High-Mix Low-Volume Research Articles

Flexible and robust detection for assembly automation with YOLOv5: a case study on HMLV manufacturing line

AbstractAutomating assembly processes in High-Mix, Low Volume (HMLV) manufacturing remains challenging, especially for Small and Medium-sized Enterprises (SMEs). Consequently, many companies still rely on a significant amount of manual operations with an overall low degree of automation. The emergence of artificial intelligence-based algorithms offers potential solutions, enabling assembly automation compatible with multiple products and maintaining overall production flexibility. This paper investigates the application of the YOLO (You Only Look Once) object detection algorithm in an HMLV production line within an SME. The performance of the algorithm was tested for different cases, namely, (a) on different products having similar product features, (b) on completely new products, and (c) under different lighting conditions. The algorithm achieved precision and recall greater than 98$$\%$$ % and mAP50:95 greater than 97$$\%$$ % .

Escalation System for Handling and Analyzing Production Disturbances

In a manufacturing environment the effectiveness of internal problem solving is a key success factor in the service level performance offered to the customers. Broken internal processes need to be fixed immediately or at least in a very short time, in order to fulfill the committed delivery dates. There are several methodologies applied for internal problem solving used by different companies. This article presents a solution worked out for a plant operating in a high-mix low-volume (HMLV) production environment prone to both internal and external disruptions and disturbances. Principles, architecture and information flow in our digitized disruption handling – so-called escalation – system will be shortly discussed. Lessons learned in a six-year’s period of using the system will also be summarized. The recorded data confirm that with the introduction of the escalation system the capability of the plant to adapt to changing circumstances and disruptions greatly improved.

An experimental investigation of Lean Six Sigma philosophies in a high-mix low-volume manufacturing environment.

This article experimentally examines methods for implementing the philosophies of Lean Six Sigma (LSS) in a High-Mix Low-Volume (HMLV) manufacturing environment. HMLV environments present unique challenges to LSS paradigms because of the need for extraordinary operational flexibility and customer responsiveness. The subject HMLV manufacturer for this experimentation manufactures (among 8500 others) an example component for which 3 machines work independently to perform the necessary operations to manufacture this component. The experiment that is the subject of this research seeks to adapt LSS philosophies to develop treatments to improve the performance of the manufacturing of this component. These LSS-inspired treatments included 1) using cellular manufacturing methods, and the 3 machines as a single work cell to manufacture the component, and 2) using a single multipurpose machine to perform all operations required to manufacture the component. The results of this experiment demonstrate that the cellular manufacturing method was the most effective to reduce costs, to standardize operations at a process level, and to increase throughput. The single machine processing method improved production rates and on-time delivery relative to the baseline, but greatly increased lead time, thereby increasing total cost per part. These results highlight the importance of critically assessing the application of LSS within HMLV environments compared to the Low-Mix High-Volume (LMHV) environments where LSS is traditionally successful. HMLV manufacturers and researchers can use these findings to identify the most effective methods for their specific needs and to design interventions that will improve system-level manufacturing performance in high mix environments.

Knowledge-Intensive Diagnostics Using Case-Based Reasoning and Synthetic Case Generation

Due to commercial pressures, North-American printed circuit-board assembly manufacturers have had to reposition themselves in the more difficult market segment of lower volume, higher complexity products, also called High-Mix, Low-Volume (HMLV). The high per-unit costs of these products justify substantial diagnosis and repair efforts when defects are detected in production. Although automated diagnostics is desirable, the low production volumes impose severe limits on available data. We propose a novel approach based on knowledge modeling and casebased reasoning for automated diagnosis of assembled printed circuit boards in an HMLV production environment. Our hybrid approach can overcome the knowledge-acquisition bottleneck even in a data-poor environment. The proposed approach does not require contributions from product designers or experts and is designed to operate using only information available during manufacturing. Our test results show that our diagnostic system can detect, locate and classify all single faults and multiple faults affecting up to three neighboring nodes with a better success rate than the reference commercial tool. Moreover, case base data, including board layout information and user feedback from previous repairs, is used to feed a recommender system to provide repair suggestions. A production and repair data simulator is described and used to evaluate the recommender system’s effectiveness.

