Swindon Improves Response to Illegal Waste Disposal with Artificial Intelligence

The existing system allowed people to report fly-tipping through the council website, provided they had an account. But it was an unstructured form, which made it difficult for people to provide the right information required to clear up the rubbish. The reports were then printed out and handed to the clear-up team. With more than 300 cases a month, this was a manual and time-consuming drain on resources.

The council has a small DevOps capability and a strong desire to innovate. It created an Emerging Technology team to work differently and to engage with the private sector. The team engaged with Amazon Web Services (AWS) and two AWS Partners to collaborate on designing a better system to deal with the fly-tipping problem.

The entire process took less than 3 months—from initial conversations to a working service—with a delivery team of just four people.

The new system still uses the website but no longer requires a login. Based on feedback from residents and staff, the site features a map so users can mark precisely where the rubbish is and submit photos of what has been dumped. The system can automatically analyse the images and allocate the right resources to collect it—a large truck, a smaller vehicle, or something more specialised.

The image analysis system helps the council prioritise its workload. For example, items that need to be immediately disposed of such as drugs paraphernalia or hazardous materials are cleared first.

The clear-up teams are equipped with tablets that allow them to make their own reports and to close cases on the go. The tablet also provides a map with an optimised route for the team to reduce mileage and pollution and avoid overloading the vehicles. The devices are far more efficient, user friendly, and have already delivered savings in printing costs.

Swindon saves about £3,000 a year on fuel costs with a reduction in associated emissions. The new system has also saved the council 2,137 staff hours on the ground. Previously, the teams had nothing but written reports to allocate resources and plan the best route.

For the residents of Swindon, the new system means average clean-up times have fallen from over 10 days to just 4. And 98 percent of those surveyed were pleased with the new system.

The new system gets precise location data with almost every report—83 percent of fly-tipping sites are marked on the online map. Previously, the collection teams were often working from a street name so regularly spent additional time searching for rubbish locations.

The challenge was made worse by COVID-19 but the team pressed on with testing the new technology in a socially distanced manner. This determination paid off. With recycling centres closed, Swindon saw a 54 percent increase in fly-tipping during the first lockdown.

The council worked with AWS and AWS Consulting Partner Methods to co-produce the image analysis and reporting system. The extended team delivered a working service in less than three months.

“AWS was really engaging, easy and open to work with, and seemed to really care about the people of Swindon and the potential social good this project could do,” says Swindon Council emerging technologies lead, Sarah Talbot.

Staff say the artificial intelligence-powered system has transformed the way they work. Council teams can quickly update systems once the site is cleared. Big data analysis can also help identify fly-tipping trends and dumping hot spots to guide enforcement and help reduce problems.

Today, the clear-up teams are working more efficiently despite the increased demands on their time, and council staff are no longer tied up with printing daily reports.
Because the system is modular, it can quite easily be modified for other report-it services and is already being used for tracking down abandoned shopping trolleys.
The system uses Amazon API Gateway to receive data from the council’s customer relationship management system. AWS Lambda processes that data into the Amazon Relational Database Service (RDS) and moves the images into Amazon Simple Storage Service (Amazon S3), and Amazon Rekognition runs the analysis. Finally, Amazon DynamoDB acts as a backend to bring this information together with GIS information to create the actual map.

The project has also helped change attitudes within the council. "The broader council team appreciates the value of the work the Emerging Tech team provides," says Talbot. “They can see the value of the work, and that our focus and drivers are around helping with real issues in tangible, new ways.”

Talbot credited the help of the clear-up team and their openness to new ways of working for the project’s success. Swindon Borough Council is keen to share its technology with other cities, towns, or borough councils, as they are part of the Local Digital Declaration. “All councils, regardless of whether they’re in the UK or further afield, have the same issues—we all have to deal with graffiti, fly-tipping, potholes, and with decreasing budgets,” explains Talbot.

“Sharing work, particularly work that can be deployed quickly at low cost, means we can tangibly help each other. That’s something my team and Swindon are passionate about. It’s powerful, and it’s important.”