The new digital nose uses MEMS, chips, and AI to detect smells down to the parts-per-billion range.
AFP via Getty Images
A 40-year-old Taiwanese company says it’s invented the “world’s first” commercially available AI nose, which it is debuting this week at SEMICON. Uses include detecting illness in humans in healthcare settings, boosting safety in factories, and improving chip yields in semiconductor fabs, where the slightest impurities can ruin millions of dollars of expensive processors.
Imagine today’s emerging humanoid robots with functioning noses. Or Roomba-like vacuums that can sense safety factors while they clean our homes. Or just smarter smoke alarms and radon detectors that also check for dozens of different contaminants.
The “AI Nose” from Ainos starts with tiny micro-electro-mechanical sensor arrays that mechanically “sniff” the air. The technology then uses a proprietary Smell Language Model – think SLM versus typical AI LLMs like ChatGPT – to analyze the scent signature.
Smell is a “new category in industrial sensing,” says Ainos CEO Eddy Tsai.
Ainos’ SLM has been trained on a scent dataset collected over 13 years, the company says, and the mechanical nose can detect volatile organic compounds down to the parts-per-billion range. The “SLM” then translates that physical data into a “Smell ID,” an indexable, machine-readable representation of scent patterns.
(Interestingly, dogs can identify smells at the parts-per-trillion level: an order of magnitude better.)
The first market the AI nose is aimed at is the semiconductor industry, which is a massive $115 billion industry in Taiwan. TSMC, or the Taiwan Semiconductor Manufacturing Company, commands around 50% of global market share in advanced semiconductors all by itself.
Even tiny chemical leaks in semiconductor fabs can cause costly yield losses or safety hazards, so that’s a potentially lucrative market.
The company is also targeting hospital infection control, environmental monitoring, and food quality assurance. Ainos also highlights flexible detection ranges, suggesting that future modules could be tailored for detecting everything from ammonia in clean rooms to methane in energy facilities. That could broaden AI Nose’s market well beyond chips and into environmental compliance, agriculture, and municipal infrastructure.
Ainos isn’t alone in the artificial nose business.
There’s also Noze, which has an investment from the Bill & Melinda Gates Foundation, Canaery, which is focused on port and shipping safety, and Koniku, which says it is using biological neurons to for applications like airport security, military sensing purposes, and agriculture.
Universities, startups, and even DARPA have experimented with digital noses, often focusing on explosives detection, disease diagnosis through breath, or environmental sensing.
To date, most efforts remain niche, experimental, or lab-bound.
Eventually, however, some company will crack this market, the sensors will rapidly come down in price, and we’ll be able to easily access much more detailed information about the safety and status of our homes and factories.
Or, perhaps, we can just use our canine friends.