DO I smell? It is an embarrassing problem we have all had to deal with. A run for the bus or a hot meeting room can leave you trying to check your armpit without anyone noticing. Luckily, AI is here to help.
UK chip-maker Arm, better known for developing the hardware that powers most smartphones, is working on a new generation of smart chips that embed artificial intelligence inside devices. One of these chips is being taught to smell.
The idea is that the chips will be small and cheap enough to be built into clothing, allowing an AI to keep tabs on your BO throughout the day. Arm also wants to add the chips to food packaging to monitor freshness.
The e-noses are part of a project called PlasticArmPit, in which Arm is developing smart chips made from thin sheets of plastic. Each chip will have eight different sensors and a built-in machine learning circuit.
It will look like a piece of cling-film with bits stuck to it, says James Myers at Arm. “PlasticArmPit will be the first application of machine learning in plastic electronics.”
Smells are made up of different combinations and concentrations of gases. The sensors on the chip will detect different chemicals in the air and the AI will take that complex data and identify it as a particular whiff.
The chip will then score the smell. If it is in the armpit of your shirt, it will tell you the strength of your body odour from 1 to 5, says Myers. “It’s the job of the machine learning to collect and interpret all the data and then alert the user if action is needed.”
“The chips will be small and cheap enough to be built into clothing, allowing an AI to keep tabs on your BO”
E-noses are not new. Julian Gardner, who pioneered the technology at the University of Warwick, UK, has been building them for three decades. In 1993 he co-founded a company called Alpha MOS that sells e-noses to the food industry.
The trouble is that these devices cost around $20,000, says Gardner. He has since developed smaller, cheaper versions that cost just a few dollars. But they need to be made even cheaper to be sewn into clothing, which is what Arm hopes to do.
“I think that if the sensors are almost free, then people could buy clothes with them,” says Gardner. But they will also need to survive in the wash, which could be a challenge even for plastic electronics, which are normally more resilient than regular electronics.
Alex Bond at Fresh Check, a London-based start-up that is developing a chemical test to check for bacteria on food, thinks e-noses are a good way to monitor food quality because they do not need to touch the food.
An AI-powered nose could also be tuned to pick up different types of smell. “Flexibility is important because beef doesn’t spoil in the same way as fish,” Bond says. “And a pork loin may be classed as spoiled, but still be suitable to be turned into sausage.”
However, Bond thinks that it will be a challenge to get smart chips into food packaging – no matter how cheap they are. “Any increase to packaging costs is hard to justify,” he says. “Most food manufacturers have exceptionally tight profit margins, so there has to be an incredibly strong incentive for them to adopt more expensive packaging.”
One option may be to limit the use of sensors to premium foods or countries where there is a higher risk of contamination.
Still, Arm hopes to embed more than just e-noses into packaging. Chips built into plastic could be used to signal what kind of plastic a bottle or wrapper was made from, for example.
This article appeared in print under the headline “Chip with a nose will tell you when you have BO”