The study focuses on three key digital technologies: the Internet of Things, artificial intelligence, and blockchain. Each of these technologies serves a distinct function within the supply chain ecosystem. IoT devices primarily enable the sensing and real-time collection of operational data, providing a continuous stream of information about physical processes. Blockchain technology strengthens traceability, verification, and information integrity across supply chain transactions, ensuring that records are secure and verifiable. Artificial intelligence complements these data-generating technologies by processing large and heterogeneous datasets, identifying patterns, and generating predictions that support optimization and decision-making.

Central to the proposed model is the concept of supply chain transparency. The researchers posit that digital technology adoption enhances sustainability performance indirectly by improving the availability, accessibility, traceability, and verifiability of supply chain information. This transparency acts as a mediating mechanism, allowing firms to monitor environmental impacts, verify ethical work conditions, and optimize resource utilization. By making information more visible and accountable, digital tools create the informational conditions necessary for effective sustainability-oriented decisions. The study suggests that the value of digitalization lies not just in operational efficiency, but in the ability to reveal and manage sustainability-related data across the entire supply chain.

However, the study highlights that technology alone is insufficient to guarantee improved sustainability outcomes. The presence of a skilled workforce is identified as a critical moderating factor that strengthens the relationship between digital transformation and sustainability performance. Employees must possess the digital, analytical, and managerial skills required to operate these technologies, interpret complex data, and translate insights into operational actions. Without adequate human capital, firms may struggle to leverage their digital investments effectively. The authors emphasize that a socio-technical perspective is essential, as the performance outcomes of digitalization result from the dynamic interplay between technological systems and the people who use them.

To test these hypotheses, the researchers conducted an empirical quantitative study based on survey data collected from 250 manufacturing firms. The sample was selected using a stratified sampling method to ensure diversity across multiple industry sectors and geographic regions. Data were gathered through an electronic questionnaire distributed to senior managers responsible for supply chain, operations, or sustainability functions. The questionnaire assessed the extent of digital technology adoption, organizational capabilities, and sustainability performance metrics. All dimensions were operationalized using 5-point Likert scales, modified from validated scales in prior research to suit the specific context of the study.

The data analysis revealed strong construct validity and high internal consistency across all measured variables. Factor analysis indicated that the scales explained more than 75% of the variance, and Cronbach’s alpha values ranged from 0.895 to 0.933, confirming the reliability of the instruments. Correlation analysis showed significant positive associations among the core research variables, providing initial support for the proposed relationships. Hierarchical multiple regression analyses were then employed to test the direct, mediating, and moderating effects specified in the conceptual framework. These statistical procedures allowed the researchers to isolate the specific contributions of digital adoption, transparency, and workforce skills to sustainability performance.

The results confirmed that digital technology adoption has a significant positive association with environmental sustainability performance. The broader digital technology adoption construct, which encompasses the overall use of Industry 4.0 technologies, reported a standardized coefficient of 0.615. When examining artificial intelligence adoption specifically, the study found a standardized coefficient of 0.545, with an adjusted R-squared of 0.488. These findings indicate that AI constitutes a meaningful component of the digital-sustainability relationship, particularly through its analytical and optimization capabilities. The study distinguishes between the broad infrastructure provided by IoT and blockchain and the specialized analytical power of AI, suggesting that both are necessary for a comprehensive digital transformation strategy.

Furthermore, the analysis provided evidence for the partial mediation effect of supply chain transparency. This finding suggests that while digital technologies offer direct operational benefits, a substantial portion of their value is realized by enabling greater visibility and accountability across the supply chain. The strong moderating effect of a skilled workforce also supported the hypothesis that human capital is a pivotal strategic variable. Firms with a highly skilled workforce are better positioned to exploit technological capabilities and integrate digital information into decision-making. Conversely, organizations with inadequate workforce capabilities may see diminished returns on their technology investments, highlighting the importance of parallel investments in training and recruitment.

The study acknowledges several limitations, primarily stemming from its cross-sectional design. While the survey provides empirical evidence for the proposed relationships, the single-point-in-time data collection limits the ability to establish temporal precedence and definitive causal relationships. The authors suggest that future research should employ longitudinal designs or complementary qualitative approaches to examine how these dynamics evolve over time. Additionally, the focus on manufacturing firms in specific geographic regions may limit the generalizability of the findings to other industries or contexts. Despite these limitations, the study offers valuable insights for managers and policymakers seeking to leverage digital technologies for sustainability objectives.

In conclusion, the research underscores that sustainable supply chain performance is not solely a function of technological adoption but a result of the interaction between digital tools, transparency mechanisms, and human capabilities. Managers are advised to adopt a holistic vision that explicitly targets sustainability improvements and to match their digital investments with parallel investments in human capital. By understanding the mechanisms through which digital transformation drives sustainability, firms can better navigate the challenges of the Industry 4.0 era and achieve long-term value creation while meeting their environmental and social responsibilities.

Subject of Research: Impact of AI and digital transformation on sustainable supply chain performance

Article Title: Artificial intelligence and digital transformation for sustainable supply chain performance

Article References: Belgaroui, R., & Samet, K. (2026). Artificial intelligence and digital transformation for sustainable supply chain performance. Discover Artificial Intelligence, 6(1), Article 1615. https://doi.org/10.1007/s44163-026-02524-1

Image Credits: AI Generated

DOI: 10.1007/s44163-026-02524-1

Keywords: Artificial Intelligence, Supply Chain Management, Sustainability, Digital Transformation, Industry 4.0, Blockchain, Internet of Things, Workforce Skills, Manufacturing, Environmental Performance, Artificial, intelligence

Daniel Coughlin, executive director of CRAFT and a senior research and development staff member at ORNL, framed the center as a response to aging industrial infrastructure. He said the U.S. pulp and paper industry is critical to thousands of rural communities, but much of the existing infrastructure is decades old. According to Coughlin, CRAFT aims to unify, align and accelerate the nation’s leading centers of forest-products innovation, linking industry and the research community through a framework intended to support the growth of U.S. manufacturers. The announcement, made by ORNL on 22 October 2026, describes the center as a collaboration hub for technology commercialization and workforce development, with industry engagement and guidance described as critical to its success.

The initiative’s stated research agenda focuses on solutions that reduce cost, enhance productivity and increase the abundance of U.S. forest products. Named focus areas include next-generation pulping, chemical recovery, dewatering and drying, and the digital transformation of the industry. CRAFT plans to couple advanced materials research with smart manufacturing tools intended to accelerate the digital transformation of the sector. During its first phase, the center will work with industry to shape pilot-scale investments needed to de-risk innovation. To improve the global competitiveness of existing assets, CRAFT intends to provide industry access to automation, modeling and data analysis tools that allow manufacturers to extract process insights aimed at improving efficiency, increasing productivity, reducing downtime and reducing cost.

Gregg Beckham, a senior research fellow and the National Laboratory of the Rockies lead for CRAFT, said the center is eager to engage industry to identify critical bottlenecks and innovation opportunities. He described plans to apply molecular-to-process-scale modeling, AI-enabled discovery and integrated experimental validation to accelerate the translation of new technologies with real-world impact. Art Ragauskas, the University of Tennessee-ORNL Governor’s Chair in Biorefining and chief technology officer for CRAFT, called the center a gateway to innovation for the U.S. pulp and paper industry, saying that CRAFT and its partners will work to de-risk technologies to rapidly transition the most promising solutions from research to full commercial impact. These statements describe the center’s planned approach rather than demonstrated results, since the initiative is at the beginning of its first phase.

Robert Wagner, associate laboratory director for ORNL’s Energy Science and Technology Directorate, said ORNL is positioned to lead CRAFT because of its expertise in advanced manufacturing, materials science and data-driven process optimization, and that partnerships across DOE’s national laboratory system and with universities give industry access to technical capabilities. CRAFT will work through an industry and research guidance team alongside representatives from pulp, paper and packaging manufacturers, downstream customers, equipment suppliers and solution providers, with the stated aim of aligning the center’s work with real-world priorities and providing a pathway to commercialization of new technologies. Coughlin said DOE’s investment in CRAFT reflects the view that the nation’s economic security depends on efficient, competitive manufacturing, and that innovation developed in close partnership with industry is intended to strengthen U.S. manufacturing, help industry adapt to market shifts and secure high-value manufacturing jobs in communities across the country. UT-Battelle manages ORNL for DOE’s Office of Science.

Subject of Research: Forest products manufacturing innovation; national laboratory collaboration for pulp, paper and packaging industry modernization

Article Title: ORNL launches national center to modernize US forest products manufacturing

Article References: ORNL launches national center to modernize US forest products manufacturing. Original publication (Bibliographic details unavailable.)

