Biomimetic Olfactory Chips: Are Artificial Intelligence and E-Noses the Next Canary in a Coal Mine? – Securities.io

The Difficulties Of Artificial Olfaction

While machine vision has made tremendous progress in the last few years, other artificial senses have lagged behind. One of them is the sense of olfaction or smell'.

This is because we have long known how to get a precise electric signal in response to light, something that has been used on a massive scale since the first digital cameras. In contrast, smell is essentially the detection of volatile chemical substances.

This is a lot more difficult for a few reasons:

For all these reasons, most chemical / olfactory digital detection is currently limited to a few chemical compounds. And generally only used in an industrial setting where the dangerous chemicals to be detected are expected to come from accidents or leaks, for example, carbon monoxide, ozone, chlorine, etc.

This could change thanks to the development of biomimetic olfactory chips by researchers on the team of Prof. Fan Zhiyong, Chair Professor at the Hong Kong University of Science and Technology (HKUST).

The way the sense of smell works in animals & humans is through an array of chemical detectors called olfactory receptors, able to detect with high sensitivity a wide range of volatile chemicals.

The number of genes coding for such olfactory receptors can vary from 300 to 1,200 depending on the species and how important the sense of smell is for it.

So, instead of having one receptor for every possible chemical molecule, every compound will have a unique footprint produced when activating each of these receptors slightly differently. The olfactory bulbs then assemble this complex signal into a nerve signal and interpret it by a part of the brain called the olfactory complex.

HKUST researchers have created a way to replicate this system, bypassing the constraints of building a miniaturized receptor for each possible chemical compound.

They assembled nanotube sensor arrays on a nanoporous substrate, reaching up to 10,000 individually addressable gas sensors per chip.

This data is then processed by a neural network algorithm to be translated into a perception of a specific chemical digital smell.

This design gives the olfactory chips the potential to simultaneously detect both the presence and concentration of a dozen or more chemicals at once.

As a demonstration, the team created a biomimetic olfactory chip that demonstrated exceptional sensitivity to various gases, and with excellent distinguishability for mixed gases and 24 distinct odors.

They then integrated both the olfactory chip and vision sensors on a robot dog, creating a combined olfactory and visual system that can accurately identify objects in blind boxes, pretty much like a real dog.

The most immediate application of olfactory chips is where most chemical detectors are currently used: safety applications. This includes factories, water treatment stations, petrochemical industries, pipe leak detection, and environmental monitoring (air pollution, etc.).

These new types of detectors could detect more chemicals at once than previous technologies, allowing for a larger data stream and better assessment of safety.

As demonstrated by the robodog prototype, such a detection system could be used to detect otherwise invisible threats. From drug smuggling to detection of explosives, every activity where sniffer dogs are used could be systematized, thanks to the merger of AI, autonomous robotics, and olfactory chips.

Search and rescue could also benefit from olfactory chips to find survivors under destroyed buildings after a natural catastrophe.

One reason why most animals have a developed sense of smell is to detect if a food is edible or spoiled. We can imagine that very sensitive olfactory chips specialized in food products could be very useful for the food industry.

Similarly, farming drones could also be used to smell the ripening of fruit, the presence of fungal crop diseases, insect pheromones, etc.

It has been known for a while now that some diseases are associated with the emission of specific smells. Anecdotal data of cats or dogs able to detect cancer have now been proven more than just urban myths through the use of artificial sensors.

Most notably, several cancers have started to be detected through these methods, with the electronic nose able to do so with a 95% accuracy.

The findings suggest that the Penn-developed tool which uses artificial intelligence and machine learning to decipher the mixture of volatile organic compounds (VOCs) emitting off cells in blood plasma samples could serve as a non-invasive approach to screen for harder-to-detect cancers, such as pancreatic and ovarian.

Penn Medicine News

We also see companies like BrainChip using digital olfactory detection to detect bacteria in blood samples.

