Upload README.md with huggingface_hub

README.md

@@ -4,120 +4,17 @@ tags:
 - robotics
 - manipulation
 - lerobot
-- ros2
-- dual-arm
-- robot-learning
-size_categories:
-- 1K<n<10K
 ---
 
 # Robot Manipulation Dataset
 
-A LeRobot-compatible dataset created from ROS2 bag data with automatic metadata generation.
+A LeRobot-compatible dataset with 9 episodes and 2,905 frames.
 
-## Dataset Overview
+## Quick Stats
+- Episodes: 9
+- Frames: 2,905  
+- Joints: 34
+- Duration: 31.04s
 
-- **Total Episodes**: 9
-- **Total Frames**: 2,905
-- **Duration**: 31.04 seconds
-- **FPS**: 100.03
-- **Robot Type**: Dual-arm robot with 34 joints
-- **Tasks**: Various manipulation tasks
-
-## Features
-
-- **Action**: 34 joint positions (float32)
-- **Observation.state**: 34 joint positions (float32)
-- **Timestamp**: Relative timestamps (float32)
-- **Episode/Frame indices**: Proper indexing
-- **Task descriptions**: Human-readable task names
-
-## Joint Configuration
-
-The dataset includes 34 joints:
-- Arm1: shoulder_pan, shoulder_lift, elbow, wrist_1, wrist_2, wrist_3
-- Arm2: shoulder_pan, shoulder_lift, elbow, wrist_1, wrist_2, wrist_3
-- Gripper joints: Various finger and palm joints
-
-## Episodes
-
-| Episode | Frames | Task | Duration |
-|---------|--------|------|----------|
-| 0 | 405 | Put the marker in the blue bin | 4.04s |
-| 1 | 254 | Place the object in the container | 2.53s |
-| 2 | 262 | Move the item to the target location | 2.62s |
-| 3 | 304 | Position the marker correctly | 3.04s |
-| 4 | 277 | Complete the manipulation task | 2.77s |
-| 5 | 279 | Put the marker in the blue bin | 2.79s |
-| 6 | 329 | Place the object in the container | 3.29s |
-| 7 | 357 | Move the item to the target location | 3.57s |
-| 8 | 438 | Position the marker correctly | 4.38s |
-
-## Usage
-
-### LeRobot Visualizer
-View the dataset interactively:
-```
-https://huggingface.co/spaces/lerobot/visualize_dataset?path=%2FSraghvi%2Fworking-detailed-stats-fixed%2Fepisode_0
-```
-
-### Python
-```python
-from datasets import load_dataset
-
-# Load the dataset
-dataset = load_dataset('Sraghvi/working-detailed-stats-fixed')
-
-# Access episodes
-episodes = dataset['train']
-print(f'Loaded {len(episodes)} frames')
-
-# Get first frame
-first_frame = episodes[0]
-print(f'Action shape: {len(first_frame["action"][0])} joints')
-print(f'Timestamp: {first_frame["timestamp"]}s')
-```
-
-## Data Format
-
-Each frame contains:
-- `action`: Target joint positions (34 joints)
-- `observation.state`: Current joint positions (34 joints)
-- `timestamp`: Relative time from episode start
-- `episode_index`: Episode number (0-8)
-- `frame_index`: Frame number within episode
-- `index`: Global frame index
-- `task_index`: Task identifier
-
-## Statistics
-
-The dataset includes comprehensive statistics for each episode:
-- Min/max/mean/std for all joint values
-- Timestamp ranges and distributions
-- Frame count and indexing
-- Mock image statistics (ready for real camera data)
-
-## Generation
-
-This dataset was created using an automatic ROS2 to LeRobot pipeline that:
-1. Parses ROS2 bag files
-2. Extracts joint state data
-3. Creates realistic episode splits
-4. Generates comprehensive metadata
-5. Uploads to HuggingFace Hub
-
-## License
-
-MIT License - See LICENSE file for details.
-
-## Citation
-
-If you use this dataset, please cite:
-```bibtex
-@dataset{working_detailed_stats_fixed,
-  title={Robot Manipulation Dataset from ROS2 Bag},
-  author={Sraghvi},
-  year={2024},
-  url={https://huggingface.co/datasets/Sraghvi/working-detailed-stats-fixed}
-}
-```
+## View Dataset
+[Open in LeRobot Visualizer](https://huggingface.co/spaces/lerobot/visualize_dataset?path=%2FSraghvi%2Fworking-detailed-stats-fixed%2Fepisode_0)