A Review of the High-Mix, Low-Volume Manufacturing Industry

The high-mix, low-volume (HMLV) industry has seen growth in the need for product customisation with research to increase manufacturers’ flexibility for the variation in market demands. This paper reviews 152 documents from 2000 to October 2022, discussing work related to HMLV production. From an industrial perspective, this paper analyses the industries with HMLV, revealing production sectors and research areas, categorising the developed work, type of validation, and applications. The results show that most work is not industry-specific, with production planning as the central aspect of the research. While other parts of the production processes and value chain received less attention, the semiconductor and electronics industries are the two most researched with substantial validation, leaving gaps in other industries. Earlier work primarily focused on the theoretical development of production planning; however, the development of Industry 4.0 technologies advocates decision support systems for reactive production planning. This period sees the rise in robotics and automation, with improved robotics capability and human—robot collaboration. Assembly assistance systems were developed for manual production to aid operators in managing the variety of information. This paper serves as a reference for the HMLV manufacturing industry in a structured manner while identifying potential for future research in this field.

Hyper-heuristic for flexible job shop scheduling problem with stochastic job arrivals

Flexible job shop is prevalent in high-mix, low-volume (HMLV) production. Real-world HMLV production environments are subjected to fluctuating demands that cause jobs to arrive stochastically. This research presents a simulated annealing-based hyper-heuristic (SA-HH) for solving a flexible job shop scheduling problem with stochastic job arrivals. Two variants of SA-HH that involve a heuristic scheme with and without problem state features are introduced. Experimental results reveal that in terms of the average makespan, SA-HH outperforms a benchmark heuristic on all 10 benchmark instances.

Physics Informed Synthetic Image Generation for Deep Learning-Based Detection of Wrinkles and Folds

Abstract Deep learning-based image segmentation methods have showcased tremendous potential in defect detection applications for several manufacturing processes. Currently, majority of deep learning research for defect detection focuses on manufacturing processes where the defects have well-defined features and there is tremendous amount of image data available to learn such a data-dense model. This makes deep learning unsuitable for defect detection in high-mix low volume manufacturing applications where data are scarce and the features of defects are not well defined due to the nature of the process. Recently, there has been an increased impetus towards automation of high-performance manufacturing processes such as composite prepreg layup. Composite prepreg layup is high-mix low volume in nature and involves manipulation of a sheet-like material. In this work, we propose a deep learning framework to detect wrinkle-like defects during the composite prepreg layup process. Our work focuses on three main technological contributions: (1) generation of physics aware photo-realistic synthetic images with the combination of a thin-shell finite element-based sheet simulation and advanced graphics techniques for texture generation, (2) an open-source annotated dataset of 10,000 synthetic images and 1000 real process images of carbon fiber sheets with wrinkle-like defects, and (3) an efficient two-stage methodology for training the deep learning network on this hybrid dataset. Our method can achieve a mean average precision (mAP) of 0.98 on actual production data for detecting defects.

Improving Delivery Performance in High-Mix Low-Volume Manufacturing by Model-Based and Data-Driven Methods

In a high-mix and low-volume (HMLV) manufacturing environment where demand fluctuation is the rule rather than the exception, daily production management in face of conflicting key performance indicators such as high delivery precision, short lead time, and maximal resource utilization is a most challenging task. This situation may even be hampered by unreliable supplier performance. This paper presents a generic decision support workflow, which first identifies the most critical external and internal factors which have a serious impact on delivery performance. Next, it suggests a method which combines traditional manufacturing system simulation with advanced machine learning techniques to support the improved daily routine lot-sizing and production scheduling activities in a HMLV company. Argumentation is motivated and illustrated by a detailed industrial case study.

Impact of Resource Allocation Policies on Lead Time in a High-Mix Low-Volume Production System

Impact of Resource Allocation Policies on Lead Time in a High-Mix Low-Volume Production System

Challenges for High-Mix Low-Volume Printing of Graduation Books

Challenges for High-Mix Low-Volume Printing of Graduation Books

Using Artificial Intelligence to Facilitate Assembly Automation in High-Mix Low-Volume Production Scenario

Artificial Intelligence (AI) based algorithms are being used increasingly to support industrial robots in the automation of assembly processes. The objective of this work is to detect bores of different geometry, appearance, and inner structure for automating a high-mix low-volume assembly line. Two most widely used AI algorithms namely, Single Shot Detection (SSD) and You Only Look Once (YOLO) have been used to perform the bore detection process. The results obtained by these algorithms have been compared with the conventional detection algorithm (Gradient filter) using the standard metrics used for evaluating the performance of the detection algorithms. The obtained results demonstrate the efficiency and robustness of AI-based algorithms for the detection as they exhibit better performance than the conventional detection method.