Image Credits: AI Generated

DOI: Not provided

Keywords: ORNL, launches, national, center, modernize, forest, products, manufacturing, scientific research

The approach offers high deposition rates and a large build envelope, making it attractive for aerospace, automotive and marine applications, while surface finish, dimensional accuracy, anisotropy and fatigue performance remain important limitations. The central finding is that WAAM Ti-6Al-4V cannot be treated as a uniform material. Repeated heating changes the alloy as each new layer is deposited, producing spatial gradients in phase morphology, grain orientation, hardness, residual stress and mechanical behavior.

The alloy contains 6 wt% aluminum and 4 wt% vanadium, which stabilize its alpha and beta phases. Its columnar prior-beta grains often grow epitaxially from the substrate through successive layers, while the alpha laths within them vary with local cooling and reheating. Reported prior-beta grain sizes range from about 0.2 to 4 millimeters.

The review identifies the moving heat source, deposited geometry and later thermal cycles as central mechanisms linking process settings to structure and properties. Three main arc configurations provide different compromises. Gas tungsten arc welding uses a non-consumable tungsten electrode and typically deposits 1 to 2 kilograms per hour.

It is clean and relatively spatter-free, can handle one or two wires, and can be pulsed to alter the thermal profile and reduce porosity. Plasma arc welding constricts the arc inside a nozzle, raising current density and enabling deposition rates of 2 to 4 kilograms per hour. The concentrated arc improves stability and can remove surface oxide through cathode atomization, but greater heat accumulation can increase distortion, grain coarsening and tensile residual stress.

Gas metal arc welding uses consumable wire and reaches 3 to 4 kilograms per hour, or about 9.5 kilograms per hour with a tandem arrangement, but conventional operation is prone to unstable arcs and spatter. Cold metal transfer offers a lower-heat GMAW variant by retracting the wire during short-circuit periods. The review reports that it can reduce heat input by roughly 30 to 50% and dilution from the typical GMAW range of 20 to 30% to below 10%, while producing less spatter.

Its behavior can be adjusted through short-circuit, globular, spray, synergic and pulse-advanced transfer modes. In one comparison, CMT-WAAM Ti-6Al-4V thin walls had a finer, more uniform alpha-lamellar distribution, an average microhardness of 328.9 HV, a tensile strength of 1,045 megapascals and 30.15% elongation compared with a conventional GMA-arc counterpart. These results apply to the tested material and conditions; they do not establish universal performance for every WAAM configuration.

Because Ti-6Al-4V conducts heat at only about 6.7 watts per meter-kelvin, layer-by-layer deposition can create a hot, distorted build with pronounced thermal gradients. The review describes several ways researchers have attempted to dissipate heat, including electric preheating of wire, interlayer dwell periods, forced-air or liquid-carbon-dioxide cooling, progressively reducing heat input as the build rises, fractal-like scanning paths and toolpaths optimized through thermal simulation. A comparison found dwell time more influential than cooling-gas flow under the conditions examined, showing that equipment settings cannot be generalized without considering geometry and build height.

Torch weaving at 1 hertz combined with a low-heat “coldArc” regime also disrupted epitaxial columnar growth and produced a lower-aspect-ratio prior-beta structure and more uniform transformation products. Hybrid processing offers a route to reduce anisotropy, the tendency for strength and ductility to vary with loading direction. Hot forging can create a fully equiaxed microstructure, and one study reported an ultimate tensile strength rising from 961 to 1,137 megapascals and elongation from 8.8% to 15.2%.

In-situ interpass rolling is especially effective because reheating during the next deposition can generate annealing twins and recrystallized grains. Post-process rolling and ultrasonic peening can also introduce compressive residual stress. In a large-scale comparison, hot isostatic pressing achieved full densification and the best overall performance among the treatments examined, while solution treatment and aging improved strength but produced comparatively small gains in fatigue-relevant properties.

These studies also show that strengthening and ductility do not always rise together. Atmospheric control is equally important because hot titanium readily absorbs oxygen, nitrogen and hydrogen. Local argon shielding can restrict oxygen pickup to tens or a few hundred parts per million, although lower regions of large builds may remain exposed as heat accumulates.

A sealed GTAW-WAAM chamber that maintained oxygen below 10 parts per million produced 99.98% near-theoretical density, a refined Widmanstatten microstructure, an ultimate tensile strength of 964 plus or minus 29 megapascals and a dry-sliding wear rate about 40% below conventionally manufactured Ti-6Al-4V. Another controlled study found that increasing oxygen from 1 to 1,000 parts per million raised hardness and tensile strength but reduced elongation from 11.5% to 9.5%. The review notes that direct comparison of local shielding and full enclosure under otherwise identical conditions is still needed.

Fatigue performance remains a demanding test because microscopic defects can override improvements in the bulk microstructure. The review specifically identifies gas pores and surface roughness as important factors in fatigue initiation and crack growth. Tomography of WAAM Ti-6Al-4V made with contaminated wire showed near-surface cracks growing around pores; pore growth accelerated late in life, between 20,000 and 28,000 cycles, equivalent to 65 to 85% of fatigue life.

Samples containing critical pores larger than 100 micrometers had substantially lower fatigue life than nominally defect-free specimens, even though their as-built microstructures and tensile strengths were similar. Clustering of gas porosity also governed fatigue life across process-varied and heat-treated conditions in a large-scale study, reinforcing that microstructural refinement alone cannot guarantee durability. The review’s outlook points toward integrated process design rather than a single universal recipe.

Fractal scanning, pulsed sources, interpass cooling, heat treatment and in-situ deformation are presented as complementary approaches whose effects must be assessed together. Important unanswered questions include how to optimize scanning for residual stress, which grain-refining strategy best balances strength and ductility, and how residual stress affects fatigue-crack growth. Direct evidence linking these variables remains sparse, and the review excludes comprehensive treatment of process monitoring, closed-loop control, digital twins, qualification, certification and high-temperature creep.

Its conclusion is nevertheless clear: WAAM can deliver large Ti-6Al-4V components at high deposition rates, but structural performance emerges from the combined physics of the arc, melt pool, thermal cycling and post-processing, not from deposition speed alone.

Subject of Research: Review of WAAM Ti-6Al-4V process, microstructure, residual stress and mechanical performance

Article Title: Wire arc additive manufacturing of Ti-6Al-4 V alloy: a review

Article References: Mendiratta, M., Chakkravarthy, V., Satish Idury, K. S. N., Singh, N. K., Smirnov, S. A., Tsai, J.-T., Evlashin, S., Chen, K., Zhang, L., Gushchina, M., Klimova-Korsmik, O., & Narayan, R. L. (2026). Wire arc additive manufacturing of Ti-6Al-4 V alloy: a review. Journal of Materials Science: Metallurgy, 1(1), Article 24. https://doi.org/10.1007/s44492-026-00025-x

Image Credits: AI Generated

DOI: 10.1007/s44492-026-00025-x

Keywords: wire arc additive manufacturing, Ti-6Al-4V, directed energy deposition, titanium alloys, microstructure, residual stress, tensile properties, fatigue life, thermal management, Wire, additive, manufacturing

The choice of infrared heating in this study also reflects a deliberate engineering trade-off. Laser-assisted tape placement offers precise, concentrated energy delivery and is well established in aerospace production, but the equipment is expensive and requires careful safety management. Hot-gas torch heating is gentler but less energy-efficient and can be difficult to control across varying layup geometries. An IR lamp sits between these extremes: it delivers radiant energy that can be tuned in intensity, it heats the tow and substrate without physical contact, and the hardware cost is a fraction of that of a laser system. Radiant heating also couples reasonably well with dark carbon fiber surfaces, which absorb infrared radiation efficiently, reducing the amount of energy wasted on the surrounding air. For a material like CF/PC, whose processing temperatures are moderate compared with high-temperature matrices such as PEEK, the IR lamp provides sufficient heat flux without the need for the extreme power densities that laser systems are designed to deliver.

