Towards Automatic Assembly of Small Screws: Failure Detection and Stage Classification

Xianyi Cheng
Master's Thesis, Tech. Report, CMU-RI-TR-19-69, August, 2019

Download Publication

Copyright notice: This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. These works may not be reposted without the explicit permission of the copyright holder.

Abstract

Hundreds of billions of small screws are assembled in consumer electronics industry every year, yet reliable screwdriving automation remains one of the most challenging tasks.

Barriers that prevent further adoption of robotic threaded fastening systems are system cost and technical challenges, especially those pertinent to small screws. Smaller screws require tighter tolerance and higher alignment accuracy, which increases the system cost and failure rate. A solution is to close the loop. An affordable intelligent screwdriving system that can support online stage classification and failure prediction is the first step to automating the assembly of small screws. However, existing failure prediction techniques are simple and they perform poorly. In addition, most solutions are essentially data-driven, thereby requiring lots of training data and laborious labeling. Moreover, they are not robust against varying environment conditions and suffer from generalization issues. To this end, we propose a stage and result prediction framework that combines knowledge-based process models with a hidden Markov model. The novelty of this work is the incorporation of operation-invariant characteristics such as screwdriving mechanics and a stage transition graph, enabling our system to generalize across a range of experimental settings and largely reduce the requirements on data and labeling. In our experiments, a system trained on M1.4×4 screws adapts with very little non-labeled data to three other screw sizes (M1.2×3, M2.5×5, and M1.4×4) with varying tightening current, motor velocity, insertion force, and tightening force.

We also discuss the role of sensor reduction and compliance in screwdriving automation. To that end, we present our preliminary work on sensor reduction. We use statistical methods to select proper sensors and produce affordable intelligent screwdrivers that can be deployed in industry. Our preliminary experiments also show that compliance allows screws to be inserted with larger misalignment errors.


@mastersthesis{Cheng-2019-117142,
author = {Xianyi Cheng},
title = {Towards Automatic Assembly of Small Screws: Failure Detection and Stage Classification},
year = {2019},
month = {August},
school = {},
address = {Pittsburgh, PA},
number = {CMU-RI-TR-19-69},
keywords = {robotic screwdriving, failure detection, stage classification},
} 2019-08-12T16:47:42-04:00