February 28, 2021

Artificial intelligence to prevent us from flooding garbage | Science

Artificial intelligence to prevent us from flooding garbage | Science

The wastebaskets in downtown San Francisco exude waste of all kinds, almost at any time of day, or at least that seems to eyes little trained in the collection of garbage. The cleaning services are not enough and the detritus often accumulates around the bins once they are full. According to data from the mayor's office, last year more than 8,000 people slept continuously in this city and in 2018, more than 26 million people visited San Francisco. The cleanliness of this city is also one of the most expensive in the United States, with an annual budget of 31 million dollars, double or triple that much larger cities like Los Angeles or Chicago.

To solve this problem, which is not only aesthetic but also can generate pests and bad smells, the Nordsense company has a Silicon Valley-style solution; sensors that detect when the bins are about to overflow and alert the cleaning services. Nordsense will place these small sensors in the next months at the top of a thousand paper mills in several neighborhoods of San Francisco.

"We had several months of testing to see what information we were collecting and for the city of San Francisco to see that our system works," Manuel Maestrini, founder of Nordsense, explained by telephone to EL PAÍS. "We had three months to place a hundred sensors and the test worked quite well. In fact, the city realized that of those hundred paper mills, in reality only very few overflowed daily. "

Maestrini says that knowing where more waste is generated is crucial to make cleaning tasks effective, fast and also, pollute less. "If instead of going with the truck making the daily rounds, they already knew exactly which bins are being filled frequently and at what time, they would do the collection in a much more effective way," says this software engineer.

Thanks to artificial intelligence, all the data collected by the sensors will be filtered by algorithms that will generate the daily routes, updating them as data is sent. "It's immediate: as soon as the sensor detects that the bin is about to fill up completely, it sends the signal to the application that analyzes the information and generates a new route."

Not only detect that the trash has reached the top, but every fifteen minutes the sensor scans the entire surface of the bin and is able to know if it is really full or if for example, seems overwhelmed because there is a pizza box. "We do not want the dumpsters to arrive to find that there is really almost no garbage, so we have managed to get the sensors to recognize objects," says Maestrini.

During the first test in San Francisco they managed to reduce the overflow of garbage to 80 percent and the cleaning service to 6 percent in the areas where they installed these devices.

Maestrini devised this system during his master's degree in Denmark, a country where he still resides and where he began his initial tests. "I started with a small pilot program in Copenhagen that the city council liked a lot and then I gave the jump to San Francisco through a call for startups that helped solve the garbage problem," says Maestrini.

Nordsense is also operating in Israel, with two pilot programs. But as the founder of this company says, they have more than twenty cities in Israel and in other countries waiting to begin the installation of thousands of sensors. "We are overwhelmed," explains the engineer, "we can not cover more cities for now, but in the near future we will be able to expand more and expand and improve our algorithms."

Although it may seem almost banal, which many of us take for granted, the collection of garbage can be a real headache for many municipalities. From Nordsense they say they want to change a system that is outdated and that does not work as well as it could work. "This is how we help governments, citizens and the environment," Maestrini says. Of course, the city of San Francisco must disburse 300,000 dollars (267,000 euros) to maintain the system for one year.


Source link