Lemon-03/il_aloha_hammer_strikes_table_v1

LeRobot Dataset: ALOHA Hammer Strikes Table (Joint Control)

This dataset was created using LeRobot. It contains 50 expert episodes of the ALOHA robot performing a "Hammer Strike" task in a MuJoCo simulation.

⚠️ Important Note: This dataset uses Joint Position Control (Joint Space), not End-Effector Control (Cartesian/XYZ). This ensures stability and compatibility with standard ACT training pipelines, eliminating the "drifting" issues associated with IK mismatches.

🎥 Dataset Preview

Top  |  Left Wrist  |  Right Wrist

Task Description

The task ACTAlohaHammerStrikesTable-v0 involves complex bi-manual coordination:

1. Grasp: The right arm approaches and grasps the hammer.
2. Handover: The right arm lifts the hammer and passes it to the left arm.
3. Strike: The left arm accepts the hammer and performs a striking motion on the target surface.

This dataset serves as a benchmark for testing Imitation Learning algorithms (specifically ACT) on tasks requiring high precision and dual-arm coordination.

Technical Specifications

Action Space & Observation State

Unlike datasets that use XYZ coordinates (End-Effector position), this dataset records raw Joint Angles.

• Shape: (14,)
• Format:
  • Indices 0-5: Left Arm Joints (Waist, Shoulder, Elbow, Forearm Roll, Wrist Angle, Wrist Rotate)
  • Index 6: Left Gripper (0.0 = Closed, 1.0 = Open)
  • Indices 7-12: Right Arm Joints
  • Index 13: Right Gripper

Cameras

To optimize for training speed and storage, video resolution is set to:

• Resolution: 360x240 (Width x Height)
• Codec: AV1
• FPS: 50

💻 How to use with LeRobot

from lerobot.datasets.lerobot_dataset import LeRobotDataset <---（Lerobto v4.3.0）

# Load the dataset
dataset = LeRobotDataset("Lemon-03/il_aloha_hammer_strikes_table_v1")

# Check content
print(f"Total episodes: {dataset.num_episodes}")
print(f"Cameras: {dataset.camera_keys}") # ['overhead_cam', 'wrist_cam_left', 'wrist_cam_right']

Dataset Structure

meta/info.json:

{
    "codebase_version": "v3.0",
    "robot_type": "aloha",
    "total_episodes": 50,
    "total_frames": 133000,
    "total_tasks": 1,
    "chunks_size": 1000,
    "data_files_size_in_mb": 100,
    "video_files_size_in_mb": 200,
    "fps": 50,
    "splits": {
        "train": "0:50"
    },
    "data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet",
    "video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4",
    "features": {
        "observation.state": {
            "dtype": "float32",
            "shape": [14],
            "names": [
                "left_waist", "left_shoulder", "left_elbow", "left_forearm_roll", "left_wrist_angle", "left_wrist_rotate", "left_gripper",
                "right_waist", "right_shoulder", "right_elbow", "right_forearm_roll", "right_wrist_angle", "right_wrist_rotate", "right_gripper"
            ]
        },
        "action": {
            "dtype": "float32",
            "shape": [14],
            "names": [
                "left_waist", "left_shoulder", "left_elbow", "left_forearm_roll", "left_wrist_angle", "left_wrist_rotate", "left_gripper",
                "right_waist", "right_shoulder", "right_elbow", "right_forearm_roll", "right_wrist_angle", "right_wrist_rotate", "right_gripper"
            ]
        },
        "observation.images.top": {
            "dtype": "video",
            "shape": [240, 360, 3],
            "names": ["height", "width", "channel"],
            "info": {
                "video.codec": "av1",
                "video.fps": 50
            }
        },
        "observation.images.wrist_cam_left": {
            "dtype": "video",
            "shape": [240, 360, 3],
            "names": ["height", "width", "channel"],
            "info": {
                "video.codec": "av1",
                "video.fps": 50
            }
        },
        "observation.images.wrist_cam_right": {
            "dtype": "video",
            "shape": [240, 360, 3],
            "names": ["height", "width", "channel"],
            "info": {
                "video.codec": "av1",
                "video.fps": 50
            }
        }
    }
}

Citation

BibTeX:

@misc{lerobot_hammer_strikes_2025,
  author = {Lemon-03},
  title = {ALOHA Hammer Strikes Table Dataset (Joint Control)},
  year = {2025},
  publisher = {Hugging Face},
  howpublished = {\url{[https://huggingface.co/datasets/Lemon-03/il_aloha_hammer_strikes_table](https://huggingface.co/datasets/Lemon-03/il_aloha_hammer_strikes_table)}},
}