The two parameters isolated in the design of experiments, compaction force and tool temperature, act on consolidation through complementary mechanisms. Compaction force governs the mechanical squeeze flow of the softened polymer at the interface between the newly placed tow and the substrate beneath it. When the matrix reaches a sufficiently low viscosity, pressure from the roller spreads the polymer across surface asperities, displaces entrapped air, and increases the real contact area between layers. Insufficient force leaves interlaminar gaps and voids that act as crack initiation sites under flexural and shear loading. Excessive force, however, can squeeze matrix material out from the tow edges, thin the resin-rich layer at the interface, and in extreme cases damage the fiber architecture or distort the placed tow. Tool temperature, by contrast, controls the thermal boundary condition of the process. A warmer tool reduces the rate at which heat is drawn away from the bond line, extending the time available for polymer diffusion and wetting before the material cools below its processing range. The heat-sink effect documented in prior AFP studies, in which bond quality degrades through the laminate thickness as successive layers cool the growing stack, is directly moderated by tool temperature. A heated tool essentially raises the baseline from which each new layer must be heated, narrowing the thermal gradient that drives heat away from the interface.

The selection of flexural strength and lap-shear strength as the primary response variables is well matched to the failure modes that poor consolidation produces. Flexural testing loads the laminate in a bending configuration that places the outer plies in tension and compression, making the measurement sensitive to interlaminar weaknesses that allow plies to separate or delaminate under the bending-induced shear stresses. Lap-shear testing, in which two overlapping bonded regions are pulled in opposite directions, directly interrogates the quality of the fusion bond between layers. Fractographic examination of the tested specimens by digital microscopy provides the mechanistic link between process parameters and mechanical outcomes: surfaces that fail cohesively through the polymer, leaving polymer residue on both mating faces, indicate that the fusion bond achieved full integrity, whereas clean interfacial separation with exposed fiber surfaces indicates incomplete wetting and bonding. This combination of quantitative mechanical data and qualitative fractography allows the researchers to distinguish between defects that merely reduce stiffness and those that fundamentally compromise the bond.

The demonstration of laying CF/PC tows onto a rotating three-dimensional printed polyetherimide cylindrical mold deserves particular attention, because it addresses one of the persistent practical barriers to AFP adoption. Conventional AFP tooling is typically machined from metal, which is expensive, heavy, and time-consuming to produce for each new geometry. Additively manufactured tooling offers rapid, low-cost production of complex shapes, but it introduces challenges: printed polymers generally tolerate lower surface temperatures than metals, their thermal conductivity is far lower, and their surface finish and dimensional stability under heat can be inferior. Polyetherimide is a high-temperature amorphous thermoplastic with a glass transition temperature well above the processing temperatures needed for polycarbonate, making it a plausible candidate material for a mold that must withstand repeated exposure to IR heating and hot tows. The low thermal conductivity of a polymer mold, which would be a liability in processes requiring rapid heat extraction, may actually be advantageous here, since it reduces the heat-sink effect that pulls thermal energy away from the bond line during placement. Successfully laying tows on such a mold without secondary machining demonstrates a pathway from printed design to consolidated composite part in a single workflow, which is significant for low-volume production, prototyping, and geometrically complex structures such as ducts, fuselage sections, and pressure vessels.

The drone frame application mentioned in the study illustrates the market segment where CF/PC via IR-assisted AFP is most immediately competitive. Unmanned aerial vehicles demand high stiffness-to-weight ratios, moderate mechanical loads, and increasingly, production volumes that exceed what manual layup or autoclave processing can economically deliver. Thermoplastic matrices add further advantages relevant to this sector: they can be welded for assembly and repair, they resist impact damage better than brittle thermoset matrices in many cases, and they can be re-melted at end of life for recycling, an attribute of growing importance as environmental regulation tightens across the automotive and electronics industries where this material was originally targeted by its supplier.

It is worth situating the reported achievement in the broader context of AFP process science. The finding that IR power and robot speed emerged as the most critical parameters from screening is consistent with the fundamental physics of the process: the energy delivered per unit length of tow, which is the ratio of heating power to layup speed, determines whether the matrix reaches the viscosity needed for bonding. Too little energy per unit length leaves the polymer too stiff to flow and wet the substrate; too much energy degrades the matrix through thermal oxidation or chain scission, and can also cause excessive tow sagging and buckling under the compaction roller. The subsequent identification of compaction force and tool temperature as the next most influential parameters reflects the two-stage nature of consolidation: heating first brings the polymer to a processable state, and the combination of pressure and retained heat then determines how completely the polymer flows and bonds before cooling. The comparison of steel and silicone compaction rollers adds a further dimension, since roller compliance affects how pressure is distributed across the tow and how the roller conforms to curved surfaces, which matters greatly for the cylindrical mold demonstration.

The claim that this is the first investigation of an amorphous matrix, specifically CF/PC, using IR-assisted AFP highlights a gap in the literature. Most published AFP parameter studies have concentrated on semi-crystalline aerospace matrices, partly because certification requirements in aerospace have historically favored materials with well-characterized melting behavior and established databases. Extending AFP process knowledge to amorphous, lower-cost matrices broadens the accessible application space toward consumer products, electronics housings, and ground transportation, where cost per kilogram and cycle time weigh more heavily than ultimate temperature capability. The reported demonstration of improved mechanical properties at optimized settings, achieved out-of-autoclave without secondary consolidation, supports the argument that IR-assisted AFP can produce CF/PC laminates of sufficient quality for semi-structural duty, and the combination of a cost-effective heat source, a moderately priced matrix, and additive tooling suggests a manufacturing chain whose economics could favor adoption in industries where autoclave-grade thermoplastic AFP has remained out of reach.

Subject of Research: Infrared-assisted automated fiber placement of carbon fiber-polycarbonate thermoplastic composites

Article Title: Enhanced manufacturing quality of thermoplastic composites through infrared-assisted automated fiber placement

Article References: Venkatesan, C., Ridhwan, M., Zulkifli, F., Gupta, U., & Silva, A. (2025). Enhanced manufacturing quality of thermoplastic composites through infrared-assisted automated fiber placement. Frontiers in Manufacturing Technology, 5, Article 1649798. https://doi.org/10.3389/fmtec.2025.1649798

Image Credits: AI Generated

DOI: 10.3389/fmtec.2025.1649798

Keywords: automated fiber placement, carbon fiber, thermoplastic composites, infrared heating, process optimization, mechanical properties, polycarbonate, in-situ consolidation, flexural strength, lap-shear strength, 3D printed mold, manufacturing

The review, authored by N. Charan S. S. Kowshik of Ocugen India and Pushpendra Singh of Ocugen, lays out an end-to-end framework for establishing robust AAV manufacturing processes, from plasmid production and cell culture through transfection, purification, and fill-finish. The authors contend that process variability, low yields, and poor scalability remain the central constraints on translating AAV’s therapeutic promise into accessible treatments. Their prescription is a data-driven, quality-by-design approach in which every unit operation is characterized, modeled, and controlled before a process is validated and submitted to regulators.

At the heart of conventional AAV production is transient transfection of human embryonic kidney 293 (HEK293) cells. Three plasmids are typically prepared: a cis-acting plasmid carrying the therapeutic gene flanked by inverted terminal repeats, a trans-acting plasmid encoding AAV replication and capsid proteins, and a helper plasmid enabling viral replication in the host cell. These are introduced into cells expanded from master and working cell banks, and after two to three days in the bioreactor the cells are harvested for downstream purification. Each step in this chain introduces variability. Inconsistent plasmid uptake by host cells leads directly to variation in the ratio of empty to full capsids, a critical quality attribute that determines potency, dose consistency, and immunogenicity.

The upstream innovations highlighted in the review attack this variability from several directions. High-density perfusion cultures maintain cells in an optimal growth phase by continuously supplying fresh nutrients, extending the transfection window and enhancing genome replication and packaging. In one reported setup, an N-1 perfusion process for suspension HEK293 cells achieved cell densities of 35 million cells per milliliter with greater than 95 percent viability after seven days, supporting production bioreactors of 2,000 to 5,000 liters. For adherent systems, single-use fixed-bed perfusion bioreactors such as the iCELLis platform have demonstrated eight-fold higher productivity through post-transfection media exchange. Optimized transfection parameters, including plasmid amount and DNA-to-polyethyleneimine ratio, can push unit yields as high as 2 x 10¹² vector genomes per milliliter.

Plasmid system design is evolving just as rapidly. Traditional triple-plasmid transfection, while fast and efficient, often produces inconsistent co-delivery of the required genetic elements. Advanced dual-plasmid platforms that combine the transgene, AAV rep and cap genes, and accessory elements on one plasmid, with adenoviral helper functions on a second, have shown improved productivity and full capsid proportions across multiple serotypes and scales up to 50-liter bioreactors. The AAVone single-plasmid system consolidates all necessary components into one construct, yielding a two- to four-fold increase in vector production, reduced batch variability, and lower DNA impurities, characteristics the authors consider especially suitable for GMP-grade manufacturing. Design-of-experiments optimization of triple-plasmid ratios has nearly doubled genome titers for AAV2 and AAV9 while reducing empty capsid formation, and engineered helper plasmids with deletions in adenoviral E4 and E2a genes have delivered more than two-fold productivity gains.

