fuxi-robot/excavator-motion

Introduction

  Excavator-motion is a large-scale dataset of excavator motion trajectories, encompassing three distinct excavator models. Each H5 file in the dataset records a complete motion trajectory from the initiation of digging to the full loading of a truck. Compared to our another released dataset Excavator-video, excavator-motion not only contains RGB images from identical view but also includes elevation maps in global view and fine-grained excavator joint angle measurements. Building upon excavator-motion, we invite you to develop specialized excavator models and collaborate with us to advance the frontier of intelligent construction machinery.

Download the Dataset

We recommend you use the official download tools huggingface-cli to download the dataset. To download the full dataset, you can use the following command：

# install huggingface-cli
pip install huggingface-hub

# download the dataset
huggingface-cli download fuxi-robot/excavator-motion --repo-type dataset --local-dir .

Or you only want to download a specific file, such as xcmg_data_2025-04-11-17-46-49.hdf5, you can use the following command：

huggingface-cli download fuxi-robot/excavator-motion xcmg_data_2025-04-11-17-46-49.hdf5 --repo-type dataset --local-dir .

Any issures about downloading or more detailed use of huggingface-cli, please refer to the official document Command Line Interface (CLI).

Data Structure

Group	Shape	Meaning
/observation/timestamp	[N]	timestamp in nanoseconds
/observation/images/main	[N, 384, 480, 3]	bgr image from main view
/observation/images/elevation	[N, 200, 200, 3]	elevation image from lidar
/observation/qpos	[N, 4]	joint angle in rad, [boom, arm, bucket, swing] in order
/action	[N, 4]	joint angle in rad at next timestep, [boom, arm, bucket, swing] in order

Visualization

We offer a tool for visualizing the .h5 data， helping you know the data more intuitively. You can visualize the data, such as xcmg_data_2025-04-11-17-46-49.hdf5, using the following command.

python scripts/visualize.py --excavator=75 --data_path=./data/75/xcmg_data_2025-04-11-17-46-49.hdf5 --out_dir=. --format=mp4  --fps=10

Here is the visualization of some examples. Notably, the original h5 data has a frame rate of 10 fps, and the videos here were accelerated by a factor of three.

Convert to LeRobot

We also offer a tool for converting the .h5 file to lerobot format. You can start conversion using the following command.

python scripts/excavator_h5_to_lerobot.py --data_dir=data/75 --out_dir=lerobot_examples