halfcheetah

Benchmarks

TaskDatasetSOTA resultTrendResults

Locomotion

HalfCheetah F=1 (test)

6,540.5Average Reward

Offline Reinforcement Learning

halfcheetah jump

4,016.6Average Return

Reinforcement Learning Surrogate Modeling

HalfCheetah (HC) (test)

85Reward Ratio

Reinforcement Learning

HalfCheetah v4

5,381.07Mean Reward

Reinforcement Learning

HalfCheetah discrete-regime, exponential dwell

18,598Mean Return

Locomotion

halfcheetah expert v2

106.8Normalized Return

Reinforcement Learning

HalfCheetah Meta (train)

9,721Reward

Continuous Control

HalfCheetah

4.18Value Estimation MAE

Offline Reinforcement Learning

halfcheetah medium-expert-morph

63.9Score

Offline Reinforcement Learning

halfcheetah medium-expert-kinematic

57.4Score

Offline Reinforcement Learning

halfcheetah expert morph

76.4Score

Offline Reinforcement Learning

halfcheetah expert-kinematic

72.2Score

Constrained behavior synthesis

HalfCheetah 100 evaluation tasks (test)

98S

Offline Reinforcement Learning

HalfCheetah Medium-Replay Delayed Reward D4RL

42.4Score

Offline Reinforcement Learning

HalfCheetah Medium Delayed Reward D4RL

42.9Score

Reinforcement Learning

HalfCheetah target domain 1.5 density

11,515Reward

Reinforcement Learning

HalfCheetah 0.5 gravity (target)

7,559Reward

Reinforcement Learning

HalfCheetah 1.5 gravity MuJoCo (test)

6,894Reward

Safe Offline Multi-Agent Reinforcement Learning

2halfcheetah medium

0.51Reward

Locomotion

HalfCheetah non-stationary MuJoCo

15,369Final Training Reward

Offline-to-Online Reinforcement Learning

HalfCheetah D4RL Locomotion

49.37Return (HalfCheetah Random)

Robot Co-design

HalfCheetah search phase

1.36D2C Score

Noisy Action Robust MDP

Halfcheetah MuJoCo Noisy Action Robust MDP (high randomness)

1,765.71Worst-case Return

Noisy Action Robust MDP

Halfcheetah Medium Randomness MuJoCo Noisy Action Robust MDP

1,725.6Worst-case Return

Noisy Action Robust MDP

Halfcheetah MuJoCo Noisy Action Robust MDP (low randomness)

1,691.8Worst-case Return