Plasmid quality itself emerges as a critical starting-material attribute. The supercoiled percentage of plasmid DNA directly influences transfection performance: highly supercoiled plasmids form smaller, more compact complexes with transfection reagents, which are taken up more efficiently during endocytosis. The review notes that regulatory expectations commonly call for a defined lower limit above 80 percent supercoiled DNA in bulk release specifications. Achieving that standard requires careful control of bacterial cultivation, harvesting during logarithmic growth phase, and precise alkaline lysis and neutralization, followed by anion-exchange or monolith-based chromatography to separate supercoiled plasmids from open-circular and linear species. Next-generation transfection reagents, including FectoVIR-AAV, the TransIT-AAViator system with its RevIT enhancer, and modified linear PEI formulations, offer reported productivity gains of two-fold or more over conventional PEI while reducing reagent and plasmid consumption.

Downstream processing confronts a different challenge: separating the genome-containing capsids that constitute the active drug substance from empty and partially filled particles that contribute nothing to efficacy and may trigger unwanted immune responses. Affinity chromatography has advanced considerably with serotype-agnostic peptide ligands that enable broad, high-yield capture across all major AAV serotypes under milder elution conditions. The POROS CaptureSelect AAVX resin, for example, achieves recovery rates of 65 to 80 percent across divergent serotypes while supporting repeated regeneration. Density gradient ultracentrifugation with cesium chloride or iodixanol remains a workhorse for impurity removal, typically requiring two to three cycles for enhanced purity, though its scalability is limited. Continuous-flow ultracentrifugation systems with rotors accommodating up to 50 liters have reported recovery rates exceeding 55 percent in a single run, beginning to address that constraint. Polishing chromatography then exploits the difference in surface charge between full and empty capsids: two-pass anion-exchange strategies, membrane-based ion exchange optimized through design of experiments, and dual salt elution gradients have achieved full capsid enrichment of 70 to 80 percent or higher at production scale.

The review places quality-by-design at the center of the entire manufacturing enterprise. Process characterization begins with risk assessment, using failure mode effects analysis and cause-and-effect matrices to identify potential critical process parameters and evaluate their impact on critical quality attributes such as vector titer, capsid integrity, full-to-empty ratio, and aggregation. Validated scale-down models, built by matching power per unit volume, oxygen transfer coefficients, and shear stress between commercial-scale bioreactors of 50 to 2,000 liters and small-scale stirred tanks, allow multiple experiments to be run efficiently. Statistical tools including multivariate data analysis, principal component analysis, and equivalence testing confirm that the scale-down model faithfully represents the commercial process. Design of experiments then maps the design space, from which proven acceptable ranges and normal operating ranges are defined. Variations within the design space are not regarded as changes requiring supplementary regulatory submissions, and once a parameter exceeds its proven acceptable range during commercial manufacturing, the batch is considered a failure.

Process validation follows the three-stage framework outlined in the US FDA guidance: process design, process qualification, and ongoing process verification. Because AAV products are manufactured in limited batches during development and early commercial phases, the authors emphasize a risk-based, statistically justified approach to determining the number of process performance qualification batches. The overall residual risk depends on product knowledge, process understanding, and manufacturing experience; tightly controlled processes may require fewer batches to produce statistically meaningful data, while highly variable processes demand multiple campaigns. No harmonized industry standard currently exists for justifying batch number, though three consecutive qualification batches remain the common recommendation for processes under control.

Looking ahead, the authors see artificial intelligence and continuous manufacturing as pivotal to the next generation of AAV therapies. AI-driven soft sensing with predictive modeling can estimate critical quality attributes such as viral titer and capsid integrity in real time from readily available process variables, enabling continuous monitoring, early deviation detection, and predictive quality assessment. Mechanistic modeling of empty-full capsid separation using anion-exchange membrane chromatography has already demonstrated that AI-enhanced models can predict recovery yields with remarkable fidelity. Advanced analytical technologies, including cryoelectron microscopy, mass spectrometry-based proteomics, and molecular dynamics simulations, coupled with AI and predictive modeling, are expected to illuminate capsid behavior and host interactions, empowering the engineering of capsids optimized for stability, immune evasion, and targeted tissue delivery. The worldwide AAV vector market is expected to expand at a compound annual growth rate of approximately 14 percent until 2035, and the authors argue that standardized orthogonal analytics, traceable reference materials, and multi-laboratory ring trials to harmonize potency and genome titer assays will be essential to bring manufacturing standards closer to accessible, safer biologics. Automated platforms integrating high-throughput screening with synthetic biology, they conclude, could ultimately permit fully customizable AAV vectors designed for specific diseases or tailored to individual patients, transforming the platform into a remarkably precise and adaptable instrument for the coming era of personalized gene therapy.

Subject of Research: Advancing AAV vector manufacturing: challenges, innovations, and future directions for gene therapy

Article Title: Advancing AAV vector manufacturing: challenges, innovations, and future directions for gene therapy

Article References: Kowshik, N. C. S. S., & Singh, P. (2025). Advancing AAV vector manufacturing: challenges, innovations, and future directions for gene therapy. Frontiers in Molecular Medicine, 5, Article 1709095. https://doi.org/10.3389/fmmed.2025.1709095

Image Credits: AI Generated

DOI: 10.3389/fmmed.2025.1709095

Keywords: Advancing, vector, manufacturing, challenges, innovations, future, directions, gene, therapy, scientific research

The stakes could hardly be higher. More than 90 percent of all manufacturing enterprises in China fall into the small and medium-sized category, which means the digital fate of the entire sector rests on their shoulders. Yet official statistics paint a sobering picture. According to the Analysis Report on Digital Transformation of Small and Medium-sized Enterprises (2021) cited in the study, 79 percent of small and medium-sized enterprises remain in the exploratory stage of digital transformation, 12 percent have reached the practical stage, and only 9 percent have achieved deep application. In other words, the overwhelming majority of these firms have barely begun a journey that the researchers describe as a long-term, systematic project, one in which significant results typically materialize only after three to five years or longer.

Why do so many small manufacturers stall? The study identifies five interlocking barriers. First, leadership often lacks sufficient understanding of digital transformation; many executives are unsure why sensors should be installed on equipment or what industrial big data is actually for, and they prefer projects that deliver quick returns over multi-year restructuring of production methods, management models, and business models. Second, digital talent is desperately scarce. On average, digital-related personnel account for only 20 percent of total human resources in Chinese enterprises, and most of those specialists are concentrated in large firms that can offer generous salary packages that small enterprises simply cannot match. Third, the transformation foundation is weak: only 38 percent of small and medium-sized manufacturing enterprises have adopted identification technologies such as barcodes and RFID for data acquisition, 22 percent have integrated and shared key business systems, and a mere 4.5 percent use big data analysis to guide production. Fourth, transformation funds are insufficient, since small asset scales and chronic financing difficulties leave many firms unable to sustain the high costs of transformation through its middle and later stages. Fifth, and perhaps most culturally entrenched, many small manufacturers habitually go it alone, rarely collaborating or sharing resources with other enterprises, a legacy of workshop origins where inter-firm collaboration had little visible impact on profitability.

This pattern of isolation has produced what the researchers call information islands, where data and knowledge are not shared across departments, let alone across companies. With the rise of new-generation information and communication technologies such as artificial intelligence, big data, cloud computing, and blockchain, the authors contend that relying on going it alone is no longer realistic for sustainable profit growth. The millions of small and medium-sized enterprises that have gone bankrupt in the past three years, they note, are sobering examples of what happens when firms stick to old conventions in a rapidly digitizing economy.

The theoretical engine behind the study combines two classic frameworks from strategic management: the Resource-Based View and Dynamic Capabilities Theory. The Resource-Based View, rooted in Barney’s 1991 work on firm resources and sustained competitive advantage, holds that competitive advantage stems from access to valuable, rare, inimitable, and non-substitutable resources. Dynamic Capabilities Theory, developed by Teece and colleagues, focuses on a firm’s ability to sense, seize, and reconfigure resources in dynamic environments. The sharing economy, the authors argue, directly addresses both lenses. Platforms offering advanced technologies, specialized expertise, and data resources give small firms access to resources they could never possess internally, while real-time market information flowing through these platforms sharpens their sensing capabilities and their flexible models support rapid reconfiguration of operations.

