Delivering results: Swindon Borough Council’s emerging tech team
Twelve months ago Swindon Borough Council made the decision to start an ‘Emerging Technologies Team’. The purpose of forming the team was to focus on identifying opportunities that benefit multiple council services, test technology (both existing and new), measure the results and report back.
It’s a new approach and one that’s delivering big results for the council.
“We fail or win fast, and focus on solving real business problems using new technologies. Sarah Talbot Head of Swindon’s Emerging Technology Team told attendees on a recent DL Week webinar.
“Focus is on small, not necessarily complete solutions that prove whether or not something is worth investment. It enables us to work at pace and to trial and evidence at a low cost,” she explained.
At Jadu, along with digital transformation partners Methods, we’ve been working closely with Swindon’s Emerging Technologies Team on a number of projects, including on the council’s ‘Report it’ functionality (which has been nominated in the Digital Leaders Award 2020 ‘AI Innovation of the Year’ category[ST1] .)
‘Report it’ functionality is commonly used in areas such as the reporting of graffiti, potholes, fly tipping and where citizens need to notify the council of something. By its nature, it has been reactive and relied on the public to fill out forms.
The team started experimenting to find another way.
“The question we asked was whether we could apply data science and Artificial Intelligence (AI) techniques to transform this common approach into something that was proactive,” said Jillur Quddus, Lead Data Scientist of Methods on the webinar.
“We started by training a neural network to identify potholes using images and live video streams. Once we’d trained this bespoke neural network, we embedded it into an Android mobile application, mounted a smartphone on an existing council vehicle and spent a day driving around Swindon,” he said.
The application detected potholes in real time. For every detection event, it captured an image and the location using the phone’s Global Positioning System (GPS). Integrating with Jadu CXM (Swindon’s case management system), cases for potholes were created in real time and a reactive process was transformed into a proactive one. Focus shifted to
automating detection of depth and width of potholes, in order to appropriately prioritise pothole cases.
As is the team’s continued objective, the approach was then applied across services that are seeing increasing channel shift, including fly tipping.
Swindon’s Street Smart team deals with 300[ST2] fly tipping cases each month and at points during the initial COVID-19 lockdown, that increased by 54 percent. The process had been that residents would report fly tipping, but important details would often be missing, which would be problematic when it came to knowing what equipment and vehicles were needed.
Now, accurate location details are registered and images that are uploaded go through image recognition, using a data set developed and used by the US army. As the US army transports a lot of goods, it has lots of reference data around the dimensions, weight, and volume[ST3] of objects. The image tool therefore knows how much items weigh based on average dimensions and makes estimates[ST4] , which aids the team in knowing van fills for most efficient collections. As the technology helps prioritise the fixing of potholes, so too does it help with the prioritisation of fly tipping. For example, if hypodermic needles are pictured, they’re going to have a higher priority than a mattress. This capability is now being prepared for use in the reporting of graffiti, where offensive images are prioritised.
The council has seen a 83 percent uplift in efficiency (by alleviating manual processes) during the proof of concept phase of the fly-tipping application. It’s gone from 10 percent of citizens submitting photos to 55 percent. Citizens have given a 98percent ‘positive[ST5] /good’ rating for the streamlined processes, and 83 percent have used the maps and geospatial tools to pinpoint locations.
Free School Meals and Robots
Another area of real success for the team has been the implementation of Robotic Process Automation (RPA), which emulates repetitive work, to automate its Free School Meals processes.
Work began before COVID-19 to improve applications. The process had previously required a parent/guardian to apply for free school meals via an online Jadu form and a person to manually check the eligibility of the application using the online government portal. Council staff would enter that into the Department for Education (DfE) portal and then enter the DfE results into the line of business application.
Having a person processing the information did not change the outcome as they were not involved in decision making, but often[ST6] just copying and pasting and/or re-entering data, which RPA can handle, running around nine times faster than human processing. This helped the council to scale during the pandemic when it saw a 2000 percent increase in Free School Meals applications.
Now, a parent/guardian applies online via a Jadu form, the application is then sent to a central mailbox area and database where Swindon’s robot workers go to the DfE and check the eligibility status of the applicant. If for any reason universal credits are not yet confirmed, the robots check every day for an updated status. They then go off and perform all the relevant validation checks, check the school information and create eligibility letters. Where there’s something that needs to be checked, it is handed over to a person to handle.
During the height of Covid, that 2000 percent increase would have resulted in 578 hours worth of work, but thanks to RPA, it took just 9.6 hours. That’s a 98.3% efficiency and 578 hours that human workers can add value by resolving the more complex inquiries.
It’s made it easier for parents/guardians to claim, accelerated request-through-to-fulfilment time, reduced the risk of digital privacy incidents, reduced resources for very low value copy and paste work, improved data quality and allowed the council to meet unprecedented demand. It’s also given the council a model for a digital workforce, which it can then apply to lots of processes.
The council continues to replicate success with its new ‘Report It’ functionality expected to benefit 30 other services for example. But the team is looking at a whole range of things, from traffic adapted lighting through to digital twinning, physical robotics, smart Personal Protective Equipment (PPE) wearables and other Covid related detection.
It’s fantastic to see the council take such an innovative approach and to be a part of it. It’s always great to see Swindon be so open in sharing its work on the likes of the Jadu Library (a repository for shared assets). This is exactly what local government needs, especially right now!
