Industrial automation has been evolving rapidly, but the vision of a factory that can run 24/7, maintaining itself and minimising the risk of accidents or human error, remains just out of reach.
For all the great leaps AI and machine learning have taken in recent years, these tools can only ever be as good as the data they have to learn from. The more data points these machines have access to, the greater their potential. When it comes to machine vision, this means sensing and imaging devices must be able to peer beyond the visible, and into the invisible world of infrared light.
Turning vision into reality
Doing so could unlock the full potential of machine vision. By adopting short wave infrared (SWIR) sensing, the fully automated vision of the future goes from fantasy to reality in an instant. It could allow packing lines to identify faulty products like bruised fruit, and automatically funnel them into a waste disposal system. It can detect internal stress and damage in machine components before they become an issue, simplifying preventative maintenance. It can revolutionise microchip production by detecting defects on silicon wafers, enabling fabricators to optimise their yields.
And this is just scratching the surface of what is possible with SWIR-powered machine vision. Collecting data from both visible and invisible light spectra can provide machine learning algorithms the information they need to manage many industrial functions that currently require human input. Production planning, job onboarding, logistics, maintenance and repairs, and quality control could all be carried out autonomously.
However, theorising about this autonomous future is very different from making it happen. As is the case in many industries, the cost of this technology is a significant roadblock.
At present, indium gallium arsenide (InGaAs) semiconductors are the only SWIR-sensitive option on the market for machine vision applications. However, InGaAs devices can cost around $10,000 USD per unit, and even then, offer a relatively poor resolution. These two drawbacks combined mean that they are unsuitable for all but the most highly specialised applications – it is not the kind of technology that could spark another industrial revolution.
The quantum solution
The solution to this problem requires scalable, cost-effective, and high-performance devices – and the only solution to tick all three boxes is the quantum dot (QD) sensor.
QDs are nanoscale semiconductor materials that can be tuned to absorb almost any wavelength of light across the visible, near infrared (NIR), and SWIR spectra. This alone makes them a versatile option for almost any machine vision application, but more importantly, QDs offer a superb level of performance. The pixel pitch and spectral range mean they resolve a crisp, clear image that offers greater visual clarity to both the human eye and machine learning algorithms.
Perhaps most importantly, though, QDs can achieve all this at a fraction of the cost of InGaAs devices. This is because QD sensors are made by depositing a QD film onto a complementary metal oxide semiconductor (CMOS) device. CMOS technology is very cost-effective and already forms the foundation of many consumer-grade sensors on the market today.
While work remains to scale up this technology to the level required, many of the challenges of developing viable QD sensing technologies have already been overcome by our INFIQ® solution. Other QD technologies use a QD film that is spin-coated, layer by layer, onto a substrate. This can require up to 14 individual layers on a single sensor, with each one being a potential failure point. INFIQ® QDs can be deposited in a single layer of a colloidal solution, dramatically reducing the risks of defects emerging.
INFIQ® is also the QD technology that is closest to overcoming the second major obstacle to widespread adoption – the presence of lead.
Lead is a highly effective material for constructing SWIR-sensitive QDs because of its broad absorption spectrum, which means it can be easily tuned to absorb a wide range of light wavelengths. However, it is extremely toxic and its use is banned for the majority of applications. At Quantum Science, we have pioneered the development of lead-free QDs that are responsive to wavelengths of up to 1,550nm.
This means we are in the unique position of being able to shape the future of machine vision technology – and, with INFIQ®, that future starts now.