This post is also available in: עברית (Hebrew)
A novel technology based on machine learning could replace sniffer dogs in finding explosives. High-tech sensors made from genetically-modified living cells can detect odors in the air and will be able to provide airports and airlines with situational awareness on the chemical, explosive, bacteriological threat.
The cells are fused with a silicon chip that processes odor signals and passes them through a machine learning system for classification, performance improvement and error correction. If a smell is identified as a security threat, the purple, jelly-like device — called a Konikore — lights up.
The US startup Koniku, in collaboration with Airbus, will start field trials of the devices in December, at Changi Airport in Singapore and San Francisco International Airport.
Weighing less than 350 grams and about half the size of a smartphone, the devices could be installed in multiple locations: on the revolving doors at the entrance to a terminal, at check-in desks, or at the entrance to an aircraft. This would not only make them easier to deploy than their canine counterparts, but more cost-effective.
“Dogs work for 20 minutes maximum, they can be easily distracted, and they are very, very expensive to train — it’s an average cost of $200,000 per dog,” says Julien Touzeau, head of product security for Airbus America.
Potential uses for the device do not stop at security, Koniku has been investigating whether the same technology could be used to detect viruses like Covid-19, following reports that dogs may be trained to sniff it out. While they cannot detect the actual virus, respiratory diseases cause a change in the body odor of sufferers, which dogs — or “electronic noses,” devices that can detect odors — may be able to pick up on.
However, some scientists specializing in electronic noses are skeptical of the technology.
Timothy Swager, a chemistry professor at the Massachusetts Institute of Technology, says that to pull off what Koniku claims would require “some technical miracle.”
Integrating natural proteins into silicon circuits is extremely difficult, he says, and the fragility of cells and the complexity of their interactions with chemical substances makes them hard to work with.