It is likely that the more olfactory chips become sensitive and able to detect dozens or hundreds of compounds at once, the more such discoveries could be used for diagnosis, not just of cancer but of many other diseases, especially metabolic diseases.

Contrary to the current version, it could maybe achieve this only from the smell of our skin or breath, not even needing a blood sample.

As a purely silicon-based system, olfactory chips could be integrated into our omnipresent small electronic tools like the smartphone.

It could be useful to constantly monitor and automatically detect threats like carbon monoxide, smoke, or gas leaks or judge the safety of food.

We could also imagine more trivial but nevertheless potentially useful and popular applications, like helping while cooking, recognizing spices, etc.

In the longer run, if coupled with a smell generator, it could even enable the digital transfer of (preferably good) smells between phones.

Another more distant in the future, but not impossible application would be to integrate such olfactory chips capability into the human body.

Especially considering the quick progress of human-machine interfaces, like, for example, Elon Musk's Neuralink.

We could easily imagine such a sensor being integrated into our bodies and giving us warnings about harmful chemicals at levels below what is biologically possible. Or for chemicals we are completely unable to detect naturally.

In the future, with the development of suitable bio-compatible materials, we hope that the biomimetic olfactory chip can also be placed on human body to allow us to smell odor that normally cannot be smelled.

It can also monitor the abnormalities in volatile organic molecules in our breath and emitted by our skin, to warn us on potential diseases, reaching further potential of biomimetic engineering,

Prof. Fan Zhiyong

The potential of olfactory chips is likely to be confined in the first years to serious applications with clear use cases, from disease diagnostics to threat detection. So these applications are most likely where we can find companies that could benefit from this innovation.

(this list did not include chip companies with strong potential in olfactory chips and sensors, but whose largest part of their revenues will most likely stay driven by classical computing chips, like for example Intel's neuromorphic chipor IBMs SyNAPSE Scalable energy-efficient neuro synaptic computing chip).

This artificial intelligence company specializes in creating chips that mimic the human brain through Neural Network Layer Engines (NPEs).

It claims to be the first to commercialize neuromorphic technology. It also sees itself as ahead of serious competitors like IBM and Intel chips, thanks to on-chip learning, standard ML workflow & on-chip convolution.

It is focused on vision, audio, olfactory, and smart transducer applications.

This makes the company a very good candidate to benefit from progress in olfactory chips. It could directly license the HKUSTs discovery, try to replicate it, or see its own chips become a key part of the hardware required for interpreting the nanotube sensor array data.

The company sees a massive potential market for its products, including machine vision and olfactory capacities.

BrainChip has a high-margin IP business model, where it licenses its technology for an upfront fee and royaltiesstreams, and then partnering with system integrators to create the final product.

Honeywell is a leader in detection & sensors, with a strong or dominant presence in industries like building automation, aerospace, and safety (many of its aerospace and building activities are linked to sensor technologies).

As a recognized leader in sensors & monitoring, it could be in a prime position to commercialize and expand the scope of gas detectors from its current limited (but already lucrative) state to an omnipresent tool.

Honeywell is also at the forefront of other technological innovations, notably quantum computing through its ownership of 54% of Quantunuumand a business sector we discussed in our article The Current State of Quantum Computing.

It is also active in Liquid Metal Printing, something we discussed in Liquid Metal Printing May Become a Productive Force in the Landscape of Manufacturing and Design.

Honeywell is already a massive company in the sensor and automation sector, with ambitious goals in a large array of innovative technologies.

So even if biomimetic olfactory chips could be a competitor in the short term, it is likely that it will be able to adapt and benefit from the growth of the olfactory sensors market, either through its own R&D or through acquisitions of smaller companies.

See original here:
Biomimetic Olfactory Chips: Are Artificial Intelligence and E-Noses the Next Canary in a Coal Mine? - Securities.io

Related Posts

Comments are closed.