Manual Assembly Research Articles

Automated visual inspection and defect detection of large-scale silicon strip sensors

For the Phase-II Upgrade of the ATLAS Detector, the Inner Detector will be replaced with the Inner Tracker (ITk), consisting of a pixel and a strip tracker. The 17,888 silicon strip detector modules comprising the ITk strip tracker will be assembled from silicon strip sensors and flexes with readout chips in a manual assembly process performed at 20 module assembly sites in a complex distribution chain, which requires quality control steps to be performed after each distribution and assembly step. Sensor quality control requires a visual inspection of the full sensor area (about 100 cm2) of each sensor to detect and log any defects (e.g. scratches, breakdown areas or chipped corners) or contamination. Since manual surveys of full sensor areas for several thousand sensors are both time-consuming and prone to errors, alternative methods were investigated to automate the process and improve its reliability. This paper presents a setup developed to take high-resolution images of full silicon strip sensors with high repeatability quickly and an algorithm developed for the automated detection of defects, built using functions and filters from popular open-source visual processing packages OpenCV and Scikit-image. Methods were developed both for small-scale high-resolution images and full-size sensor images with lower resolution — both are presented here.

Physical and materials aspects of photonic crystals for microwaves and millimetre waves

Abstract Experimental and numerical results on photonic crystals are presented for the frequency range 26–60 GHz. The material used is alumina where two techniques have been applied for fabricating the photonic crystals: manual assembly of alumina rods and rapid prototyping. The observed positions of the fundamental and higher photonic band gaps are in excellent agreement with the calculated results. A new type of defect in the 3D woodpile structure, is created by inserting interstitial rods. As a new 2D structure a square parquet lattice is investigated. The concept of a negative index of refraction is adressed including a model calculation and an experimental demonstration by the transmission through a slab of a 2D photonic crystal.

Genetic algorithm-based task allocation in multiple modes of human–robot collaboration systems with two cobots

In this paper, we introduce a human–robot collaboration (HRC) mold assembly cell to cope with small-volume mold production and reduce the risk of musculoskeletal disorders (MSDs) on a human worker during manual mold assembly operation. Besides, the wide variety of types and weights of the mold components motivated us to design an HRC system that consists of two robots. Therefore, we propose two collaboration modes for HRC systems using two robots and develop a task-allocation model to demonstrate the application of these collaboration modes in the mold assembly. The task-allocation model assigns a task based on the task characteristics and capability of agents in the collaboration cell. First, we decompose the assembly operation into functional actions to analyze the characteristics of tasks. Then, we obtain the agent assignment preference based on task characteristics and capability of agents using the analytic network process. Finally, we apply the genetic algorithm in the final task allocation to minimize assembly time, use of a less capable agent, and ergonomic risk. This paper contributes to expanding the HRC system with two robots in the mold assembly to allow the execution of a greater diversity of tasks and improve the assembly time and MSD risk level for the human worker.

Automation and Expansion of EMMA Assembly for Fast-Tracking Mammalian System Engineering.

With applications from functional genomics to the production of therapeutic biologics, libraries of mammalian expression vectors have become a cornerstone of modern biological investigation and engineering. Multiple modular vector platforms facilitate the rapid design and assembly of vectors. However, such systems approach a technical bottleneck when a library of bespoke vectors is required. Utilizing the flexibility and robustness of the Extensible Mammalian Modular Assembly (EMMA) toolkit, we present an automated workflow for the library-scale design, assembly, and verification of mammalian expression vectors. Vector design is simplified using our EMMA computer-aided design tool (EMMA-CAD), while the precision and speed of acoustic droplet ejection technology are applied in vector assembly. Our pipeline facilitates significant reductions in both reagent usage and researcher hands-on time compared with manual assembly, as shown by system Q-metrics. To demonstrate automated EMMA performance, we compiled a library of 48 distinct plasmid vectors encoding either CRISPR interference or activation modalities. Characterization of the workflow parameters shows that high assembly efficiency is maintained across vectors of various sizes and design complexities. Our system also performs strongly compared with manual assembly efficiency benchmarks. Alongside our automated pipeline, we present a straightforward strategy for integrating gRNA and Cas modules into the EMMA platform, enabling the design and manufacture of valuable genome editing resources.

