Digitalisation in manufacturing is everyone’s dream in the apparel industry and different solution providers have tried and are trying to solve it in their own way. There is no doubt that Covid has accelerated some of the digitalisation processes; garment sampling and approval is one such good example. Amongst the different functions in apparel manufacturing, the sewing floor has the highest MMR (man-to-machine ratio) and is considered the most difficult to digitalise. Prabir Jana, Shahi Chair and Professor, NIFT Delhi maps the tech evolution of real-time production tracking on the sewing floor.
One of the key objectives of the digitalisation of any function is ‘visibility’ and sewing is no exception. In 1960s and ’70s, managers used to depend on bundle tickets and bundle tracking sheets to gain visibility of their sewing line. The size of elaborate bundle tickets used to be near A4 size and full of alphanumeric. Barcode technology was probably the first effort to digitise such large amount of information into a small form factor and make it visible on a computer screen. Each bundle of garment components (either cut parts or semi-finished) used to carry a barcode sticker and scanning of the same would update about how many pieces are completed in which operation in real time. The bundle would travel between different operations in a PBU sewing line and the difference of timestamp would automatically calculate the time spent in the operation, hence efficiency.
While barcode remained the standard for few decades, technologists explored different technologies including ‘dallas chip’ before stumbling upon RFID technology. The basic purpose of digitalisation remained the same, that is, which sewing operator is producing how many pieces every hour and where WIP is building up? The definition of real time was based on the frequency of barcode scanning (because some human executives needed to scan the bundles manually) and it was generally updated every hour. A large display board used to depict the hourly figures at the beginning of the line.
RFID technology was developed thereafter to scan multiple/bulk SKUs where tag/sticker faces were non-exposed, non-aligned with scanners and could be scanned also from a distance. For example, if there are 30 different sizes and colours of the same shirt in one closed carton and one would like to know what is inside the carton without opening the carton and scanning each shirt separately, RFID technology comes handy here as it enables scanning of multiple such cartons once they are trucked out of warehouse by installing a RFID scanner at the gate. Some overenthusiastic technologists implemented RFID technology for real-time production tracking in sewing line which is an ‘overkill’. People were so obsessed with the new technology that even some overhead hanger system (OHS) companies replaced their barcode with RFID stickers. Imagine when hangers with chips are running on a rail with scanners installed by the side of the rail perfectly aligned! What a misuse of technology!
The definition of real time was based on the frequency of barcode scanning (because some human executives needed to scan the bundles manually) and it was generally updated every hour. A large display board used to depict the hourly figures at the beginning of the line. |
Then came showbiz technology; these were just like the same barcode stickers scanned by tab screens instead of handheld scanners or even QR codes in bundles scanned by tab screens. Either tabs were installed at each sewing workstation or at the end of the sewing line. On top of that, the management felt that visibility should not be restricted to the manager’s cabin, therefore large LCD screens were installed on the shopfloor to display the real-time status. Surely those gadgets uplifted the ‘digital status’ or ‘digital look’ of the traditional sewing floors and probably heralded digitalisation of sewing floor.
The last decade saw tremendous improvement in data visualisation (rise of dashboarding); numerous solution providers joined the bandwagon. However, in the whole process, one important aspect that was overlooked was the accuracy of the data. The below map shows the evolution of technology against data accuracy; while manual and barcode/RFID technology offered almost similar data accuracy, the IoT sewing machines by few global sewing machine manufacturers (Brother/Juki/Jack/Hikari/etc.) offered improved accuracy of data.
Figure 1: Tech map of real-time production tracking on sewing floor
While after nearly half-a-decade, now we are talking (and preaching) about new disruptive technologies in manufacturing, but if we introspect, there is no real technology invasion in this domain of real-time production tracking on the sewing floor (sans IoT). The IoT sewing machines are still at nascent stage with current technology limitations (dependency on back tack) resulting in low accuracy. The adoption of IoT also remains very low due to the high cost of changing the entire sewing machine. Most of the recent start-ups surprisingly concentrate more on the visualisation part and ignore the accuracy part and come in the lower band of tech map. It is also important to note that lower band of technologies (manual/barcode/RFID/OHS) primarily rely on external actions to capture data. The external actions may be tap/scan/touchscreen/keyboard operations and hence these are prone to manual errors (both unintentional as well as intentional). The higher band of technologies do not expect any human intervention in data capture and rather data is extracted from machines.
While after nearly half-a- decade, now we are talking (and preaching) about new disruptive technologies in manufacturing, but if we introspect, there is no real technology invasion in this domain of real-time production tracking on the sewing floor (sans IoT). The IoT sewing machines are still at nascent stage with current technology limitations (dependency on back tack) resulting in low accuracy. |
But there is some good news; there are brand new start-ups which are daring to change the whole scenario lately. Artitex is a German start-up that is exploring a low-cost solution using smartphone sensors to capture sewing machine production non-invasively; although there are some challenges in tech and execution. OpenSeam is a US based start-up and Apparel4Tech is an India based start-up and both are trying to solve the problem using AI/ML. Both are working on proprietary yet cheap sensors (single or multiple sensors) that can be retrofitted on, making data collection from existing sewing machines to be a game changer. Then the data can be processed by AI/ML models to calculate the production accurately.
Figure 2: Real-time production tracking in Gartner’s Hype Cycle
It is interesting to see how tech is evolving and maturing over time. If we break down the tech of real-time production tracking as per Gartner’s Hype Cycle, we can very well say it has crossed the Peak of Inflated Expectations and is nearing Trough of Disillusionment. While quite expectedly, many of the current start-ups will vanish in disillusionment, the sewing machine OEMs are conspicuously absent in the race. Hopefully very soon, we will be sliding up the Slope of Enlightenment riding on the AI/ML wave of new tech generation start-ups.