To translate theory into practice, the study proposes three types of sharing economy platforms. The first is a chain-hosting shared manufacturing platform, built on leading enterprises, which improves collaboration efficiency across the upstream and downstream of an industrial chain, enabling sharing from raw material procurement and research and development design through production, logistics, marketing, and after-sales service. The second is an industry-brain-like collaborative sharing platform, constructed on industries within a region that have significant advantages, which leverages platform effects to drive the digital transformation of entire industrial clusters, particularly helping firms with weaker capabilities, while also providing services such as talent training and technological innovation. The third is a shared service cloud platform built on the industrial internet, which combines cloud computing with industrial internet interfaces to provide high-quality, low-cost microservices and computing power in a cloud-based manner, enabling small firms to go to the cloud and use the cloud without massive upfront investment. Complementing these platforms, the study describes the data middle platform, an enterprise-level sharing service hub that integrates data sources, establishes unified data standards and models, provides real-time decision support, promotes business innovation, and ensures data security, thereby breaking down the information barriers that have traditionally separated departments and systems.

The empirical heart of the paper is a case study of Company F, a small enterprise in the elevator industry specializing in sales, installation, and maintenance, which engaged with a shared service platform built by Company L, an industrial internet enterprise recognized by the Ministry of Industry and Information Technology as a specialized, refined, and innovative small giant. Company L’s platform relies on key technologies including the Internet of Things, intelligent perception, cloud intelligence, and 5G edge computing to enable digital collaborative sharing across the entire elevator industrial chain. Through IoT devices such as smart sensors, smart elevator gateways, and AI cameras, enterprise customers can collect elevator status data and maintenance information online. The platform analyzes the collected data and issues early warning notices to determine the possible time and location of elevator failures, arranging personnel to reach the scene in time to avoid breakdowns. A low-code development platform further allows business departments to build applications by dragging and dropping, reducing the application building threshold and shortening project development cycles.

The quantitative results are striking. Analyzing Company F’s financial records from 2015 to 2022 with Microsoft Excel, the researchers found that the mean main business profit margin rose from 32.65 percent before transformation (2015 to 2018) to 45.94 percent after transformation (2019 to 2022), while mean total asset turnover increased from 0.64 to 0.84. Simple linear regression with time as the independent variable confirmed these effects statistically: for profit margin, the regression yielded a significant positive slope of 13.29 with an R-squared of 0.62 and p less than 0.05, supporting the hypothesis that digital transformation via a sharing service platform significantly increases profitability; for total asset turnover, the slope was 0.20 with an R-squared of 0.74 and p less than 0.01, confirming enhanced operational efficiency. The journey was not without turbulence. In 2020, facing the unforeseen circumstances of the COVID-19 pandemic, Company F saw its main business profit margin and net profit growth rate decline by 32.98 percent and 117.85 percent respectively compared with 2019. But the leadership swiftly adjusted strategy, collaborated with Company L to formulate a new digital transformation plan, and maintained rapid growth in main business income and net profit over the following two years, with main business income growth reaching 20.97 percent in 2022.

Qualitative findings reinforced the numbers. Thematic analysis of interview transcripts with Company F’s digitalization head and technical staff, alongside observational notes from site visits, was conducted using NVivo version 12. Initial open coding identified recurring concepts such as real-time monitoring and cost savings from 15 interview segments and 10 observation entries, which were grouped into categories during axial coding and refined into three key themes: enhanced maintenance efficiency, streamlined spare parts management, and increased collaboration. Real-time monitoring appeared in 80 percent of interviews, and staff emphasized the platform’s role in anticipating failures before they happen and reducing warehouse clutter. The researchers concluded that Company L’s sharing service platform mitigated technological and financial barriers by enabling predictive maintenance and lean inventory practices, directly supporting the study’s central proposition that sharing economy platforms facilitate digital transformation in small and medium-sized manufacturing enterprises.

The implications extend well beyond a single elevator firm, and well beyond China. The authors suggest that the chain-hosting model could be adapted in regions such as Southeast Asia or Europe to bolster small firm integration into global supply chains, while the shared service cloud platform’s low-cost digital solutions could empower resource-constrained enterprises in middle-income countries to leapfrog technological barriers. At the same time, the study cautions that significant regional disparities exist within China itself, with coastal provinces often possessing more developed digital infrastructure and a more established culture of technological adoption than inland regions, creating an uneven playing field that demands targeted policy interventions. The researchers also acknowledge limitations: the evaluation timeframe is limited, and future research should employ longitudinal approaches tracking enterprises over five to ten years to understand how the benefits and challenges of sharing-economy-facilitated transformation unfold over time. For now, the message for millions of small manufacturers is clear: the resources they lack may already exist on a platform they have yet to join, and the collaborative sharing of technology, talent, and data may be the most affordable path through the digital gate.

Subject of Research: Digital transformation barriers among Chinese small and medium-sized manufacturing enterprises and how sharing economy platforms overcome them, with statistical evidence from an elevator industry case study.

Article Title: Overcoming digital transformation barriers: Chinese SMMEs in the sharing economy

Article References: Xiaosheng, J., Hua, Z., Jin, Z., AL-Bukhaiti, K., & Wan, A. (2025). Overcoming digital transformation barriers: Chinese SMMEs in the sharing economy. Frontiers in Manufacturing Technology, 5, Article 1495840. https://doi.org/10.3389/fmtec.2025.1495840

Image Credits: AI Generated

DOI: 10.3389/fmtec.2025.1495840

Keywords: digital transformation, sharing economy, SMMEs, manufacturing, China, data sharing, shared service platform, Industry 4.0, elevator industry, regression analysis, Overcoming, digital

The study, published in Frontiers in Manufacturing Technology, addresses a fundamental inefficiency in how production processes are traditionally optimized. Conventional experimental design methodologies, including full and fractional factorial designs, one-factor-at-a-time approaches, the Taguchi Method, and response surface modeling, are static and non-adaptive. They separate experiment design from execution and analysis, meaning that the entire experimental plan is laid out before a single trial is run. If early results suggest a more promising direction, the methodology offers no mechanism to adapt. In an environment where physical experimentation is time-consuming, costly, and resource-intensive, this rigidity represents a significant bottleneck for process optimization across the entire life cycle of a production process, from the design and development of new processes through failure analysis and optimization to adaptation and calibration in series production.

Bayesian Optimization, or BO, offers a fundamentally different approach. The technique, which traces its origins to the work of Močkus in the 1970s, is a sequential decision-making strategy particularly suited for optimizing expensive-to-evaluate black-box functions that lack a closed-form representation, do not provide function derivatives, and only allow for point-wise evaluation. BO consists of two core components: a surrogate model that approximates the input-output behavior of the system being optimized, and an acquisition function that guides the selection of parameter sets to evaluate next. After each experiment, the Bayes theorem is applied to update the surrogate model from a prior to a posterior belief, incorporating the new observation into the model’s understanding of the process. The acquisition function then uses the uncertainty quantification of the updated surrogate to maximize the information gain of each subsequent experiment while balancing the exploration-exploitation trade-off. When BO is applied to sequential experimental design in scientific and engineering experimentation, the resulting methodology is termed Bayesian experimental design, or BED.

Despite its promise, BED has not achieved widespread adoption in production engineering practice, and the reason is configuration complexity. Setting up a BO algorithm for a specific optimization problem requires extensive knowledge of both the algorithm and the process characteristics. The mean and covariance functions of the surrogate model, the choice of acquisition function, and the size of the initial data sampling must all be individually configured, and each of these choices significantly influences overall optimization performance. For Gaussian process surrogate models, the covariance function, also known as the kernel, is of particularly high criticality for model fidelity and the sample path behavior of the experimentation. Previous benchmark studies in materials science and computational fluid dynamics have explored the impact of BED configuration, but typically focusing on only two components in combination. None had systematically examined the combined interdependencies among acquisition function, surrogate model, and initial design in the context of production processes.

To close this gap, the research team designed a benchmark study of remarkable scope. They selected eight engineering-focused synthetic test functions with different characteristics, complexities, and known optimal solutions. Four of these, designated F1 through F4, are purely mathematical problems. The remaining four, AdaptedBranin, Borehole, OTLCircuit, and WingWeight, are physically motivated optimization problems drawn from a virtual library of simulation experiments. Each test function was evaluated in both a noiseless and a noisy variant, with the noisy version introducing homoscedastic random noise at a level of 0.1 relative to the standard deviation of the function’s objective. In the configuration space, the team tested a total of 96 BO configurations obtained by combining eight kernels, four acquisition functions, and three initial sampling sizes in a full-factorial manner. The kernels included RBF, Matern05, Matern15, and Matern25, each investigated both as an isotropic version and as an anisotropic counterpart utilizing automatic relevance detection, or ARD, which implicitly determines the relevance of input parameters. The acquisition functions comprised expected improvement, probability of improvement, upper confidence bound, and Noisy Expected Improvement. Initial sampling sizes were set at small, five trials; medium, 10 trials; and large, 30 trials, all within a fixed total budget of 35 trials. Each configuration was run ten times with different random seeds applied to a pseudo-random Sobol sampling strategy, ensuring that all configurations started with the same initial data and that no configuration benefited from random fluctuations.

