Industry 4.0 technology: What’s in store for textile and apparel?
The debate is still raging between manufacturers, consultants and academicians whether Industry 4.0 required for apparel manufacturing and even then whether that is relevant for countries in South East Asia. Without getting into the debate of whether COVID has delayed or hastened the Industry 4.0 adoption, we are going to see an exciting future ahead, says Dr Prabir Jana.
The amount of publicity “Industry 4.0” has received during last 5 years through catalogue and or promotional materials (of product and service providers) and webinars and workshops is just amazing. Now we need to acknowledge that the awareness (or misuse of the phrase!) phase is over and we need to come to brass tacks of implementation. Many apparel manufacturers across the world started taking definitive steps; acknowledging the importance of same, Shahi Exports, India’s largest apparel manufacturer has taken concrete step by setting up Industry chair at National Institute of Fashion Technology, Delhi during the year 2019 with clear emphasis to conduct research in the domain of Industry 4.0 that is likely to disrupt apparel manufacturing industry across globe. Several researches are currently underway through the initiative.
The debate is still raging between manufacturers, consultants and academicians whether Industry 4.0 required for apparel manufacturing and even then whether that is relevant for countries in South East Asia. No, I am not discussing the why Industry 4.0 here and leaving that space for consultants. Way back in 2016 a Roland Berger report predicted that Industry 4.0 has higher potential of disruption in apparel industry than in textile industry, which is quite surprising from the point of view of existing level of automation already exists in textile in comparison to apparel industry. In this article I am going to discuss about the technology and human resources that need to put in place once any organisation decided to implement Industry 4.0.
Industry 4.0 or the fourth industrial revolution was first introduced by a team of scientists developing a high-tech strategy for the German government. It involves automation and data exchange in manufacturing technologies, including cyber-physical systems, the Internet of things, cloud computing and cognitive computing, additive manufacturing, augmented & virtual reality and creating the smart factory.
Data
Machine learning (ML) and narrow forms of artificial intelligence (AI) have officially reached the mainstream and data is coined as new oil. We all know that the explosion of innovation we’re seeing in AI/ML stems from a series of rapid technological advances over the last few decades: faster/cheaper computers (per Moore’s Law), variable-cost cloud computing, cost of cloud storage is plummeting by orders of magnitude (often termed as Bezo’s Law, which says “a unit of computing power price over cloud is reduced by 50 percent approximately every three years”).
We have seen proliferation of AI/ML application in apparel retail industry, stitchfix, stylumia are popular names now. However, we haven’t seen similar application of AI/ML application in apparel manufacturing industry, purely because of lack of availability of data. Unless we start harnessing data from production floor soon, it will be another decade to see some worthwhile application of AI/ML in apparel manufacturing. Just to put things in perspective, all manufacturing organisation collects magnitudes of shop floor data daily from production floor; like loom production and quality records, daily sewing output line wise, fabric inspection record roll wise, DHU (Defect per Hundred Unit) record line wise, just to name a few. However none of these records are collected and stored in a manner that can be used for AI/ML application.
While textile manufacturing organisations might be in a better position to start as most of the data can be machine generated, while apparel manufacturing organisations are still in a bind as most of the data are human generated. One of the topmost priorities is real time production data from each sewing operator in sewing line. There is a race to develop technology to do this flawlessly yet cheap; there are solutions that uses RFiD or NFC or similar that can be installed with any ordinary sewing machine; other option is IoT sewing machines that can communicate two way with any computing device through WiFi. NIFT’s i-SMART technology creates the digital skill matrix of sewing operator which is crucial for management for scientific line planning, re-training, recruitment and appraisal.
Robotics
Robotics has a tough journey in textile and apparel industry (particularly in apparel). The limpy dimensionally unstable fabric was the main culprit in making robotic application impossible in apparel manufacturing. To be specific it was not the robotic arm, but the end effector that was simply not capable of handling fabrics. An end effector is a peripheral device that attaches to a robot’s wrist, allowing the robot to interact with its task. In layman’s terms if we imagine a human hand as robotic arm then the end effector is the palm and fingers.
The challenge was separation and picking up of single ply of fabric from a stack of plies; due to porous and dimensionally unstable nature of fabric, end effectors needed special design (ventura picker or Clue & Peabody picker) and time-consuming setting up making the process economically unsustainable. Researchers have tried and given up during early 90’s. MITI, Japan has actually created an unmanned sewing line for production of men’s brief during 1990 where simply cut parts need to be fed at one end of line and completely sewn brief will come out at other end of line. Such unmanned sewing line is workable for producing millions of pieces of one style and one size, which is not a reality in fashion industry. To put in current perspective Softwear Automation’s much hyped T-shirt line that was supposed to produce one T-shirt in 22 seconds for Adidas is very similar. May be COVID played spoilsport in the Adidas project, but we may see similar technology in near future.
