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

TL;DR

ARCH introduces a hierarchical learning framework combining imitation and reinforcement learning to enable precise, long-horizon robotic assembly with minimal demonstration data, demonstrating superior generalization and efficiency.

Contribution

The paper presents ARCH, a hierarchical modular approach integrating primitive skills and high-level policies for contact-rich assembly tasks, reducing data requirements and improving generalization.

Findings

ARCH outperforms baseline methods in success rate.

ARCH generalizes to unseen objects.

ARCH demonstrates high data efficiency.

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 policies. 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. More details are available at: https://long-horizon-assembly.github.io.