Add dataset_list.md

README.md

@@ -11,7 +11,7 @@ Software that integrates various imitation learning methods and benchmark task e
 ## Install
 See [the installation documentation](./doc/install.md).
 
-## Models
+## Policies
 ### [SARNN](./robo_manip_baselines/sarnn)
 Spatial attention recurrent neural network
 
@@ -24,7 +24,11 @@ Diffusion Policy
 ### [MT-ACT](./robo_manip_baselines/mt_act)
 Multi-Task Action Chunking Transformer
 
-## Data collection by teleoperation
+## Data
+### Publicly available datasets
+See [the dataset list](./doc/public_datasets.md).
+
+### Data collection by teleoperation
 See [teleop](./robo_manip_baselines/teleop).
 
 ## Environments for robot manipulation

doc/dataset_list.md

@@ -0,0 +1,37 @@
+# Dataset list
+
+## Demonstrations in MuJoCo environments
+| Task  | Link |
+| --- | --- |
+| [TeleopMujocoUR5eCable](./environment_catalog.md#MujocoUR5eCableEnv) | [Download (9GB)]() |
+| [TeleopMujocoUR5eRing](./environment_catalog.md#MujocoUR5eRingEnv) | [Download (8GB)](https://www.dropbox.com/scl/fi/cg3qd7k5scmpxnj0t4qa5/TeleopMujocoUR5eRing_Dataset30_20241031.zip?rlkey=jgbwglrqi7svvrggpawrazg5r&dl=1) |
+| [TeleopMujocoUR5eParticle](./environment_catalog.md#MujocoUR5eParticleEnv) | [Download (13GB)]() |
+| [TeleopMujocoUR5eCloth](./environment_catalog.md#MujocoUR5eClothEnv) | [Download (8GB)](https://www.dropbox.com/scl/fi/ums7qz2rom9focuf91j87/TeleopMujocoUR5eCloth_Dataset30_20241031.zip?rlkey=qq10s4y5gi8stbondnsoso31l&dl=1) |
+
+## Demonstrations in real-world environments
+Coming soon.
+
+## Dataset format
+| Entry  | Description | Shape | Dtype |
+| --- | --- | --- | --- |
+| `time` | time | `(T,)` | `float64` |
+| `measured_joint_pos` | Measured joint position [rad] | `(T, JointDim)` | `float64` |
+| `command_joint_pos` | Command joint position [rad] | `(T, JointDim)` | `float64` |
+| `measured_joint_vel` | Measured joint velocity [rad/s] | `(T, JointDim)` | `float64` |
+| `command_joint_vel` | Not set |  |  |
+| `measured_joint_torque` | Not set |  |  |
+| `command_joint_torque` | Not set |  |  |
+| `measured_eef_pose` | Measured pose of end-effector [m], [rad] | `(T, PoseDim)` | `float64` |
+| `command_eef_pose` | Command pose of end-effector [m], [rad] | `(T, PoseDim)` | `float64` |
+| `measured_eef_vel` | Not set |  |  |
+| `command_eef_vel` | Not set |  |  |
+| `measured_eef_wrench` | Measured force and torque of end-effector [N], [Nm] | `(T, WrenchDim)` | `float64` |
+| `command_eef_wrench` | Not set |  |  |
+| `<camera_name>_rgb_image` | RGB image | `(T,)` | `object` |
+| `<camera_name>_depth_image` | Depth image | `(T, Height, Width)` | `float32` |
+| `<camera_name>_depth_image_fovy` | Field of view of depth image height | `()` | `float64` |
+| `format` | Data format (fixed with `"RoboManipBaselines-TeleopData"`) | `()` | `str` |
+| `version` | Version of RoboManipBaselines | `()` | `str` |
+| `demo` | Demonstration name | `()` | `str` |
+| `world_idx` | World index | `()` | `int64` |
+`T` is the length of the time sequence, `JointDim` is the number of joints, `PoseDim = 7` is pose (tx, ty, tz, qw, qx, qy, qz), `WrenchDim = 6` is wrench (fx, fy, fz, nx, ny, nz), respectively.