Let’s understand what’s changed since 1990s that will make robotic sewing possible? Researchers decoded the working of sewing operator long before; a vision based system will capture image of fabric at needle point (just like sewing operator is watching the fabric being sewn carefully), the captured image will be processed by computer in a split second (much like neurons in our brain) and direct robotic arms to manoeuvre/guide/fold/pivot the fabric through needle point (just like dextrous fingers manipulate fabric plies). The hurdle during 90s was lack of digital image capture (1000 frames per second), low computing power to process the image; today companies like Vetron and Softwear Automation are showing operator less sewing possible. Grabit is solving the end effector problem with electro adhesion technology that can separate plies of georgette or denim with equal ease and pick and place easily.
IoT
Whether the popular prediction of one trillion sensors in connected devices by 2020 coming true or not, IoT or IIoT is simply the hottest and most talked about technology in Industry 4.0 bucket. This is also one technology already in use in multiple machinery and equipment in textile and apparel, generating valuable data to feed AI/ML applications. The most common IoT application in apparel manufacturing today is CNC cutting machine for layer cutting of fabrics. These machines are fitted with multiple sensors capturing and communicating crucial process and machine parameters in real time.
While the data is enriching the machine manufacturer to improve the machine feature and conduct predictive maintenance (for the manufacturers), in most of the cases the apparel manufacturer is simply unaware of the existence of such data and make any use of it. Another application that is getting traction is IoT enabled sewing machine; although there are some application limitations, these sewing machines can transfer the sewing production data of individual operator wirelessly to centralised server with reasonable rate of accuracy.
A host of brands like Brother, Juki, Jack, Hikari, etc. offers IoT enabled sewing machine today, Jack also launched IoT enabled spreading machine. As more and more machines are now coming with IoT sensors built-in, there is a possibility of better data collection from production floor and development of possible AI/ML applications.
Area of application
One approach for the industry should be to identify the low investment yet high impact areas. And in my opinion those applications that are repetitive and high chances of manual error, should be targeted first. Let’s talk about fabric defect inspection; today in more than 90% of the cases human being are engaged to look for the defect through naked eye, throughout the day and creating defect report for the fabric. Study says accuracy of manual defect inspection is between 60-70% and production rate is 15 meter per minute maximum, whereas camera based visual inspection using AI/ML can achieve accuracy of higher than 95% with double production rate. Today AI/ML applications are low investment, therefore fabric inspection can be a potential area of intervention that is low investment yet can result high impact in productivity and quality.
Similarly other applications where existing data set can be used to train ML models can be good area of applications. Use of cobots in loading or unloading functions where precision placement is necessary, is another potential area of intervention. Available IoT solutions are currently cost prohibitive, hence developing any custom solutions may be options provided factories set up clear targets of applications.
Conclusion
We have discussed that manufacturing operations are data shy, and therefore limiting development of data centric solutions for apparel manufacturing. However, things are changing very fast. Some new age start-ups like Groyyo are vowing to change that. Brining hosts of manufacturers in common e-platform and monitoring their production and quality will create the much needed dataset for future AI/ML applications.
While technology and process digitalisation is important for Industry 4.0, the digital transparency across supply chain is equally important to avoid costly inventories. Working towards that initiative four of Asia’s largest manufacturing organisation and one retailer come together to create DXM, a local, on-demand manufacturing company that’s co-owned by different supply chain actors. There are five founding partners: Shahi Exports (India), Brandix (Sri Lanka), MAS Holdings (Sri Lanka), Busana Group (Indonesia) and Carhartt. Without getting into the debate of whether COVID has delayed or hastened the Industry 4.0 adoption, we are going to see an exciting future ahead.
About the author:
Dr Prabir Jana is the Professor, Department of Fashion Technology, Shahi Chair Professor, Industry 4.0, National Institute of Fashion Technology (NIFT), Delhi. With three decades of experience, Dr Prabir Jana is a visionary in development, application and management of smart technology in textile & apparel manufacturing. He has four patents and 15 book/book chapters in his name. His special research interests include technology improvisation, sewing automation, industrial engineering, ergonomics, 3D printing and team working in clothing manufacturing.
