Query-Centric Diffusion Policy for Generalizable Robotic Assembly

2025-09-23

AI Summary

To address the misalignment between high-level planning and low-level control in robotic assembly tasks involving complex part interactions and high-noise environments, this paper proposes a query-centric hierarchical diffusion policy framework. The framework explicitly bridges symbolic reasoning and physical execution via a multimodal query mechanism that jointly encodes object geometry, contact points, and structured skill semantics. A point-cloud-driven diffusion policy model, integrated with query-based attention, enables robust key-component identification and goal-directed control. Evaluated on FurnitureBench in both simulation and real-world settings, the method achieves over 50% higher success rates on long-horizon tasks—such as insertion and screwing—compared to unstructured-query baselines. It demonstrates significantly improved robustness to sensory noise and enhanced cross-task generalization capability.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Agentic search
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

The robotic assembly task poses a key challenge in building generalist robots due to the intrinsic complexity of part interactions and the sensitivity to noise perturbations in contact-rich settings. The assembly agent is typically designed in a hierarchical manner: high-level multi-part reasoning and low-level precise control. However, implementing such a hierarchical policy is challenging in practice due to the mismatch between high-level skill queries and low-level execution. To address this, we propose the Query-centric Diffusion Policy (QDP), a hierarchical framework that bridges high-level planning and low-level control by utilizing queries comprising objects, contact points, and skill information. QDP introduces a query-centric mechanism that identifies task-relevant components and uses them to guide low-level policies, leveraging point cloud observations to improve the policy's robustness. We conduct comprehensive experiments on the FurnitureBench in both simulation and real-world settings, demonstrating improved performance in skill precision and long-horizon success rate. In the challenging insertion and screwing tasks, QDP improves the skill-wise success rate by over 50% compared to baselines without structured queries.

Problem

• Addresses hierarchical policy mismatch between high-level reasoning and low-level control
• Improves robotic assembly robustness in contact-rich, noise-sensitive environments
• Enhances skill precision and long-horizon success rates for complex assembly tasks

Innovation

• Hierarchical framework bridging high-level planning and low-level control
• Query-centric mechanism identifying task-relevant components for policy guidance
• Leverages point cloud observations to improve policy robustness