ARCH: Hierarchical Hybrid Learning for Long-Horizon Contact-Rich Robotic Assembly

Jiankai Sun, Aidan Curtis, Yang You, Yan Xu, Michael Koehle, Qianzhong Chen, Suning Huang, Leonidas Guibas, Sachin Chitta, Mac Schwager, Hui Li
Proceedings of The 9th Conference on Robot Learning, PMLR 305:2628-2642, 2025.

Abstract

Generalizable long-horizon robotic assembly requires reasoning at multiple levels of abstraction. While end-to-end imitation learning (IL) is a promising approach, it typically requires large amounts of expert demonstration data and often struggles to achieve the high precision demanded by assembly tasks. Reinforcement learning (RL) approaches, on the other hand, have shown some success in high-precision assembly, but suffer from sample inefficiency, which limits their effectiveness in long-horizon tasks. To address these challenges, we propose a hierarchical modular approach, named Adaptive Robotic Compositional Hierarchy (ARCH), which enables long-horizon, high-precision robotic assembly in contact-rich settings. ARCH employs a hierarchical planning framework, including a low-level primitive library of parameterized skills and a high-level policy. The low-level primitive library includes essential skills for assembly tasks, such as grasping and inserting. These primitives consist of both RL and model-based controllers. The high-level policy, learned via IL from a handful of demonstrations, without the need for teleoperation, selects the appropriate primitive skills and instantiates them with input parameters. We extensively evaluate our approach in simulation and on a real robotic manipulation platform. We show that ARCH generalizes well to unseen objects and outperforms baseline methods in terms of success rate and data efficiency.

Cite this Paper

BibTeX
@InProceedings{pmlr-v305-sun25b, title = {ARCH: Hierarchical Hybrid Learning for Long-Horizon Contact-Rich Robotic Assembly}, author = {Sun, Jiankai and Curtis, Aidan and You, Yang and Xu, Yan and Koehle, Michael and Chen, Qianzhong and Huang, Suning and Guibas, Leonidas and Chitta, Sachin and Schwager, Mac and Li, Hui}, booktitle = {Proceedings of The 9th Conference on Robot Learning}, pages = {2628--2642}, year = {2025}, editor = {Lim, Joseph and Song, Shuran and Park, Hae-Won}, volume = {305}, series = {Proceedings of Machine Learning Research}, month = {27--30 Sep}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v305/main/assets/sun25b/sun25b.pdf}, url = {https://proceedings.mlr.press/v305/sun25b.html}, abstract = {Generalizable long-horizon robotic assembly requires reasoning at multiple levels of abstraction. While end-to-end imitation learning (IL) is a promising approach, it typically requires large amounts of expert demonstration data and often struggles to achieve the high precision demanded by assembly tasks. Reinforcement learning (RL) approaches, on the other hand, have shown some success in high-precision assembly, but suffer from sample inefficiency, which limits their effectiveness in long-horizon tasks. To address these challenges, we propose a hierarchical modular approach, named Adaptive Robotic Compositional Hierarchy (ARCH), which enables long-horizon, high-precision robotic assembly in contact-rich settings. ARCH employs a hierarchical planning framework, including a low-level primitive library of parameterized skills and a high-level policy. The low-level primitive library includes essential skills for assembly tasks, such as grasping and inserting. These primitives consist of both RL and model-based controllers. The high-level policy, learned via IL from a handful of demonstrations, without the need for teleoperation, selects the appropriate primitive skills and instantiates them with input parameters. We extensively evaluate our approach in simulation and on a real robotic manipulation platform. We show that ARCH generalizes well to unseen objects and outperforms baseline methods in terms of success rate and data efficiency.} }
Endnote
%0 Conference Paper %T ARCH: Hierarchical Hybrid Learning for Long-Horizon Contact-Rich Robotic Assembly %A Jiankai Sun %A Aidan Curtis %A Yang You %A Yan Xu %A Michael Koehle %A Qianzhong Chen %A Suning Huang %A Leonidas Guibas %A Sachin Chitta %A Mac Schwager %A Hui Li %B Proceedings of The 9th Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2025 %E Joseph Lim %E Shuran Song %E Hae-Won Park %F pmlr-v305-sun25b %I PMLR %P 2628--2642 %U https://proceedings.mlr.press/v305/sun25b.html %V 305 %X Generalizable long-horizon robotic assembly requires reasoning at multiple levels of abstraction. While end-to-end imitation learning (IL) is a promising approach, it typically requires large amounts of expert demonstration data and often struggles to achieve the high precision demanded by assembly tasks. Reinforcement learning (RL) approaches, on the other hand, have shown some success in high-precision assembly, but suffer from sample inefficiency, which limits their effectiveness in long-horizon tasks. To address these challenges, we propose a hierarchical modular approach, named Adaptive Robotic Compositional Hierarchy (ARCH), which enables long-horizon, high-precision robotic assembly in contact-rich settings. ARCH employs a hierarchical planning framework, including a low-level primitive library of parameterized skills and a high-level policy. The low-level primitive library includes essential skills for assembly tasks, such as grasping and inserting. These primitives consist of both RL and model-based controllers. The high-level policy, learned via IL from a handful of demonstrations, without the need for teleoperation, selects the appropriate primitive skills and instantiates them with input parameters. We extensively evaluate our approach in simulation and on a real robotic manipulation platform. We show that ARCH generalizes well to unseen objects and outperforms baseline methods in terms of success rate and data efficiency.
APA
Sun, J., Curtis, A., You, Y., Xu, Y., Koehle, M., Chen, Q., Huang, S., Guibas, L., Chitta, S., Schwager, M. & Li, H.. (2025). ARCH: Hierarchical Hybrid Learning for Long-Horizon Contact-Rich Robotic Assembly. Proceedings of The 9th Conference on Robot Learning, in Proceedings of Machine Learning Research 305:2628-2642 Available from https://proceedings.mlr.press/v305/sun25b.html.

Related Material