Interference-adjusted power learning curve model with forgetting

Researchers in production and operations management have studied the effect of worker learning and forgetting on system performance for decades. It remains an active research topic. Those studies have assumed that production interruptions (or production breaks) cause forgetting, which deteriorates performance. Research on human working memory provides enough evidence that continuous forgetting, precisely cognitive interference, results from overloading the memory with information. Despite the evidence, few studies have incorporated it into learning curve models. This paper presents an enhanced version of the power learning curve that accounts for a variable degree of interference when moving from a production cycle to the next. It adopts the concept of memory trace decay to measure the residual (interference-adjusted), not the nominal (maximum) cumulative experience. We test the developed model against learning data from manual assembly and inspection tasks, with varying numbers of repetitions and breaks. We also test three alternative power-form learning and forgetting curve models from the literature. The results show that the interference-adjusted model fits the data very well. The proposed learning and forgetting model and its individualized cumulative metrics can help identify struggling workers early and release precocious learners earlier than expected. As such, the model gives insights for managers on the occurrence of interference to enable individual learning support.

An Approach to Analyze Human-caused Work Errors

Due to socio-demographic and technological changes, companies face new challenges to achieve an efficient and competitive production. Increasing cost and time pressure combined with an aging workforce with declining physical and mental performance requirements, as well as the increasing shortage of skilled workers, lead to reduced productivity. Therefore, the reduction of human-caused errors is becoming more important. This paper analyzes the connection between human-caused work errors, mental and physical strain, and workplace ergonomics. Four variables are used to analyze the interaction between ergonomics and work errors within the framework of a study in manual assembly. The study sample consists of 21 employees from a manual truck assembly at six different workstations. Models with different combinations of variables are developed in a multiple linear regression framework. Regression model (RM) 1 predicts the variance of the criterion work error with the predictors NASA-RTLX, Borg, and EAWS. It has high predictive power (adjusted R² = 0.746) but is not significant.

Towards accurate failure prediction in manual assembly by establishing a System Dynamics model

The objective of this paper is the optimization of failure management in manual assembly. An advanced System Dynamics model is applied to indicate the interactions between production and quality processes. To quantify the model, the determination of the probability of failure occurrence is considered a necessary prerequisite. In manual assembly, the failure occurrence is directly related to workers' failure. Therefore, this paper analyses existing methods for calculating the human failure probability and evaluates them based on their suitability for the application in manual assembly. Subsequently, the best-evaluated method is further developed to fulfil industrial requirements. A mathematical model is developed according to the specific requirements of manual assembly, which enables the determination of the probability of failure occurrence.

Models to evaluate the performance of high-mix low-volume manual or semi-automatic assembly lines

To address mass customisation demand trends, assembly line flexibility and productivity are critical. Industry 4.0 technologies could support assembly operations to this end. However, clear implementation methodologies are still lacking. This article presents two models for evaluating the most relevant Key Performance Indicators (KPIs) of manual or semi-automatic assembly lines, allowing to maximise the return of investment of any digital technology addition. MATLAB® was used to implement a parametric model, and FlexSim® was employed to build a discrete event simulation model. The models were validated using data from two industrial study cases from a global white goods manufacturer.

A low-cost AR training system for manual assembly operations

This research work proposes an AR training system adapted to industry, designed by considering key challenges identified during a long-term case study conducted in a boiler-manufacturing factory. The proposed system relies on lowcost visual assets (i.e., text, image, video, and predefined auxiliary content) and requires solely a head-mounted display (HMD) device (i.e., Hololens 2) for both authoring and training. We evaluate our proposal in a real-world use case by conducting a field study and two field experiments, involving 5 assembly workstations and 30 participants divided into 2 groups: (i) low-cost group (G-LA) and (ii) computeraided design (CAD)-based group (G-CAD). The most significant findings are as follows. The error rate of 2.2% reported by G-LA during the first assembly cycle (WEC) suggests that low-cost visual assets are sufficient for effectively delivering manual assembly expertise via AR to novice workers. Our comparative evaluation shows that CAD-based AR instructions lead to faster assembly (-7%, -18% and -24% over 3 assembly cycles) but persuade lower user attentiveness, eventually leading to higher error rates (+38% during the WEC). The overall decrease of the instructions reading time by 47% and by 35% in the 2nd and 3rd assembly cycles, respectively, suggest that participants become less dependent on the AR work instructions rapidly. By considering these findings, we question the worthiness of authoring CAD-based AR work instructions in similar industrial use cases.

Analyzing the potential of a time-of-flight depth sensor for assembly assistance

Products are becoming more complex and more varied with technological progress. In these circumstances, the assembly process is more prone to errors while training new workers. Adding an assistive system for monitoring the work to give feedback could help to detect errors immediately and to reduce the propagation of errors. Affordable depth sensors that use the time-of-flight technology could be utilized to monitor the assembly process in real-time and to receive immediate feedback. This work evaluates the potential of a depth sensor for assistance in manual assembly. Because these sensors have problems that are difficult to generalize due to external factors such as material properties influencing the measurement, different problems and implications of using these depth sensors for the task were highlighted with a series of experiments with an Azure Kinect, a time-of-flight depth sensor. Rules and restrictions were derived from the results with the goal of improving the use of depth sensors for assistive systems in a manufacturing process. From the experiments and preceding experiences, a couple of rules and restrictions were derived for using depth sensors in that scenario. The results show that although the approach works in general, problems occur in certain conditions, especially due to optical properties of materials for near-infrared wavelengths, such as reflectivity and transmittance, because of the technology of the depth sensor.

