Academic Publications

Academic Publications#

DOI Year#

This page contains bibliographic information for academic papers and competition projects related to EmbodiChain. Entries are ordered by year (newest first).

Publications#

2026#

RoboSynChallenge: Mastering Real-World Dexterity via Generalizing Synthesized Manipulation Skills#

Authors: RoboSynChallenge Organizing Team

Venue: NeurIPS 2026 Competition

Links: Website | Report | Code

@misc{robosynchallenge2026,
  title = {RoboSynChallenge: Mastering Real-World Dexterity via Generalizing Synthesized Manipulation Skills},
  author = {{RoboSynChallenge Organizing Team}},
  howpublished = {NeurIPS 2026 Competition},
  year = {2026},
  url = {https://robosyn-bench.net/}
}

Backpropagating Through Simulation: Analytic Policy Gradients for Sample and Learning Efficient Differentiable Continuous Control#

Authors: Yueci Deng

Venue: arXiv preprint arXiv:2606.21525, 2026

Links: Paper | Code

EmbodiChain integration: See the APG algorithm, differentiable trainer, and training CLI documentation. The Newton-backed differentiable route currently provides a kinematics gradient bridge.

Examples:

  • PointMass APG training: Training configuration and differentiable environment. Run the configuration with:

    embodichain train-rl --config extensions/embodichain_tasks/configs/tasks/classic_control/point_mass/agents/apg.yaml

  • Franka FR3 differentiable kinematics: Reference environment, demonstrating the Warp-tape / PyTorch-autograd bridge through forward kinematics and reward kernels.

@article{deng2026backpropagating,
  title = {Backpropagating Through Simulation: Analytic Policy Gradients for Sample and Learning Efficient Differentiable Continuous Control},
  author = {Deng, Yueci},
  journal = {arXiv preprint arXiv:2606.21525},
  year = {2026},
  doi = {10.48550/arXiv.2606.21525},
  url = {https://arxiv.org/abs/2606.21525}
}

From Reaction to Anticipation: Proactive Failure Recovery through Agentic Task Graph for Robotic Manipulation#

Authors: Sheng Xu, Ruixing Jin, Huayi Zhou, Bo Yue, Guanren Qiao, Yueci Deng, Yunxin Tai, Kui Jia, Guiliang Liu

Venue: Robotics: Science and Systems (RSS), 2026

@inproceedings{xu2026agentchord,
  title = {From Reaction to Anticipation: Proactive Failure Recovery through Agentic Task Graph for Robotic Manipulation},
  author = {Xu, Sheng and Jin, Ruixing and Zhou, Huayi and Yue, Bo and Qiao, Guanren and Deng, Yueci and Tai, Yunxin and Jia, Kui and Liu, Guiliang},
  booktitle = {Robotics: Science and Systems (RSS)},
  year = {2026}
}

Sim2Real VLA: Zero-Shot Generalization of Synthesized Skills to Realistic Manipulation#

Authors: Runyi Zhao, Sheng Xu, Ruixing Jin, Yueci Deng, Yunxin Tai, Kui Jia, Guiliang Liu

Venue: The Fourteenth International Conference on Learning Representations (ICLR), 2026

@inproceedings{zhao2026sim2real,
  title={Sim2real vla: Zero-shot generalization of synthesized skills to realistic manipulation},
  author={Zhao, Runyi and Xu, Sheng and Jin, Ruixing and Deng, Yueci and Tai, Yunxin and Jia, Kui and Liu, Guiliang},
  booktitle={The Fourteenth International Conference on Learning Representations},
  year={2026}
}

2025#

DexScale: Automating Data Scaling for Sim2Real Generalizable Robot Control#

Authors: Guiliang Liu, Yueci Deng, Runyi Zhao, Huayi Zhou, Jian Chen, Jietao Chen, Ruiyan Xu, Yunxin Tai, Kui Jia

Venue: Forty-Second International Conference on Machine Learning (ICML), 2025

@inproceedings{liu2025dexscale,
  title={DexScale: automating data scaling for sim2real generalizable robot control},
  author={Liu, Guiliang and Deng, Yueci and Zhao, Runyi and Zhou, Huayi and Chen, Jian and Chen, Jietao and Xu, Ruiyan and Tai, Yunxin and Jia, Kui},
  booktitle={Forty-second international conference on machine learning},
  year={2025}
}

Adding a New Paper#

To add a new publication:

  1. Add a new section under the appropriate year heading

  2. Include the paper title, authors, venue, and BibTeX entry

  3. Keep entries ordered by year (newest first)

Core Framework Citations#

The following citations are kept in the main README.md as they are considered core framework references:

  • EmbodiChain - The framework itself

  • GS-World - The underlying generative simulation paradigm

Contents