Buggy Hardware Is the Best Hardware

Bio-Hybrid sensors that incorporate locust antenna into modern electronics use ML to detect scents with 10,000x the sensitivity of e-noses.

Nick Bild
1 year agoMachine Learning & AI
Wheeled robot with a locust-powered artificial nose (📷: Tel Aviv University)

When you stop and think about it, modern technological advances are really quite impressive. Having a supercomputer in your pocket, or having access to an artificially intelligent chatbot that can answer all of your questions, for example, would have seemed like science fiction just a few decades ago, after all. But in spite of all of these innovations, the fact remains that whatever we can do, nature can do it better. We have the computer, but nature has the much more energy efficient and versatile brain; we have super glue, but nature has the gecko’s foot.

One of the most striking examples of nature running laps around our technologies is in environmental sensing. Consider a dog’s nose, for example. It is orders of magnitude more sensitive at detecting specific odors than even state of the art artificial noses. And there is also the fact that a biological nose can detect hundreds of millions of scents, whereas the artificial versions can detect only a few. A team at Tel Aviv University knows that sometimes you have to recognize when you are beaten, and as the saying goes, if you can't beat them, join them.

This was the inspiration for their novel approach to improving artificial noses with sensors that they call Bio-Hybrids. These sensors pair super sensitive biological olfactory sensors with modern electronics to improve odor detection sensitivity by orders of magnitude. Their system makes use of the antennae of desert locusts, which happens to be their primary olfactory organ. When scent-related molecules come into contact with the antennae, electrical signals are generated that can be interpreted as a signal that a particular molecule was recognized.

Of course the researchers could not ask a desert locust what those electrical signals mean, so they came up with their own approach. By exposing the sensor to a particular odor, then capturing the electrical signals that they trigger, the team was able to train a machine learning algorithm to recognize those patterns. They started with a small set of eight different odors (including lemon, geranium, and marzipan) to prove the concept. But as you might expect one would do with a bug nose scent detector, they kept playing with it after the conclusion of the initial experiments, and found that it was quite adept at recognizing any number of scents, including those of various types of Scotch whiskey.

It was noted that this new device was about 10,000 times more sensitive than the best artificial noses presently available. Using this super sniffer, the team is working to build a wheeled robot that can follow scents to their place of origin. Looking further into the future, they believe the technology could be used to sniff out bombs or other hazardous materials. Perhaps a later version of the device may even prove to be useful in sniffing out certain human diseases, as dogs have been trained to do in some cases.

Nick Bild
R&D, creativity, and building the next big thing you never knew you wanted are my specialties.
Latest articles
Sponsored articles
Related articles
Latest articles
Read more
Related articles