Knowledge extract and ontology construction method of assembly process text

The actual product assembly process mainly relies on manual assembly by workers, and the personal experience of workers is difficult to effectively reuse. Ontology as a knowledge management and expression tool is gradually applied in the field of assembly. However, the manual construction of the ontology is time-consuming and labor-intensive, and the automatic construction of the ontology requires a large number of corpora for training, both of which are difficult to obtain a good assembly case ontology. This paper proposes a method in which automatically extracts relevant knowledge from case assembly process files to generates case database and integrates ontology framework of assembly domain to construct ontology. It shows that the accuracy can be guaranteed on the basis of the rapid construction of case ontology. The feasibility of this method is proved by a practical case.

Proving the Applicability of Assembly Complexity Measures for Process Time Prediction of Customer-specific Production

The production of customized goods is becoming more and more important for industrial companies. The large number of variants resulting from this, up to batch size 1 production, requires a high degree of flexibility of the production systems and their scheduling. To meet these requirements, manual production processes are frequently still used. This is especially applicable to the area of assembly. In contrast to automated systems, the processes in manual assembly are more difficult to schedule. Today, methods for generating predetermined times are used in order to better calculate and schedule processes. However, for the assembly of a product that has not yet been produced, there are neither predetermined times nor historical data from production runs that have already been completed.This paper describes an approach to optimize the production planning of manual assembly systems for customized products by calculating more realistic lead times for new product variants on the basis of historical product data already acquired. Based on objective data acquisition, the existing data sets are classified in order to be able to derive conclusions on new product variants. The classification is based on the measure of assembly complexity. The model for calculating the lead times is validated using data from a real production system of the Chair of Production Systems.

Human-centred assembly and disassembly systems: a survey on technologies, ergonomic, productivity and optimisation

Despite the increasing use of automation in the current industry 4.0 context, manual assembly and disassembly tasks are still common and in some situations even unavoidable. The interaction between humans and other elements of the Assembly and Disassembly Systems (ADS) has been discussed in the scientific literature with the dual objective of optimising human well-being and system performance. Human Factor (HF) related studies focused on new technologies such as motion tracking systems. These technologies enable the collection of data to improve understanding the work environment and its impact on employee well-being and productivity. In the literature, different metrics were suggested to measure the ergonomic and productivity scores. These measurements influence the global system performance and may allow its optimisation. In this paper, we provide a review of the literature on human-centred ADS. We mainly focus on technologies (used to capture human motion), metrics (used to assess human ergonomic risks and productivity), and operational research models (used to optimise the performance system considering economic and cycle time objectives). Future directions are discussed in the perspectives of this paper.

Manual Collection of Data on Disruptions: Determinants to Increase the Intention to Use

Manufacturing companies operate software for the manual collection of process data on the shop floor. A volatile utilization of such software by operators leads to insufficient data quality. The affected companies have difficulties to understand and improve the performance of their assembly processes. We design software to capture data on production disruptions for a manual assembly of complex products. This research investigates the operator's intention to use this software in a single case study. Thus, we conduct software tests and interviews with 51 potential users deploying the technology acceptance model 2 questionnaire. We find that the determinants perceived usefulness and job relevance are predictors for the intention to use software for manual data collection in the investigated context.

Extensible Worker Assistance (EWA): Presenting a Comprehensive Framework for Context-Aware Assistance in Manual Assembly

To cope with the complexification of manual assembly, new assistance methods are developed continuously. However, those hardware-dependent methods are not deployed context-aware. Hence, workers are not supported situationally and new methods have to be implemented at great expense on a heterogeneous system landscape, evoking an inappropriate maintenance effort. As known from the plant engineering, standardized encapsulation of specific methods provides a solution to integrate heterogeneous applications into one generic system. Therefore, besides the propose of a novel Extensible Worker Assistance (EWA) framework, the underlying novel concepts of so-called Assistance Model Units (AMUs) is utilized as a standardized way to abstract from specific implementations and thus, enable the integration of various assistance methods into one generic system. Furthermore, the applicability of the EWA framework with its underlying core concepts is shown by a use case-specific implementation within the bus assembly. Hence, a first step towards the provision of an optimal worker assistance tailored to the individual needs is done, by the presentation of the EWA framework with the ability to integrate different assistance methods, devices and consider various contextual knowledge within the context-aware assistance selection. Future work has to be done to further develop and investigate the single components of the comprehensive framework.

