The University of Southampton

Students invent machine learning device to detect unsafe drinking water

Published: 16 April 2020
The Arwin device can detect and report changes in water quality in real-time

Student entrepreneurs from the University of Southampton have combined machine learning and spectroscopy to create a device that can alert owners to contaminated water.

Computer Science and Mathematics students Til Jordan and Andrius Matšenas are launching start-up company Arwin to bring their product to market and shield customers around the globe from water pollution.

The founders have won a University hackathon and a regional German science competition with early prototypes over the past six months and are preparing to raise investment for their final model.

The Internet of Things (IoT) device analyses water quality in real-time and then reports quality levels by request or at danger of contamination.

"There are 187 countries in the world that do not have drinkable tap water," Til explains. "Even in the USA, nearly a fifth of the whole population has consumed potentially unsafe drinking water at least once in the past 10 years.

"Using our device attached to the water flow under a sink, a change in the water quality can be detected as soon as it arises, and therefore prevent health problems by notifying the user in real time while also gathering data for an overview in the longer term."

Til and Andrius met at a University Centre for Machine Intelligence hackathon in November, which they won with a first proof of technology. This was followed by a new prototype with support from the University’s SEED Start-Up Fund and a first prize this March in the Munich Jugend Forscht competition.

The current Arwin prototype pumps a sample of liquid into a chamber where it is tested with different wavelengths of light. The device measures 18 levels which are calculated by a machine learning model and then displayed through an Android app. The entrepreneurs are close to finishing a pumpless model with additional iPhone and web apps.

"Before this winter's hackathon, I had the idea of combining spectroscopy with machine learning and tested it out with a self-built device to distinguish between different fruits," Til says. "The hackathon provided a platform for something bigger and we decided to tackle different liquids. By the end, we were able to recognise many translucent solutions like lemon and ink water with great accuracy.

"Over the past three months we’ve iterated through more prototypes and can now measure different mineral solutions - nitrates, nitrites and fluorides - that are otherwise indistinguishable by human senses."

Arwin has been shortlisted for an on-campus Dragons’ Den-style pitching competition, run by the University’s Future Worlds start-up accelerator, where the team hope to raise investment to take their business journey to the next level.

"We want to increase awareness of the cyclical changes of water quality, globally," Andrius says. "The true impact of innovation can be discovered using low-tech, accessible components and that's exactly what we’re here for. We are on our way to validating this technology to bring it to masses."

Articles that may also interest you

Share this article FacebookTwitterWeibo