Whole Body Control using Rolling Contact Joints

This project consisted in the development and evaluation of our custom Whole Body Controller (WBC)including Rolling Contact Joint (RCJ) constraints. This also involved the development of our open-sourced Robot Planning and Control (RPC) (Bang et al., 2025) framework. This controller was evaluated in the Draco3 humanoid robot designed and built by Apptronik.

Our WBC leverages implicit hierarchies to handle tasks and enforces the RCJ via hard constraints in a Quadratic Program. This work also involved some re-design considerations at the mechanical level. In particular, the cable tensioning mechanism at the hip joint was improved to withstand longer hours of experiments and the knee motors were re-designed to allow enough power to properly carry the robot’s weight. The details are provided in (Bang et al., 2023). Lastly, our architecture, dubbed RPC, includes several visualization and debugging tools along with multiple integrated physics engines (MuJoCo and PyBullet) to evaluate the controller prior to experiments. This also allows for behaviors to be tele-operated and used for imitation learning (Seo et al., 2023).

Some of the experiments performed with our controller on Draco3 are shown below:

Experiments of Draco3 balancing under disturbance rejection, moving sideways, squatting, balancing on one foot, and taking a step in place.

Demo of our RPC (Bang et al., 2025) framework.

Experiments showing our WBC balancing Draco3 while being tele-operated to perform several different tasks (e.g., while cooking).

References

2025

SII

RPC: A modular framework for robot planning, control, and deployment

Seung Hyeon Bang, Carlos Gonzalez, Gabriel Moore, and 4 more authors

In IEEE/SICE International Symposium on System Integration (SII), 2025

DOI Bib

@inproceedings{Bang2025RPC,
  author = {Bang, Seung Hyeon and Gonzalez, Carlos and Moore, Gabriel and Kang, Dong Ho and Seo, Mingyo and Gupta, Ryan and Sentis, Luis},
  title = {{RPC}: {A} modular framework for robot planning, control, and deployment},
  booktitle = {IEEE/SICE International Symposium on System Integration (SII)},
  year = {2025},
  pages = {1142-1148},
  doi = {10.1109/SII59315.2025.10871130},
}

2023

Fr. Rob. AI

Control and evaluation of a humanoid robot with rolling contact joints on its lower body

Seung Hyeon Bang, Carlos Gonzalez, Junhyeok Ahn, and 2 more authors

Frontiers in Robotics and AI, 2023

DOI Bib

@article{Bang2023,
  author = {Bang, Seung Hyeon and Gonzalez, Carlos and Ahn, Junhyeok and Paine, Nicholas and Sentis, Luis},
  title = {Control and evaluation of a humanoid robot with rolling contact joints on its lower body},
  journal = {Frontiers in Robotics and AI},
  volume = {10},
  year = {2023},
  doi = {10.3389/frobt.2023.1164660},
}

Humanoids

Deep imitation learning for humanoid loco-manipulation through human teleoperation

Mingyo Seo, Steve Han, Kyutae Sim, and 4 more authors

In IEEE-RAS 22nd International Conference on Humanoid Robots (Humanoids), 2023

DOI Bib

@inproceedings{Seo2023Humanoids,
  author = {Seo, Mingyo and Han, Steve and Sim, Kyutae and Bang, Seung Hyeon and Gonzalez, Carlos and Sentis, Luis and Zhu, Yuke},
  title = {Deep imitation learning for humanoid loco-manipulation through human teleoperation},
  booktitle = {IEEE-RAS 22nd International Conference on Humanoid Robots (Humanoids)},
  year = {2023},
  doi = {10.1109/Humanoids57100.2023.10375203},
}