Generic Architecture for the Automatic Parametrization of Production Machine Assembly Programs

In a high-mix low volume production environment, time to market is a key factor. However, one bottleneck lies in the often times manual parametrization of machines for new or modified designs. A truly flexible manufacturing environment therefore requires a continuous data flow from the design stages to the shop floor. This paper presents a concept for the automated parameterization of machines at a large automotive plant. Therefore, this paper initially discusses the results of a stakeholder analysis. The stakeholders comprise of different departments related to the product design and manufacturing processes. The requirements resulting from the interviews conducted with the stakeholders are grouped and ordered by priority. Secondly a general architecture for the control interface is presented. It includes multiple submodules, which model the continuous data flow between the departments and the production systems. The first main step of the data flow is the transformation of the information which is presented in various styles depending on the source departments to structured and standardized data. Thereafter machine parameters are generated automatically by a submodule using the structured input data and inference rules. Finally, the architecture supports the automatic transfer of the machine readable output data to the assembly line. To test the architecture a prototype comprising of more than 100 robots in a live production environment is implemented. It allows for a continuous data flow from the design and productions planning department to the robots. This enables a flexible process control, which up to today has supported the fast roll out of more than 100 new product variants. In contrast to the conventional manual setup of the machines for operation, the prototype was able to show that the monetary and time expenditure could be reduced by 95 percent.

Data driven automatic parameter inference for robotic assembly programs

In a high-mix low volume production environment, time to market is a key factor. However, one bottleneck lies in the often times manual parametrization of machines for new or modified designs. A truly flexible manufacturing environment therefore requires a continuous data flow from the design stages to the shop floor. This paper presents a concept for the automated parameterization of machines at a large automotive plant. It describes the interface requirements of the stakeholders involved as well as the overall system architecture. Finally, the paper presents first results of a prototype tested in serial production, spanning more than 100 machines.

Real-Time Decision-Support System for High-Mix Low-Volume Production Scheduling in Industry 4.0

Numerous organizations are striving to maximize the profit of their businesses by the effective implementation of competitive advantages including cost reduction, quick delivery, and unique high-quality products. Effective production-scheduling techniques are methods that many firms use to attain these competitive advantages. Implementing scheduling techniques in high-mix low-volume (HMLV) manufacturing industries, especially in Industry 4.0 environments, remains a challenge, as the properties of both parts and processes are dynamically changing. As a reaction to these challenges in HMLV Industry 4.0 manufacturing, a newly advanced and effective real-time production-scheduling decision-support system model was developed. The developed model was implemented with the use of robotic process automation (RPA), and it comprises a hybrid of different advanced scheduling techniques obtained as the result of analytical-hierarchy-process (AHP) analysis. The aim of this research was to develop a method to minimize the total production process time (total make span) by considering the results of risk analysis of HMLV manufacturing in Industry 4.0 environments. The new method is the combination of multi-broker (MB) optimization and a genetics algorithm (GA) that uses general key process indicators (KPIs) that are easy to measure in any kind of production. The new MB–GA method is compatible with industry 4.0 environments, so it is easy to implement. Furthermore, MB–GA deals with potential risk during production, so it can provide more accurate results. On the basis of survey results, 16% of the asked companies could easily use the new scheduling method, and 43.2% of the companies could use it after a little modification of production.

Lean in High-Mix/Low-Volume industry: a systematic literature review

The aim of this paper is to explore current state-of-the-art on lean in High-Mix/Low-Volume (HMLV) industry. For this purpose, authors propose a systematic literature review of 110 articles published in 66 peer-reviewed international journals that explicitly discuss lean in HMLV context. The purpose is to investigate research that is available, in order to analyse theoretical aspects of lean considerations in HMLV context, identify the level of lean maturity, the scope of lean, the way lean is implemented, and tools and practices that are commonly used. The results of the analysis show that current research is mainly focussed on practice, with little attention to theory development. The maturity of lean implementation is rather low, as articles mostly address obvious waste reduction, while more advanced concepts such as variability management and buffering remain under-researched. Studies are often operationally and internally oriented, with shop-floor improvements being the area of interest, while embedding lean in strategy is not sufficiently addressed. Lean implementation frameworks presented in the literature are usually tool-based, addressing a narrow set of simplified cases that do not take fully into account complexities of HMLV industry. Lastly, a rather narrow set of tools is used, with tools often being adapted without proper consideration of relevance of a certain tool for HMLV industry. As publications reviewing lean in HMLV context are scarce, this study presents valuable insight and serves as an excellent resource for research on the subject, since it not only presents the existing knowledge in a structured way, but also identifies possible directions for future research.

