Viewing Machine Vision Through a New Lens: The Latest Developments
17th Jun 2024To coincide with this month’s UKIVA (UK Industrial Vision Association) Machine Vision Conference (MVC) & Exhibition at the Coventry Building Society (CBS) Arena in the UK, where we will be showcasing a variety of our high-precision optics at Stand G4 from 18-19th June, we are exploring up-to-date advancements in the sector.
As the name suggests, machine vision enables production setups to see, analyse situations, and act accordingly. Thanks to these skills, it is employed across a diverse spectrum of industries – encompassing food and beverage, pharmaceuticals, automobile, and manufacturing – to save time, improve efficiencies, speed, and quality control, and, of course, enhance the safety of staff working in the premises by removing the need to perform hazardous tasks.
SWIR: Better Vision, Better Results
Although machine vision is typically brought in for several reasons, many of which we’ve mentioned above, one of the prominent motivations lies in its reliable capabilities; for example, the capacity to spot things that at times can’t be caught by the naked eye.
This is especially true for one trend revolutionising the field: Short-wave infrared (SWIR) procedures – the ‘Superman eyesight’ of the imaging sector. With its capability to distinguish elements by their reflectance properties and see through materials that are opaque to the human eye, SWIR proves highly advantageous for the non-destructive examination of internal structures. The approach finds its place in various uses, including evaluating food quality and monitoring moisture levels inside of objects.
Because of the advantages, it’s perfectly positioned for substantial expansion. In fact, analysts at French consulting firm, Yole Intelligence, forecasts rapid growth in the current marketplace for SWIR imaging technology, projecting a value of $2.9bn by 2028 [1].
Insight from Knight Optical:
Choosing Substrates for SWIR Imaging:
Materials like Magnesium Fluoride (MgF2),
Borosilicate Glass, Sapphire, and Fused Silica
are popular options.
Food For Thought: Employing LiDAR To Overcome Hurdles
The food domain heavily relies on imaging systems to maintain top standards of quality and safety, and it’s likely one of the first places you consider when thinking about machine vision. The setups are crucial for verifying produce is free from imperfections, contaminants, and inconsistencies, covering a wide range of items including fruits, veg, grains, meats (for example, looking for elements such as small bones), and processed goods. In addition to inspecting the food itself, the machinery sometimes contributes to packaging functions by assessing wrapping integrity, label accuracy, and overall product presentation.
Despite the sophistication of today’s arrangements, one humble starchy vegetable has, in certain circumstances, been noted as presenting a unique predicament due to its irregular shape. Michigan State University in the USA recognised the common issue and worked toward automating the typical manual sorting and grading of sweet potatoes in the US food industry. Its creative solution addresses the challenge posed by the wonky root plant’s naturally occurring, odd shape by incorporating LiDAR (Light Detection and Ranging) and artificial intelligence (AI) to grade and sort the vegetables based on size and defects. It captures shots of the spuds as they move and rotate on a custom-designed roller conveyor configuration, allowing the system to capture images of the entire surface of each potato [2].
Recruiting The Robots…
For a few manufacturing methods, a higher level of autonomy is necessary, which has prompted tech leaders to collaborate and combine cutting-edge devices to achieve ultra-efficiency. This is exemplified by the partnership between Flexxbotics – a specialist in smart factory bots and work cell digitalisation – and Cognex – a manufacturer of machine vision apparatus. Together, they’ve enhanced a sophisticated robotic machine tending structure that ensures precision quality, with seamless compatibility across the products [3].
…And The Machine Learning
The merge of robots and machine vision isn’t merely confined to industrial and commercial settings, however. We’re now witnessing pioneering utilisation in fields like healthcare, too, where tech like computer vision – often operating on a smaller scale but sharing core techniques with machine vision – are making a huge impact.
Take, for instance, a robotic feeding device developed by researchers at Cornell University, New York. Recognising that other similar machines require users to lean in to take a bite, a limitation for those with mobility difficulties, the team set out to address these challenges, also observing that sudden muscle spasms and restrictions in mouth movement further complicated the process.
By integrating machine learning, the feeder has evolved into so much more than simply a mechanical arm. Intuitive and capable of predicting the subtle action of human feeding, the bot identifies what’s on a plate, picks it up precisely, and delivers it directly into the user’s mouth. But, what really sets it apart is its ability to track movements in real time, adjusting to abrupt shifts and reacting to unexpected twitches – issues stated by the group at the start of the project. This adaptability makes it not just functional but user-friendly, ultimately transforming the lives of individuals with limitations [4].
AI In Automobile Machine Vision Systems
In the automotive arena, which is renowned for its early adoption of innovative tech, machine vision plays a vital purpose in ensuring quality and end-of-line detection operations. That said, a notable disparity exists between Germany and the UK in adopting AI within machine vision practices. According to a recent report by Zebra Technologies, 38% of those surveyed in the UK voiced satisfaction with AI delivery, whilst 34% of German contributors didn’t see use/relevance. Moreover, the study uncovered a significant portion – 40 to 60% – of respondents have yet to implement AI into their workings, with only a third planning to do so [5].
In a recent webcast, Anne Wendel, Director of Machine Vision at the VDMA Robotics + Automation Association, emphasised the significance of the automotive sector for machine vision tech, noting that AI software is anticipated to become a standard element here. While there’s a noticeable transition towards AI integration, Wendel highlighted that in 2019, many survey participants expressed reservations surrounding accepting AI-based approaches. However, a more recent 2023 one revealed a shift in perception, with hundreds acknowledging AI as a driving force in current and foreseeable utilisations [6].
To conclude, the industry shows promising signs of embracing advanced methods and innovations both now and in the future, and as a dedicated provider of optical solutions, we’re poised to support the evolution of machine vision technologies.
High-Precision Optics For Machine Vision Systems
Given the critical role of ‘vision’ in these processes, high-quality optical components are essential for optimal performance. With our extensive experience in supplying optical solutions for machine vision applications, we carry in-depth knowledge of optics tailored specifically for these purposes. Additionally, our long-standing history of collaboration with designers and engineers further reinforces our expertise in this specialised field.
FOOTNOTES:
[1] https://optics.org/news/14/4/15
[5] https://www.imveurope.com/article/harnessing-automotive-potential-machine-vision-ai
[6] https://www.imveurope.com/article/harnessing-automotive-potential-machine-vision-ai