Intel partnered with Cornell University to conduct research using Loihi, Intel's neuromorphic research chip. Cornell analyzed the electrical activity of the brains of animals as they smelled different odors, and then Intel created algorithms to translate them to neuromorphic silicon. The chip is then trained using machine learning to identify the different odors of harmful chemicals.
The work is still far from complete but has shown a promising beginning. While Loihi could potentially learn 10 different odors, the system has trouble differentiating odors that are similar but come from different sources (like different fruits that have similar smells, for example). It has, however, proved itself capable of picking out a hazardous smell among a background of everyday smells like perfume and food.
“In the future, portable electronic nose systems with neuromorphic chips could be used by doctors to diagnose diseases, by airport security to detect weapons and explosives, by police and border control to more easily find and seize narcotics, and even to create more effective at home smoke and carbon monoxide detectors,” said Intel in a statement.
Editor's note: This article has been updated to reflect Cornell's participation in Intel's work, and the current capabilities of the Loihi chip.