A sampling-based approach for managing lot release in time constraint tunnels in semiconductor manufacturing

For the sake of product yield and quality considerations, time constraints (TCs) are imposed between process operations in various multi-product manufacturing systems. Often spanning a number of operations, time constraints tend to follow each other in close succession and overlap, forming thus time constraint tunnels (TCTs). The regulation problem of releasing lots in these time constraint tunnels is particularly challenging in semiconductor manufacturing systems, because of re-entrant flows, machine heterogeneity and high mix low volume (HM-LV) production configurations, which are typical in many wafer fabrication facilities. In such an evolving and time-varying context, this paper proposes a sequential sampling-based approach to estimate the probability that, prior to its release, a lot leaves a given time constraint tunnel on time. The proposed approach proves to be competitive in various respects by (i) taking into account industry specific features, (ii) being industrially tractable, and (iii) being sensitive and responsive to the current manufacturing system. Based on real-life instances, numerical experiments highlight the computational effectiveness and the industrial soundness of the proposed problem modelling together with the solution approach.

Computer Vision-enabled Human-Cyber-Physical Workstations Collaboration for Reconfigurable Assembly System

Manufacturing flexibility is strategically important to firms for improving competition in the fast-changing market. This research inspired by the assembly cells system, which significantly appears rapid response, high flexibility and efficiency on the high-mix, low-volume (HMLV) environment. However, due to the non-repetitive operations caused by various products, high dependency on workers’ skills and inevitable exceptions and disruptions in manual operation, the complexity and uncertainty within the assembly cell will cause the system disturbance propagation and amplification. This paper presents a reconfigurable assembly system consisted of multiple intelligent assembly workstations (iWorkstation), named W-RAS. Based on a proposed human-cyber-physical system (HCPS) framework, the innovated iWorkstation can assist workers with differences in abilities and characteristics. Computer vision technology is applied to monitor and analyze the operation process and status for work measurement and ergonomic assessment. This real-time information visibility and traceability can support feedback-based optimal management and enhance coordination among the iWorkstations for enhancing the adaptability and reconfigurability of W-RAS. Finally, an operational mechanism is proposed for multiple iWorkstations collaboration and a numerical example shows the benefits in shortening the lead time of various orders.

Ultra-Compact Device-Manufacturing-System “Minimal Fab” Integrating Wafer and Packaging Process for High-Mix Low-Volume Productions and Its Packaging Applications

Ultra-Compact Device-Manufacturing-System “Minimal Fab” Integrating Wafer and Packaging Process for High-Mix Low-Volume Productions and Its Packaging Applications

Application of SMED Methodology and Scheduling in High-Mix Low Volume Production Model to Reduce Setup Time: A Case of S Company

Nowadays rapid changes in consumer demand determines the production model. The demand for “single and large quantity” products has gradually turned into a “high-mix and low volume” product demand. The main problem of high-mix low volume model is significantly increasing frequency of set up, causing an increase in set-up time for production. For set-up reduction strategies, Shingo (1985) proposed single minute exchange of die (SMED) methodology. The SMED methodology makes it possible to reduce setup (changeover) operations time within 10 minutes. Therefore, in this study take a cable processing company as a case for improvement. The potential of faster setup cannot be completely achieved with SMED method. So, in this paper, it also provides an optimized scheduling reference indicator for the molding machine that decision makers to determine the sequencing rule of scheduling by each criterion through the system to achieve maximum benefit for company.

Complexity-based task allocation in human-robot collaborative assembly

PurposeOver the past years, collaborative robots have been introduced as a new generation of industrial robotics working alongside humans to share the workload. These robots have the potential to enable human–robot collaboration (HRC) for flexible automation. However, the deployment of these robots in industrial environments, particularly in assembly, still comprises several challenges, of which one is skills-based tasks distribution between humans and robots. With ever-decreasing product life cycles and high-mix low volume production, the skills-based task distribution is to become a frequent activity. This paper aims to present a methodology for tasks distribution between human and robot in assembly work by complexity-based tasks classification.Design/methodology/approachThe assessment method of assembly tasks is based on the physical features of the components and associated task description. The attributes that can influence assembly complexity for automation are presented. Physical experimentation with a collaborative robot and work with several industrial cases helped to formulate the presented method.FindingsThe method will differentiate the tasks with higher complexity of handling, mounting, human safety and part feeding from low-complexity tasks, thereby simplifying collaborative automation in HRC scenario. Such structured method for tasks distribution in HRC can significantly reduce deployment and changeover times.Originality/valueAssembly attributes affecting HRC automation are identified. The methodology is presented for evaluating tasks for assigning to the robot and creating a work–load balance forming a human–robot work team. Finally, an assessment tool for simplified industrial deployment.