AI photo
SHARE

Oriental fruit flies are one of the biggest scourges to farmers around the globe, often forcing officials to put crops into quarantine just to keep Bactrocera dorsalis shut out. In Taiwan, where the situation is especially dire, scientists are using artificial intelligence tech that can determine, with uncanny accuracy, where and when an outbreak is about to happen.

The current method for detecting the pests is simple: lay down chemically enticing traps and hand-count the bugs every 10 days. The Taiwanese scientists are doing it a better way: automating the process with infrared lasers that scan the traps. Every time the beam is broken, add one to the tally. That number is radioed to a local station every 30 minutes, where officials can monitor it. The traps are also fitted with weather sensors to help keep track–high levels of humidity or other changes would increase the likelihood of an infestation.

So far they’ve set up 240 traps to regularly monitor the flow of fruit flies. When a trap counts more than 1,024 fruit flies in 10 days, it sets off an alert. But algorithms help it learn what’s normal for the area and current weather, letting it adjust to specific situations. By testing that system on data from past traps, they found it could predict an outbreak with 88 percent accuracy.

New Scientist