SharedAssembly: A Data Collection Approach via Shared Tele-Assembly

TL;DR

The paper tackles the data bottleneck for contact-rich robotic assembly by proposing SharedAssembly, a bilateral teleoperation framework with shared autonomy that partitions control among leader and follower to improve success and data collection. It introduces a hierarchical autonomy scheme where the leader handles coarse to medium guidance and orientation, while the follower fine-tunes orientation during guided insertion using force-domain knowledge and a wiggle-based feedforward strategy. Experimental results across six tight-clearance tasks and diverse operator expertise show superior success rates and efficiency, achieving around a $97.0\%$ success rate on sub-millimeter tasks, with strong data-collection potential. This approach enables scalable acquisition of assembly data and lays groundwork for training foundation models tailored to contact-rich robotic manipulation.

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.

Figures (6)

• Figure 1: Illustration of the proposed approach for data collection in contact-rich assembly by shared tele-assembly.
• Figure 2: Structure of basic bilateral teleoperation.
• Figure 3: Experimental setup for data collection based on shared tele-assembly
• Figure 4: Task execution time. The darker-color columns with interquartile range indicators represent the execution time during the guided insertion phase; The lighter-color columns correspond to the total execution time, encompassing both the position guiding phase and the guided insertion phase.
• Figure 5: Overall success rate for all tasks across participants with different backgrounds.