Earl Yardley, director of Industrial Vision Systems (IVS) explains how Industry 4.0 is changing the manufacturing world.
Powered by unparalleled levels of developments in Artificial Intelligence (AI), big data, 3D imaging, and robotic process automation, the ‘factory of tomorrow’ is well and truly here. The manufacturing of medical device products is driven by innovative developments in automation, that have enabled organisations to create new ways to deploy virtual labour in the form of machine vision to oversee automated knowledge-based inspection tasks. The drive to fully flexible production control for Industry 4.0 manufacturing means that machine vision and AI-driven vision systems are critical to allow mass production where flexible variant manufacture is the key driver.
Critical changes to working practices and automation deployment are creating new opportunities, which include cutting edge production ideologies with vision robotics, flexible manufacturing, efficiencies through self-learning, and the ability to bring machine and human interaction even closer. Medical device manufacturing needs to be lean, high-speed, and possess the ability to switch product variants quickly and easily, all validated to Good Automated Manufacturing Practice (GAMP).
Case study
The latest generation vision systems from IVS are aiming to help a major medical device manufacturer to include validated documentation support using XML document exchange in their new Industry 4.0 factory; allowing for the creation of automated documentation once procedures for automated visual inspection have been validated. This documentation is applied for the examination of pump bodies, similar to those found in nasal sprays with a pumping atomiser; this system was designed and installed by IVS. The primary inspection standards for the system was the segregation of products against known measurement criteria to control the quality of the final products. In the age of Industry 4.0, it is critical that new variants can be quickly deployed and documented as part of the manufacturing process.
The solution consists of multiple inspection criteria all fed into a dial plate system via a bowl feed. The first check involves examining the intermediate piston stroke length, an indicator for the piston stroke for the pump, in the range of tolerance of +-0.01mm. Based on this result, an inspection is developed to distinguish between the five pump types available, allowing for quick changeover and flexible manufacturing.
The second check involves examination of the transparent plastic body containing the spring and ball of the spray mechanism. Using a contour matching function, verification of these critical components is completed. Typical errors contained in this area include double ball bearings, bent springs, misaligned springs and wrong intermediate piston. Finally, a medical device injection moulded cap is checked for vital dimensional tolerances, along with shorts and flash detection – a cornerstone of IVS capability.
A key to the application was incorporating the machine vision system into the machine’s cycle time, while being mindful of flexible switching for different parts; the optimisation of the matching function meant the image processing could be completed within the necessary performance required.
The solution involved the pump bodies and caps fed on a rotary plate; thus, the test position would be well defined in terms of positional tolerance. Once the pump body reaches the inspection station on the carousel, the Programmable Logic Controller (PLC) sends a signal directly to the I/O control within the IVS-CommandAi vision controller, evaluation by all cameras is completed and communicated back to the PLC. Each feature failure is indicated independently via single I/O channels and information is stored in Excel and exchanged to the factory information system by Ethernet connection.
Documentation in the pharmaceutical and medical device industry is critical, and once a complete solution is finalised the full inspection criteria is automatically created as an XML document. This gives detailed information on every inspection function used and how it was set-up, making IVS vision systems powerful when deploying multiple systems across the medical device industry.
Future vision systems
Industry 4.0 is driving IVS to develop new and innovative solutions for the future of medical device plastics manufacturing. Robotics, machine learning and artificial intelligence are reshaping automation, allowing medical device manufacturers to improve quality, maximise value, keep costs down or offer new services. New technologies are emerging, which will help shape the future of production control.
Hyperspectral imaging
Next-generation modular hyperspectral imaging systems provide chemical material analysis in industrial environments, and this can be harnessed to view the chemical composition of a product. Chemical colour imaging visualises the molecular structure of materials by different colouring in the resulting images. This allows the chemical composition analysis in standard machine vision software. Typical applications include plastic detection in different recyclable materials and blister pill inspection quality control. The main barrier for such systems is the amount of data and speed required for processing, but the development of faster processes, better algorithms and on-camera calibration still make this a hot topic.
In the production of high-value chemicals, there must be strict control over the final product’s size, purity, form, and morphology. Medical device manufacturing continues to require precision automated inspection at high speeds. The disposable needles used in pen-injectors, for example, need to be straight as well as being sharp. Hyperspectral imaging combined with traditional machine vision can check the contents of a syringe as well as assure that bent or blunt needles and mould defects are rejected as part of the medical device inspection process, long before they reach the end-user.
Optics development
There have been significant developments in optics over the past few years, particularly when it comes to reducing the size of optical components for them to fit into tighter spaces on medical device production lines as part of a machine vision inspection solution. This trend is expected to pick up even further pace throughout the next year due to the limitations on space in some production floors. Understanding measurement techniques and uncertainty when specifying and procuring optics for use in machine vision are critical.
Deep learning
New deep learning technologies are utilising advanced artificial intelligence which is taken from the development in machine vision for autonomous vehicles, social media processing and robotics. Deep learning algorithms using convolutional neural network classifiers allows image classification, object detection and segmentation at speed. Development of these new AI and deep learning systems is expected to increase across pharmaceutical manufacturing. These algorithms can also be utilised for surface inspection and defect detection on blister packs, moulds and seals – enhancing the quality and precision of deployed machine vision systems.
Higher speeds
Faster processes allow more data and image processing in real-time. Higher resolution cameras are utilised for greater processing leading to increased accuracy of vision systems over the next few years. However, one of the biggest challenges in the adoption of machine learning is how the data is handled. Cloud-based image collection and processing will become the norm to allow higher bandwidth image processing, as well as data saving and image collection for warranty protection for medical device production.
Conclusion
Industry 4.0-related technologies in plastics manufacturing are driving much of the changes that are currently taking place in production. This applies in all sectors, but it is particularly vital in high-specification and highly regulated industries such as pharmaceutical and medical device plastics manufacturing. There are many reasons for companies to move towards factory automation technologies, including creating more efficient production, increased production flexibility, making more effective use of resources, and improving productivity. We fully expect to see the growing demand for production processes which are highly flexible utilising real-time vision inspection to cope with multi-variant production in this age of flexible manufacturing.