Continuous Control

Continuous control — learning smooth motor commands in simulated physics — was transformed by MuJoCo and the OpenAI Gym suite in the mid-2010s. SAC (2018) and TD3 became reliable baselines, but the field shifted toward harder locomotion (humanoid parkour, dexterous hands) and sim-to-real transfer after DeepMind's dm_control and Isaac Gym raised the bar. DreamerV3 (2023) showed that world-model approaches can match or beat model-free methods across dozens of control tasks with a single hyperparameter set, signaling a move toward generalist RL agents.

1
Datasets
12
Results
average-return
Canonical metric
Canonical Benchmark

MuJoCo

Physics-based continuous control benchmark. Evaluated on 15 DMControl tasks; metric is mean normalized score (0=random, 1000=expert) at 1M environment steps.

Primary metric: average-return
View full leaderboard

Top 10

Leading models on MuJoCo.

RankModelaverage-returnYearSource
1
TD3
55922026paper
2
SAC
51792026paper
3
PPO
20382026paper
4
TD-MPC2 (317M params)
9602026paper
5
TD-MPC2 (19M params)
9532026paper
6
FOWM
9452026paper
7
BRO
9412026paper
8
TD-MPC2 (5M params)
9292026paper
9
DreamerV3
8972026paper
10
TD-MPC
8572026paper

What were you looking for on Continuous Control?

Didn't find the model, metric, or dataset you needed? Tell us in one line. We read every message and reply within 48 hours.

All datasets

1 dataset tracked for this task.

Related tasks

Other tasks in Reinforcement Learning.

Reply within 48 hours · No newsletter

Didn't find what you came for?

Still looking for something on Continuous Control? A missing model, a stale score, a benchmark we should cover — drop it here and we'll handle it.

Real humans read every message. We track what people are asking for and prioritize accordingly.