For performance evaluation, the team employed the dominated hypervolume metric, a multi-objective measure that integrates both robustness, quantified as the probability of failure across runs, and efficiency, quantified as the running time of successful experiments. Higher hypervolume values indicate superior overall performance, combining high success rates with fast convergence to the known optimum. A run was considered successful if the algorithm found a solution within a tolerance range corresponding to 95 percent optimization of the known optimal solution. The team then applied a nuanced classification approach that considered both performance level and variability across combinations, dividing hypervolume values into non-qualified, undetermined, and qualified areas for each test function. This approach was necessary because simple aggregation techniques, such as computing the median hypervolume across all runs, could distort actual performance when a configuration performs well in some combinations but poorly in others.

The results revealed significant performance differences between individual BO algorithm configurations, and the test functions themselves grouped naturally into three categories of optimization difficulty. Group A, comprising F1, F2, and OTLCircuit, proved relatively simple, with median hypervolume values of approximately 0.7 and interquartile ranges below 0.12, indicating that most BO configurations could achieve similar optimization potential. These functions showed low sensitivity to noise, with noisy results similar to noiseless ones, and F1 and F2 achieved a 100 percent success rate across almost all configurations. Group B, comprising F4 and WingWeight, presented moderate challenges with median hypervolume values around 0.5 and higher variability, though no significant differences between noiseless and noisy cases. Group C, comprising F3, AdaptedBranin, and Borehole, proved the most complex, with lower efficiency levels, significant differences between noiseless and noisy cases, and a pronounced impact of noise on robustness. For AdaptedBranin and Borehole, the median hypervolume dropped from 0.56 and 0.45 in noiseless cases to 0.2 in noisy cases, with probability of failure rising from 0.1 and 0.05 to 0.5 respectively, underscoring that careful configuration selection is essential when dealing with complex and noisy optimization problems.

Among the kernels, the Matern05 kernel was consistently classified as non-qualified across the test functions, indicating poor performance in optimizing the selected problems, and Matern05-ARD performed poorly in all test functions and was never classified as the best option. RBF showed decent performance for simple functions but struggled in more complex and noisy scenarios. RBF-ARD and Matern25-ARD exhibited similar behaviors, performing well on noiseless functions while presenting acceptable performance on noisy ones. Matern15 and Matern15-ARD showed promising performance, classified as qualified for most test functions. The isotropic Matern15 kernel outperformed the anisotropic version in group B, while the anisotropic version performed better in group C. As a general conclusion, group B seemed to be optimized better by isotropic kernels and group C by anisotropic ones, with Matern15 and Matern15-ARD appearing suitable for optimizing a wide range of problems.

Among the acquisition functions, probability of improvement consistently performed worse than all other options, a finding consistent with previous works that explained its poor performance due to its greedy nature. The upper confidence bound performed better but still encountered difficulties in several functions, potentially due to its fixed exploration parameter, which may have prioritized exploitation over exploration. Expected improvement and Noisy Expected Improvement stood out as the best performing options, with NEI showing the best performance across all test functions in both noiseless and noisy cases. Regarding initial sampling sizes, the small initial sampling size clearly outperformed the medium one in groups A and C, with the medium size only representing a better option in the noisy cases of group B. The large initial sampling size proved universally inefficient, with no experiment surpassing a hypervolume value of 0.15 across all test functions, a mathematically inevitable consequence of the minimum running time achievable by large experiments occurring at trial 30 out of a total budget of 35 trials.

From these findings, the team derived actionable guidelines for applying Bayesian experimental design in manufacturing. There is no one-fits-all solution, as different optimization problems place different demands on BO configuration. RBF presents a reasonable choice for simple test functions, while Matern15-ARD is recommended as a reasonable default option for complex optimization problems, with anisotropic kernels advisable for the more complex problems typically encountered in manufacturing. Since noise negatively impacts optimization performance, process and measurement noise should be minimized by precisely calibrating both actuators and sensors of the manufacturing process. Expected improvement and Noisy Expected Improvement are recommended as qualified default acquisition functions, while probability of improvement and the investigated upper confidence bound configuration are not suggested. Randomly generated initial experiments should be kept small, leaving the search for the optimum to a sufficiently exploratory BO algorithm, though already existing datasets should be utilized to initialize the algorithms. Screening and characterization trials prior to optimization are decisive for determining the most critical parameters and associated parameter ranges, keeping the dimensionality of the optimization problem as small but also as influential as possible. The study advocates BED as an adaptive, data-efficient tool for optimizing process parameters, achieving 95 percent precision within a budget of 35 iterations using the best-qualified configurations at various levels of complexity. The team acknowledges that the study did not unravel a clear relationship between the characteristics of optimization problems and the performance of BO configurations, suggesting that current characteristics do not adequately capture the inherent patterns of response to optimization. Further research focuses on the collaboration between BED and domain experts, the integration of expert knowledge into BED, and the extension to multi-objective optimization cases, all aimed at increasing comprehensibility, facilitating user acceptance, and promoting widespread adoption in industry applications.

Subject of Research: Bayesian experimental design in production engineering: a comprehensive performance and robustness study

Article Title: Bayesian experimental design in production engineering: a comprehensive performance and robustness study

Article References: Leyendecker, L., Gonzalez Degetau, A. M., Bata, K., Emonts, J., Schmitz, A., & Schmitt, R. H. (2026). Bayesian experimental design in production engineering: a comprehensive performance and robustness study. Frontiers in Manufacturing Technology, 5, Article 1614335. https://doi.org/10.3389/fmtec.2025.1614335

Image Credits: AI Generated

DOI: 10.3389/fmtec.2025.1614335

Keywords: Bayesian optimization, Bayesian experimental design, production engineering, manufacturing, process optimization, benchmark study, Gaussian processes, Frontiers in Manufacturing Technology, Bayesian, experimental, design, production

The choice of elbow flexion as the control variable deserves closer examination. Elbow angle sits at an intersection of practical concerns that few other postural measures occupy. It responds quickly and predictably to changes in workpiece height and horizontal reach, which are precisely the two degrees of freedom the collaborative robot can manipulate in this framework. A workpiece placed too low drives elbow flexion beyond comfortable ranges; one placed too far away forces shoulder flexion and trunk lean, which also register in the elbow angle as the arm extends. Because skeletal tracking libraries estimate joint centers reliably from ordinary RGB imagery, elbow flexion can be computed continuously without gloves, markers, or electromagnetic trackers that would interfere with fine manual work such as screw-driving. This makes it an unusually well-behaved feedback variable: sensitive to the manipulated parameters, robustly observable, and directly tied to established ergonomic criteria.

It is worth noting the deliberate separation between the control variable and the evaluation metric. Elbow flexion drives the loop, but the study does not claim that optimizing a single joint angle guarantees overall postural health. Instead, the full RULA score, which aggregates upper-arm, lower-arm, wrist, trunk, neck, and leg postures along with muscle use and force considerations, serves as the outcome measure. This design guards against a common failure mode in automated ergonomic systems, where optimizing one observable quantity inadvertently worsens another. If the robot raised a workpiece to relieve elbow flexion but thereby encouraged neck flexion or an elevated shoulder, the RULA comparison between fixed and adaptive conditions would reveal the trade-off. The reported improvements across the composite score suggest that the adaptation strategy moved the whole posture pattern in a favorable direction rather than merely relocating discomfort.

The sensing architecture reflects pragmatic engineering choices that matter for real deployment. Two independent cameras, one facing the operator’s upper body and one mounted above and looking down at the hand, address a genuine occlusion problem in manual assembly. When an operator leans over a workpiece, the torso camera loses sight of the wrist and fingers, yet wrist posture contributes meaningfully to ergonomic risk in precision tasks. Processing each stream separately, without cross-camera registration or triangulation, keeps the pipeline simple and avoids error accumulation from imperfect extrinsic calibration. The trade-off is that the system does not build a unified three-dimensional skeleton; instead, each camera serves its own dedicated analysis. For a task like screw-driving, where upper-body posture and wrist orientation both matter but full-body kinematics are less critical, this division of labor is a reasonable compromise between accuracy, latency, and robustness.