Recording Data on Production Disruptions: Usability and Data Quality

This research presents an application to capture production disruptions, e.g. missing parts or machine breakdowns, in the manual assembly of complex products and systems (CoPS). We validate the usability of this application by measuring time performance, potential resulting data accuracy, and user experience. The presented results stem from a single case study. This research conducts 51 usability tests. We show that the presented application has the potential to achieve a level of data accuracy above 97 % and a level of user experience rated as “good” by the user experience questionnaire. The application is now in operational use at a manufacturer of CoPS.

Methodology to integrate ergonomics information in contextualized digital work instructions

Musculoskeletal disorders (MSDs), often caused by working in non-ergonomic circumstances, are the most important category of work-related diseases. Different methods exist for examining ergonomics through a variety of postural evaluation techniques (e.g. Rapid Entire Body Assessment (REBA)), which are mostly performed through manual survey-based observations. Consequently, measuring ergonomics of operators is time-consuming and happens only sporadically, often after problems occur. Furthermore, as an ergonomist must assign a score to the performed activity, the ergonomics evaluation is subjective. These problems can be resolved by continuous and automated ergonomic load monitoring methods that directly provide feedback to operators. In order to monitor the operator without interfering the task at hand, a vision-based ergonomics monitoring system is developed. Here, ergonomic features (e.g. joint angles) are estimated based on multiple video streams, which are used to calculate an objective ergonomic score using a standard ergonomics evaluation technique. Although cognitive operator support systems (i.e. digital work instructions) for manual assembly environments are widely used, they do typically not include ergonomics information. The goal of the methodology presented in this work is to a) formulate or update digital work instructions, b) provide the operator feedback on its ergonomic soundness via work instructions and c) generate ergonomic risk level reports in an automated way based on the outcomes of the vision-based ergonomics evaluator. Making the ergonomic guidelines context-aware ensures that interventions occur at the right time. The methodology is integrated in a framework for context-aware work instructions and is validated via a proof of concept, based on an actual industrial case. Results show that context-awareness enhances user acceptance and ergonomics of the operator.

Using RFID and Industrial IoT in Manual Assembly for Data Collection and Processing in Learning Factories

Using RFID and Industrial IoT in Manual Assembly for Data Collection and Processing in Learning Factories

Harmonizing ergonomics and economics of assembly lines using collaborative robots and exoskeletons

In the design phase of assembly lines, the harmonization of the frequently conflicting ergonomic and economic objectives is particularly important to enable both humane and efficient assembly. Recent technological developments such as collaborative robots and exoskeletons can support human workers in their assembly tasks and have the potential to enable both ergonomic enhancements and cost reductions. However, ergonomic factors in assembly line balancing are still underrepresented in literature. Exoskeletons have not yet been considered in assembly line balancing. We develop a mixed-integer programming model for balancing assembly lines under consideration of ergonomic and economic objectives and the availability of novel technologies. We validate our model using test instances from literature and find collaborative robots and exoskeletons to be ergonomically and economically viable options for partial automation of manual assembly. We present Pareto-optimal frontiers for the visual inspection of the optimization results and we quantify the potential reductions in costs and biomechanical load to support decision-makers in evaluating the technologies based on their preferences. Our approach is capable to identify nonsupported nondominated solutions, which allows for a stepwise reduction of biomechanical load with only a minor increase in costs. In this article, for the first time, the trade-offs between ergonomic and economic objectives in assembly line balancing are discussed and the potential to harmonize the conflicting objectives using collaborative robots and exoskeletons is evaluated.

Position/Force Visual-Sensing-Based Robotic Sheet-Like Peg-in-Hole Assembly

This article deals with a sheet-like peg-in-hole robotic assembly task which is usually done in manual operation. Thin-walled deformable objects will introduce some special uncertainties in robotic peg-in-hole assembly, which may make this kind of robotic assembly difficult. To deal with this issue, three four-phase selective compliance articulated robot arm (SCARA)-based peg-in-hole assembly strategies which imitate manual assembly with a hybrid switching force/position control scheme are proposed. Both position/force visual sensing methods are proposed based on a single-lens imaging system in an eye-to-hand configuration, respectively. Based on the developed visual sensing techniques, an iterative learning control (ILC) scheme is designed to approximate the synchronous position control between a SCARA robot and a two-degree of freedom (DoF) hybrid compliant gripper. The experiments are implemented on our built-up SCARA-based robotic peg-in-hole (peg: electrode ear, hole: plastic shell) assembly platform. The results indicate that the visual force sensing nonlinearity is about 3% at the full range, and the proposed robotic assembly strategies are suited to this kind of peg-in-hole assembly from industrial scenarios.