Datasets:

armandpl
/

real_cartpole_200k

Tasks:

Reinforcement Learning

Dataset card Files Files and versions Community

armandpl commited on Aug 5

Commit

d57c5be

•

1 Parent(s): e7ef15e

Update README.md

Files changed (1) hide show

README.md +3 -2

README.md CHANGED Viewed

@@ -3,8 +3,9 @@ task_categories:
 - reinforcement-learning
 - robotics
 ---
-This dataset contains sequences of actions, motor angles and pendulum angles for a [rotary inverted pendulum robot](https://github.com/armandpl/furuta). I collected it while training it to swing up and balance ([wandb run](https://wandb.ai/armandpl/furuta2/runs/bc2yhhpg)).
-Angles are in radian. The action maps to the motor voltage with:
 ```Python
   deadzone = 0.1
   center = 0.05

 - reinforcement-learning
 - robotics
 ---
+This dataset contains sequences of actions, motor angles and pendulum angles as well as velocities for a [rotary inverted pendulum robot](https://github.com/armandpl/furuta). I collected it while training it to swing up and balance ([wandb run](https://wandb.ai/armandpl/furuta2/runs/bc2yhhpg)).
+Angles are in radian. Velocities were computed from the angles and fed to the policy.
+The action maps to the motor voltage with:
 ```Python
   deadzone = 0.1
   center = 0.05