Datasets:

leggedrobotics
/

egohtr

ArXiv:

License:

Dataset card Files Files and versions

xet

Community

You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

The USER must fill out this request only with an institutional e-mail address. Requests from non-institutional e-mail accounts will be rejected without further notice. The DATASET shall only be downloaded if the USER agrees to the following terms:

ETH Zurich grants the USER the non-exclusive, non-transferable, and non-sub-licensable license for using the DATASET. Permission is hereby granted, free of charge, to any person or company obtaining a copy of this software/dataset and associated files directly from the WEBSITE. The DATASET remains the property of ETH Zurich and/or its licensors and is made available as a service to the research community for non-commercial academic purposes. ETH Zurich retains the right to disclose and advertise the DATASET use by any licensee.
This license prohibits commercial use of parts or whole of the provided DATASET (including code, models, and data). This prohibition includes using the DATASET for the partial or total training of neural networks or other artificial intelligence systems for commercial use. Using the provided DATASET to promote a commercial entity or product or in any other manner that directly or indirectly results in commercial gains is strictly prohibited. The USER agrees to cite the DATASET in any publication based on, or that results from the use of the information contained in the DATASET.
Lease, sub-license, sell, package, or distribute parts or whole of the provided DATASET, the original or any updated version, as is or as part of other software or dataset, is prohibited. This ban includes redistributing the original or converted model parameters or any part of the provided model, code, and data to third parties via any platform or public repository.
Any use of the DATASET beyond non-commercial academic purposes is expressly prohibited, including, but not limited to, its use for identity recognition; for visual surveillance; to create “deep-fakes” or other misleading or deceptive content; or to create pornography or other inappropriate content. The USER shall further not carry out any procedures with the DATASET (linking, comparison, processing) with which any identity of a person could be derived. The de-anonymization of the DATASET or any action that allows re-identifying the participants in the DATASET is prohibited in any form. The USER agrees to comply with all applicable international and national laws, statutes, regulations, and guidelines, especially those concerning data protection.
The DATASET may be modified at any moment to have deleted, changed, or added information, generating an updated version of the DATASET. For example, if a participant in the DATASET asks to have their data removed. The USER will be contacted and requested to update their local copy of the DATASET in such a case. In addition, upon the first request by ETH Zurich, the right to use the DATASET will end immediately, and the USER will destroy any DATASET copy they may have and confirm the deletion to ETH Zurich.
The DATASET is provided “as is” without warranty of any kind, express or implied, including but not limited to the warranties of merchantability, fitness for a particular purpose, and non-infringement. In no event shall the authors or licensor be liable for any direct, indirect, incidental, special, exemplary, or consequential damages, whether in an action of contract, tort, or otherwise, arising from, out of, or in connection with the DATASET or the use of it. The licensor does not warrant any quality of the DATASET beyond the stated data-quality documented and reviewed in the DATASET documentation. In particular, the licensor does not warrant the accuracy and completeness of the DATASET, neither its suitability for the USER’s intended purpose. Using the DATASET is at the own risk of the USER.
These terms of use shall be construed, governed, interpreted, and enforced according to the laws of Switzerland without any regard to International Private Law. All disputes arising out of or concerning the access and use of the DATASET, as well as these terms of use, shall be brought before the competent court of the city of Zurich

Dataset Card for EgoHTR

Dataset Details

Dataset Description

EgoHTR (Egocentric Human-Terrain Reconstruction) is a multimodal dataset of 4D human motion precisely aligned to complex, unstructured terrain. It combines egocentric wearables (Project Aria) and IMU-based motion capture with portable 3D scene scanning to reconstruct human motion in diverse indoor and outdoor environments — enabling terrain-aware humanoid locomotion research where existing datasets, limited to controlled or unaligned capture, fall short.

Curated by: Alex Brandes and co-authors, ETH Zürich
Modality: Egocentric RGB video, Exocentric RGB video, IMU motion capture, 3D scene scans, optimized SMPL-X body models

Dataset Sources

Paper: EgoHTR: Egocentric 4D Demonstrations of Human Terrain Traversal (arXiv:2607.13472), accepted at CoRL 2026
Project page: https://egohtr.github.io
Generation pipeline (code): Github

Direct Use

Benchmarking 4D human–scene reconstruction and human mesh recovery methods, evaluated against MoCap ground truth
Training and evaluating perceptive, terrain-aware whole-body control / humanoid locomotion policies (demonstrated on a Unitree G1)
Retargeting human motion to robot embodiments via the accompanying pipeline

Dataset Structure

Each sequence folder contains the compressed processed outputs per sequence: full_sequence.npz (complete synchronized data), full_opt_sequence.npz, (optional: full_retarget_sequence.npz). Full field-level schema is documented in the EgoHTR pipeline repository.

Additionally it contains the raw data of the sensors, and pipline outputs:

aria/ — egocentric RGB + SLAM trajectory + hand tracking (Project Aria)
egomc/ — IMU-suit motion capture (BVH + raw)
exomc/ — optional marker-based MoCap, used for ground-truth evaluation
scene/ — 3D scanner point cloud and textured mesh
retarget/ — motion retargeted to robot embodiments (URDF/XML)

Citation

BibTeX:

@misc{brandes2026egohtregocentric4ddemonstrations,
      title={EgoHTR: Egocentric 4D Demonstrations of Human Terrain Traversal}, 
      author={Alex Brandes and Haig Conti Georges Sajelian and Manthan Patel and Dominik Hollidt and Chenhao Li and Matthias Heyrman and Oliver Hausdoerfer and Manuel Kaufmann and Xi Wang and Jonas Frey and Angela P. Schoellig and Christian Holz and Marc Pollefeys and Marco Hutter},
      year={2026},
      eprint={2607.13472},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2607.13472}, 
}

APA:

Brandes, A., Sajelian, H. C. G., Patel, M., Hollidt, D., Li, C., Heyrman, M., Hausdoerfer, O., Kaufmann, M., Wang, X., Frey, J., Schoellig, A. P., Holz, C., Pollefeys, M., & Hutter, M. (2026). EgoHTR: Egocentric 4D demonstrations of human terrain traversal. In Conference on Robot Learning (CoRL).

Contact

If you have any questions please open an issue on the Github repo or contact Alex Brandes

Downloads last month: 68

Total file size:

719 GB

Paper for leggedrobotics/egohtr

EgoHTR: Egocentric 4D Demonstrations of Human Terrain Traversal

Paper • 2607.13472 • Published Jul 15