The participant sample of thirty, performing screw-driving under both fixed and adaptive conditions, gives the evaluation a within-subject structure that is well suited to this kind of comparison. Each participant serves as their own control, which reduces the influence of individual differences in baseline posture, task strategy, and anthropometry on the contrast between conditions. Screw-driving is a useful representative task because it combines repetitive upper-limb motion with precision demands, and because workpiece position strongly determines whether the operator works with a neutral arm or a raised, abducted shoulder. The finding that the adaptive condition reduced the influence of participant height on ergonomic outcomes is particularly significant for mixed-workforce environments. Fixed workstations inherently encode an assumed body size; operators far from that assumption absorb the mismatch as postural strain. Continuous adaptation effectively removes much of that hidden penalty.

The psychological measures reported alongside the physical ones add an important dimension to the evidence. Trust and acceptance are not soft add-ons in human-robot collaboration; they determine whether operators actually use adaptive systems as intended or override and avoid them. A robot that moves the workpiece unexpectedly could plausibly unsettle operators, especially during precision tasks where sudden motion near the hands might be perceived as hazardous. The observation that trust and acceptance increased while perceived safety was not compromised suggests that the adaptations were experienced as helpful rather than intrusive. This outcome likely depends on the nature of the adjustments: small, task-relevant repositioning of the workpiece, rather than motion of the robot arm through the operator’s personal space, keeps the interaction within familiar boundaries while still delivering ergonomic benefit.

Placing this work in the context of prior approaches clarifies its contribution. Adjustable workstations, whether manual or motorized, have long been proposed as a remedy for anthropometric variability, but they generally require deliberate operator action or scheduled adjustment. Posture-based robot motion adaptation has been explored, yet much of that literature focuses on planning robot trajectories around human postures rather than reshaping the shared workspace to improve the human posture. Vision-based ergonomic assessment systems demonstrate that camera-derived skeletal data can support RULA-style scoring in real time, but they typically stop at measurement. Optimization frameworks that balance throughput against ergonomic risk operate largely at the planning stage. The present framework closes the gap between these threads by using the same posture estimate both as a diagnostic and as a live feedback signal, converting perception directly into continuous physical adjustment during task execution.

The implications for industrial practice extend beyond the specific screw-driving scenario. Any collaborative task in which the workpiece can be repositioned by a robot, and in which operator posture varies with workpiece location, is a candidate for this control strategy. Assembly of small components, inspection tasks requiring sustained visual attention, and packaging operations all exhibit the same structure: a human performing upper-limb work whose quality depends on where the object sits relative to the body. Because the framework relies on commodity depth cameras operating in RGB mode and an open-source pose estimation library, the sensing cost is modest compared with wearable motion capture or instrumented fixtures. The main requirements are camera placement with adequate sightlines, computational capacity for real-time inference, and a cobot with the payload and workspace envelope to move the workpiece smoothly.

Several limitations and open questions naturally follow from the design. The two-camera architecture leaves gaps in coverage when the operator’s body occludes both views simultaneously, and rapid movements can challenge frame-rate-limited pose estimation. The ergonomic criteria that trigger adaptation are predefined rather than learned, so the system adapts to measured posture but does not personalize thresholds to individual comfort preferences over time. The evaluation focused on a single task type and a relatively short task duration, whereas musculoskeletal risk accumulates over hours and across task variety. Fatigue, which changes both posture and the relationship between posture and discomfort, was not a central variable. Future work could explore longer sessions, multiple task types, learned adaptation policies that incorporate individual history, and richer feedback variables that combine elbow flexion with trunk or neck measures to guide more comprehensive repositioning.

Nevertheless, the study demonstrates a principle with broad relevance: posture can serve as the error signal in a feedback loop whose actuator is the robot itself. This reframes the collaborative robot not merely as a tool-sharing partner but as an active component of the ergonomic infrastructure of the workstation. In traditional ergonomics, the environment is engineered to fit a statistical average of human bodies, and residual mismatch is managed through training, rotation, and breaks. A closed-loop adaptive approach shifts the burden of fit from the operator to the system, continuously negotiating the geometry of the task against the live state of the body. As vision-based pose estimation continues to improve in accuracy and efficiency, and as collaborative robots become more prevalent on factory floors, frameworks of this kind point toward workstations that accommodate human variability not by design assumption but by ongoing measurement and response, making ergonomic support a dynamic property of the collaboration rather than a static feature of the equipment.

Subject of Research: A vision-based closed-loop framework for adaptive ergonomic human-robot collaboration

Article Title: A vision-based closed-loop framework for adaptive ergonomic human-robot collaboration

Article References: Teeravarunyou, S., Vongbunyong, S., Kamolwit, K., Kitirojpan, P., & Lertwittayanuruk, C. (2026). A vision-based closed-loop framework for adaptive ergonomic human-robot collaboration. Frontiers in Robotics and AI, 13, Article 1948863. https://doi.org/10.3389/frobt.2026.1948863

Image Credits: AI Generated

DOI: 10.3389/frobt.2026.1948863

Keywords: human-robot collaboration, ergonomics, computer vision, collaborative robots, RULA, pose estimation, adaptive workstations, manufacturing, closed-loop control, human-centered robotics, vision-based, closed-loop

The research, led by Jessica Tuuf of Åbo Akademi University in Vaasa, Finland, together with colleagues from the university’s energy and information technology faculties and the Finnish abrasives manufacturer Mirka Ltd, examines what it would actually take to bring end-of-life coated abrasives back into the material loop at industrial scale. The work forms part of the Data-driven Sustainability Management project, financed by Business Finland through the Mirka Shape Ecosystem, an initiative aimed at producing verifiable sustainability metrics using data-driven models.

Coated abrasives are technically challenging waste. A typical product layers abrasive grains—often aluminium oxide, silicon carbide or ceramic materials—onto a backing of paper, cloth or polymer film, held together with adhesive resins. This multi-material construction means that no single recycling stream can process the waste directly. At the same time, the products are consumed in such volumes worldwide that even partial recovery of their valuable abrasive grains could meaningfully reduce demand for virgin raw materials, many of which are energy-intensive to produce.

The authors frame the problem within the European Union’s ambition to become a carbon-neutral continent by 2050, which has accelerated pressure on manufacturing sectors to cut waste and increase recycling. In the hierarchy of circular economy strategies, waste reduction comes first, and the study acknowledges that extending product life is the preferred option for coated abrasives. But because these products are consumed by design, the researchers argue that a complementary strategy—an organised system for handling spent abrasives—would carry a distinct advantage over simply sending them to landfill or incineration.

What distinguishes the study is its focus on prerequisites rather than process chemistry. While the technical aspects of recycling abrasive products have been explored to some extent in prior research, the large-scale return of end-of-life abrasives has never been systematically examined. The Finnish team therefore mapped the conditions that would need to be in place: reverse logistics capable of collecting dispersed waste from thousands of user sites, waste management frameworks that classify the material correctly, corporate strategies that commit to circular targets, and—critically—data governance that makes material flows visible and verifiable.

Data emerges as the connective tissue of the proposed system. Without reliable information about where spent abrasives originate, in what quantities, and with what composition, take-back schemes cannot be planned economically, and recyclers cannot certify the recovered grain quality. The study positions data governance alongside legislation and waste management as a crucial factor determining whether circular transitions in the sector succeed or stall. This aligns with a broader shift in industrial sustainability thinking, in which digital product information and traceability are treated as infrastructure on a par with physical collection networks.

Legislation adds another layer of complexity. Waste regulations determine whether used abrasives are classified as hazardous or non-hazardous, which in turn dictates how they can be transported, stored and processed. The study discusses how regulatory frameworks shape the economics of any recovery model, and how evolving EU policy on waste and circularity could either enable or constrain take-back systems. For manufacturers, anticipating these legal conditions is as important as perfecting the recycling process itself.

On the basis of the case study, the authors advance both material flow models and recycling concepts for the sector, and propose several alternative models for the future recovery of end-of-life coated abrasives. These range from concepts in which used products are returned through organised channels to those in which data collected during product use informs downstream handling decisions. The researchers emphasise that recovering abrasive grains could directly reduce the demand for virgin raw materials, closing the loop on one of the most resource-intensive inputs in surface-finishing manufacturing.

