Assembly Stethoscope: Detecting Assembly Errors through Frequency Sweeping-A Feasibility Study
Year: 2025
Type: conference-paper
Abstract: Manual assembly, such as assembling flat-packed furniture, is often error-prone. Hidden mistakes, such as poorly tightened bolts or missing joints, may remain unnoticed until the product fails during use. Backtracking the source of these mistakes and fixing the underlying causes requires technical understanding of the assembly and the joints used. To simplify this, we explore the feasibility of an assembly stethoscope, combining hardware and software tools, to diagnose assembly errors. It works by transmitting an acoustic frequency sweeping, namely a chirp signal, through the assembled structure and capturing the signal with a surface microphone. Using signal processing and a classification model, the assembly stethoscope identifies errors like poorly tightened or missing fasteners (e.g., bolts), allowing users to verify quality and address issues promptly. In this paper, we evaluate the system under different scenarios, including multiple chair legs, and generalization to unseen users and achieve a classification accuracy of 74.5% ~ 88.7%. Further, we identify the key challenges and explore potential solutions to build a robust real world system.
Authors: Bo Liu, Yixuan Gao, Yin Li, Rajalakshmi Nandakumar, Thijs Roumen
Institution: Cornell Tech
Language: English
Subfield: Mechanical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W7084042519