You are viewing 1 of 2 articles without an email address.


All our articles are free to read, but complete your details for free access to full site!

Already a Member?
Login Join us now

Las Vegas cleans up with machine learning

Trash littering city streets, parks and public places looms large in the minds of mayors and city leaders everywhere. Here’s how cities such as Las Vegas are using technology to solve the problem

LinkedInTwitterFacebook
 Las Vegas has developed a high-tech approach to cleaning up
Las Vegas has developed a high-tech approach to cleaning up

In 2017, New York City set aside $32 million to reduce its burgeoning rat population, highlighting the scale of the problem. A key part of the solution was better trash removal, including investing in 336 new solar-powered waste compactor bins and more frequent trash collections by city workers, as well as stricter enforcement and increased fines for illegal dumping by private businesses.

 

Today’s newest trash removal approaches are high tech and focused on removing trash quickly to reduce eyesores and prevent rats and disease.

 

Las Vegas’ home-grown machine learning trash pickup

 

One example is Las Vegas, Nevada. The city doesn’t have major concerns with rats like much larger New York, but the gambling mecca does want to keep parks and public areas clean of litter and graffiti – especially in its downtown improvement district, separate from the Vegas Strip.

 

Las Vegas has developed a high-tech approach to keeping its downtown parks clean. An adaptive learning system uses video cameras and artificial intelligence (AI) to detect trash, graffiti or other objects that the city has flagged within the AI programme. This enables clean-up crews to be sent out on demand, which is cheaper than sending them out on a regular schedule.

 

The Las Vegas system has already worked successfully in two parks near the downtown area and may soon be adapted for other parks and city roadways, according to Michael Lee Sherwood, Director of Information Technology for Las Vegas.

 

Sherwood explained how it works: Video captured from cameras already installed in the parks for security purposes streams in real-time to a cloud-based AI service in Microsoft Azure. The AI service, designed by the city, analyses the video data to make a determination, based on what the AI system has been trained to spot. If a park is judged dirty, a work order is generated to clean it.

 

Many AI systems rely on machine learning. Usually, in simple terms, a computer is shown repeated images of a scene or an object in a so-called normal state. When a computer sees that same image in aberrant state, it can be programmed to provide an alert. In Vegas, the appearance of trash is deemed aberrant.

 

In another example of AI, police in the US states of Utah and Arizona are using thermal video images of cars driving the wrong way down a highway to send automatic, real-time alerts to prevent serious wrecks and fatalities.

 

Other cities, including Moscow’s state-run clinics, have successfully used images of healthy lungs to compare to images of lungs of people suspected of lung cancer. Montreal has experimented with using AI to find sick street trees, while other AI researchers there hope to use high-resolution images to detect when bridge spans are sagging, even slightly, over time.

 

On-demand trash pickup for greater efficiency

 

“We are using this new IoT system to conserve human resources to only be expended when the area or roadway actually needs cleaning,” Sherwood said. “We believe this will lead to higher service levels and experiences for residences, while creating operational efficiencies for our city.”

 

“We’re expanding the project now to roadways and illegal dumping,” he added. “There are a lot of practical applications.”

 

In developing the project, Vegas city staff relied on generally available, less expensive software tools and already-installed video cameras. The cameras aren’t the most expensive high-resolution models that are equipped for facial recognition. They don’t store data directly; they send all the collected data over wireless to a server for analysis. The city hopes to patent its AI learning approach.

 

Adoption and adaption

 

“It’s very inexpensive,” Sherwood asserted. “The technology works.” While he assumes there will be “large cost savings over time,” the city hasn’t yet publicised its savings.

 

Although technology works well in calling clean-up crews to the scene when litter accumulates, it isn’t always easy to persuade workers to perform clean-ups on an ad hoc basis.

 

“The system has to get the human side to adapt,” Sherwood said. “With the older park schedule, workers know they are cleaning today, but with on-demand services, you need user adoption of the process.”

 

One clear obstacle in using video data could be worries about privacy, with cameras often sparking concerns about surveillance.

 

“There are no privacy concerns because the system is not looking for people,” Sherwood said. “Humans are bigger objects than trash. But even if it is a human moving around who dumps a mattress or leaves litter all over the place, we’re not looking at those aspects of human behaviour, but the static image of trash.”

 

Other cities grappling with public litter and trash are adapting similar concepts to Vegas. For example, last year in Los Angeles, sanitation crews began using dashboard cameras to identify trouble spots along its 22,000 miles of road. The findings and clean-up response are automated with machine learning.

 

However, the use of video cameras to tackle issues such as these is still an emerging area.

 

Smart trash bins

 

Many cities globally have also relied on smart trash compactor bins, such as those from Bigbelly. Some of the bins that Bigbelly makes are powered by solar and wireless communications, offering sensors to detect when a bin is full and sending a signal to have a work crew arrive to empty the bin. The company advertises reductions in collections of up to 80%.

 

The bins are often sold for use in cities and on college campuses where pedestrians gather. Although the bins don’t detect trash strewn along parks and highways in the way the Vegas application works, city workers have still sometimes had to adapt to on-demand work schedules.

 

“Bigbelly has created some issues staff [in various cities], as their routes became so much more efficient,” said analyst Alison Brooks at IDC. “This issue has been circumvented by having them work on other tasks. The workers still have regular schedules, but they are dynamically configured as opposed to relying on set routes.”

 

Las Vegas is not using these bins but has studied the technology. “We see opportunities for all types of technology to play a role in automation and operational efficiency,” Sherwood said.

 

“There is no one magic solution, but rather a concert of sensors and technologies working together to solve challenges. “

 

“Artificial intelligence and machine learning offer a great opportunity for the future,” he added. “We are already seeing substantial results.”

LinkedInTwitterFacebook
Add New Comment
LoginRegister