our algorithm to strong off-policy baselines on tasks from the DeepMind control suite (Tassa et al., 2018) – to give a reference point as to how our algorithm performs on common benchmarks.”
Critic Regularized Regression (2020, methodology): “D4PG has been used successfully to solve a variety of problems [35] as well as for learning from a combination of ofﬂine (and off-task) and online data [6].”
COMBO: Conservative Offline Model-Based Policy Optimization (2021, methodology): “To answer question (2), we evaluate COMBO on two image-based environments: the standard walker ( walker-walk ) task from the the DeepMind Control suite [61] and a visual door opening environment with a Sawyer robotic arm ( sawyer-door ) as used in Section 5.1.”
SoftGym: Benchmarking Deep Reinforcement Learning for Deformable Object Manipulation (2020, background): “In contrast, there are a number of popular benchmarks for reinforcement learning with rigid or articulated objects [13, 14, 15].”
Acme: A Research Framework for Distributed Reinforcement Learning (2020, methodology): “The DeepMind Control Suite (Tassa et al., 2018) provides a set of continuous control tasks in MuJoCo (Todorov et al., 2012) and has been widely used as a benchmark to assess performance of continuous control algorithms.”
Reinforcement Learning with Augmented Data (2020, methodology): “We highlight the main contributions of RAD below: • We show that simple end-to-end RL algorithms coupled with augmented data either match or beat every state-of-the-art baseline in terms of performance and data-efﬁciency across 15 DeepMind Control environments [22].”
Image Augmentation Is All You Need: Regularizing Deep Reinforcement Learning from Pixels (2020, methodology): “The PlaNet benchmark [22] consists of six challenging control tasks from [43] with different traits.”
Meta-World: A Benchmark and Evaluation for Multi-Task and Meta Reinforcement Learning (2019, background): “ for robots to acquire new skills as quickly as people can. 2Related Work Previous works that have proposed benchmarks for reinforcement learning have largely focused on single task learning settings [11,12,13]. One popular benchmark used to study multi-task learning 2 is the Arcade Learning Environ”
RLBench: The Robot Learning Benchmark & Learning Environment (2019, methodology): “These learned (continuous control) agents are usually tested on benchmarks such as OpenAI Gym [10] or the DeepMind Control Suite [11].”
Maximum a Posteriori Policy Optimisation (2018, methodology): “Speciﬁcally, we start by looking at the continuous control tasks of the DeepMind Control Suite (Tassa et al. (2018), see Figure 1), and then consider the challenging parkour environments recently published in Heess et al. (2017).”
Google Research Football: A Novel Reinforcement Learning Environment (2019, methodology): “llenging, and resulting environments often provide a useful playground for research questions centered around a speciﬁc reinforcement learning set of topics. For instance, the DeepMind Control Suite (Tassa et al., 2018) focuses on continuous control, the AI Safety Gridworlds (Leike et al., 2017) on ”
Distributed Distributional Deterministic Policy Gradients (2018, methodology): “We ﬁrst consider evaluating performance on a number of simple, physical control tasks by utilizing a suite of benchmark tasks (Tassa et al., 2018) developed in the MuJoCo physics simulator (Todorov et al., 2012).”