SharedAssembly: A Data Collection Approach via Shared Tele-Assembly

2025-03-15

AI Summary

Existing robotic assembly datasets suffer from a severe scarcity of contact-intensive task samples, hindering the training of general-purpose robot models. To address this, we propose a shared-autonomy-enabled bilateral teleoperation framework that integrates real-time force feedback, sub-millimeter motion control, and human-robot collaborative state estimation to enable high-precision, scalable assembly data collection. Our approach is the first to incorporate shared autonomy into bilateral teleoperation, substantially lowering the operational barrier: novice and intermediate users achieve expert-level assembly success rates (97.0%). Compared to conventional teleoperation, our method overcomes fine-grained assembly performance bottlenecks, significantly improving data collection efficiency and cross-task generalization across diverse sub-millimeter assembly tasks. This yields high-quality, large-scale, contact-rich training data essential for advancing general-purpose robotic models.

Technology Category

• Intelligent Robots: Manipulation
• Humans and AI: Interaction Techniques and Devices
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Assembly is a fundamental skill for robots in both modern manufacturing and service robotics. Existing datasets aim to address the data bottleneck in training general-purpose robot models, falling short of capturing contact-rich assembly tasks. To bridge this gap, we introduce SharedAssembly, a novel bilateral teleoperation approach with shared autonomy for scalable assembly execution and data collection. User studies demonstrate that the proposed approach enhances both success rates and efficiency, achieving a 97.0% success rate across various sub-millimeter-level assembly tasks. Notably, novice and intermediate users achieve performance comparable to experts using baseline teleoperation methods, significantly enhancing large-scale data collection.

Problem

• Addresses data bottleneck in robot assembly training
• Captures contact-rich assembly tasks effectively
• Enhances success rates and efficiency in assembly

Innovation

• Bilateral teleoperation with shared autonomy
• Scalable assembly execution and data collection
• High success rate in sub-millimeter tasks