The collaboration between academia and industry is central to the study’s credibility. By working within Mirka’s ecosystem, the researchers could ground their models in a real production and distribution environment rather than in idealised assumptions, an approach increasingly favoured in circular economy research where the gap between pilot concepts and industrial reality is notoriously wide. The involvement of Åbo Akademi University’s information technology group alongside its energy technology specialists also reflects the study’s core thesis: that sustainability transitions are as much computational and organisational problems as they are materials problems.

For a product category as unglamorous as sandpaper, the stakes are surprisingly high. Coated abrasives are used in nearly every manufacturing chain, from automotive body shops to furniture workshops to semiconductor polishing, and their cumulative material footprint is substantial. If the data-governed take-back models outlined in the study prove viable, a product that has always been defined by its disposal could become an example of how digital information unlocks circularity in even the most fragmentary waste streams. The research directly supports the European Union’s objectives for advancing the circular economy, and it offers a template that other multi-material, short-lived product sectors—from filters to packaging—may soon find difficult to ignore.

Subject of Research: Data governance and circular economy models for recycling end-of-life coated abrasives

Article Title: Data as an enabler of circular transitions in the coated abrasives industry

Article References: Tuuf, J., Hellman, F., Frände, N., Härmälä, P., Risku, C., Björklund-Sänkiaho, M., Björkqvist, J., & Lundell, A. (2026). Data as an enabler of circular transitions in the coated abrasives industry. Discover Sustainability. https://doi.org/10.1007/s43621-026-04966-0

Image Credits: AI Generated

DOI: 10.1007/s43621-026-04966-0

Keywords: coated abrasives, circular economy, recycling, data governance, end-of-life products, take-back systems, industrial sustainability, waste management, material flow models, European Union, manufacturing, sustainability

The study behind this framework was carried out by researchers at Mälardalen University in Eskilstuna, Sweden, together with a colleague at Research Institutes of Sweden in Kista, and it appears in the journal Frontiers in Manufacturing Technology as part of the journal’s Digital Manufacturing section. The team set out to address a gap they identified directly in the manufacturing literature: while staged process models for AI development in industrial settings already exist, they tend to describe recommended activities at a level of abstraction that leaves practitioners guessing about what to actually do on Monday morning. Other scholarly efforts have produced detailed methods, but those methods typically cover only narrow slices of the development journey, such as the elicitation of requirements, leaving the rest of the process undocumented.

To build something more complete, the researchers began with a systematic review of development frameworks that had already proven useful in adjacent fields. AI applications in clinical medicine, human-AI interaction design, and software engineering have each generated mature bodies of work on how to structure the creation of intelligent systems, and the authors mined these traditions for transferable insights. They then combined and modified the existing frameworks to fit the particular realities of manufacturing, where production engineers, quality specialists, machine operators, data scientists, interface designers, and software developers must all coordinate around systems whose data pipelines, machine-learning models, and user interactions are deeply interdependent.

The result is a framework of 147 task recommendations organised into six conceptual stages of development progression. Each stage functions as what methodologists call a determinant framework: rather than prescribing a rigid sequence, it lays out factors and tasks that have been found to influence development outcomes, giving project leaders a detailed and actionable map they can adapt to their own circumstances. The granularity matters because the framework’s intended users are development project leaders with manufacturing backgrounds—people who may be experts in production engineering but who must nonetheless orchestrate contributions from specialists whose vocabularies and priorities differ sharply from their own.

The choice of project leaders as the primary audience reflects a deliberate theory about where AI implementation succeeds or fails. When front-line employees such as operators and technicians are the ultimate users of an AI system, the person driving the project must understand those employees’ operational workflows, work contexts, and cognitive processes well enough to communicate the system’s benefits, concerns, and limitations in language the front line actually uses. In the researchers’ conceptualisation, an AI solution system contains both a technical subsystem—the AI-enabled software or hardware itself—and a social subsystem encompassing the human actors and workflows that interact with it. Human actors can augment AI components by providing control and feedback on inference results, while AI components can in turn enhance human situation awareness, data analysis, and decision-making. Neither subsystem can be developed in isolation from the other.

Transparency emerges as a second guiding principle. The authors argue that a major barrier to effective collaboration in cross-functional AI teams is the lack of knowledge overlap: each participant can comprehend only the parts of the solution immediately relevant to their own expertise, so nobody anticipates what colleagues are doing or how their own activities ripple outward. By making the entire development process visible and legible to all participants, the framework aims to let a quality engineer understand why a data scientist needs certain labelling conventions, or let an operator see how their feedback loops shape model retraining. Transparency also pays dividends when manufacturing organisations must rely on external parties to acquire or co-develop AI systems, because an internal team that understands the full development process can set realistic, well-scrutinised requirements for vendors and research partners rather than accepting vague promises.

The empirical grounding of the framework came through two complementary routes. First, external reviewers examined the framework and assessed its coherence and practical relevance. Second, the researchers applied the framework in pilot form to two real-world development projects, observing how its task recommendations played out against the messy contingencies of actual industrial work. These pilots were preliminary rather than definitive validations, but they surfaced concrete patterns of utility. In early-phase project planning, the framework prompted deep multi-domain dialogues in which participants from different disciplines confronted questions they might otherwise have deferred until late in the project, when course corrections are far more expensive. Midway through projects, the framework supported reflective planning, giving teams a structured occasion to step back and assess whether their trajectory still matched operational realities.

The pilots also generated candid remarks about limitations. The researchers note that the framework’s use in iterative development environments requires care: modern AI development rarely proceeds in a clean linear sequence, and teams that work in agile or spiral patterns must adapt the six stages to cycles of prototyping, testing, and revision rather than treating them as a one-way march. The authors also confine the post-deployment stage—what happens after a system goes live in production—outside the detailed scope of the model, meaning organisations will need complementary guidance for maintaining, monitoring, and retraining systems once operational.

One of the study’s sobering findings is its characterisation of the current state of practice: the development of human-integrated AI systems in industrial environments remains immature. This immaturity helps explain the AI graveyard phenomenon. When early prototypes demonstrate inspiring potential but applications rarely cross into real operational environments, the fault often lies not in model accuracy but in the unmanaged seams between data collection, curation, inference-based decision-making, error handling, and system maintenance—activities in which shop-floor personnel are deeply embedded. Operators and technicians who collect data and handle errors are not peripheral users; they are components of the solution system itself, and treating them as afterthoughts undermines the socio-technical harmony the technology requires.

The stakes of getting this right are amplified by the environments in which manufacturing AI operates. These systems are frequently intended for high-stakes operational processes where faulty decisions or controls can severely damage production performance. A mis-classified part, an erroneous maintenance prediction, or an ill-timed control adjustment carries consequences that a failed recommendation engine in a consumer application never would. This asymmetry raises the cost of the knowledge gap: a data scientist who does not understand quality engineering constraints, or a production engineer who cannot articulate what the model needs, each introduce risks that compound silently until deployment.

Application areas for AI in manufacturing span production planning, automation, quality inspection, process control, and predictive maintenance, and each domain brings its own patterns of human involvement. In quality inspection, for instance, image-recognition systems may flag defects that human inspectors must verify, creating a division of labour that must be designed rather than assumed. In predictive maintenance, technicians’ tacit knowledge about machine behaviour may be essential for curating the training data, yet that contribution is often invisible in conventional project plans. The framework’s 147 recommendations give project leaders a vocabulary for surfacing these hidden dependencies early, converting what would otherwise be implicit assumptions into explicit, discussable tasks.

For manufacturing organisations weighing whether to adopt such structured approaches, the study suggests that the investment in process discipline may be what separates systems that create lasting value from those that stall at the prototype stage. The framework does not promise to eliminate the challenges of harmonising human and AI work, but it offers a shared map that lets diverse experts see the whole journey, anticipate one another’s contributions, and catch misalignments while they are still cheap to fix. As manufacturers continue to experiment with machine learning across their operations, the evidence from this research indicates that the organisations most likely to escape the graveyard are those that plan for the human side of the system with the same rigour they apply to the algorithms themselves.

Subject of Research: A development framework for integrating human work into AI systems in manufacturing

Article Title: A development framework for human work integrated AI systems in manufacturing

Article References: Yamamoto, Y., Aranda-Muñoz, Á., & Sandström, K. (2025). A development framework for human work integrated AI systems in manufacturing. Frontiers in Manufacturing Technology, 5, Article 1601903. https://doi.org/10.3389/fmtec.2025.1601903

Image Credits: AI Generated

DOI: 10.3389/fmtec.2025.1601903

Keywords: artificial intelligence, manufacturing, human-AI collaboration, socio-technical systems, machine learning, development framework, Industry 4.0, technology readiness levels, workflow integration, digital manufacturing, project management, predictive maintenance