AquaCat: Using Radar and Machine Learning for water pollutant detection
The AquaCat project makes use of low-cost miniaturized radar technology and machine learning in order to reliably detect water pollutants. It is a collaboration between the University of St Andrews in Scotland and the Universidade Federal de Goiás in Brazil. The project builds on RadarCat, a radar-based object classification system developed by the University of St Andrews School of Computing’s Human-Computer Interaction Group (SACHI) [1,2].
Brazil faces unprecedented problems caused by water pollution. Untreated sewage, industrial chemicals, and agricultural waste have entered the waterways introducing pollutants dangerous to both humans and the environment. For example, dangerous levels of Thallium, Arsenic and Zinc have been found in the Pardo River in São Paulo .
Current sampling and analysis techniques, such as inductively coupled plasma mass spectrometry (ICPMS), require expensive specialist equipment, reducing the possible scale of measurement, both in time and in space. We are addressing this issue by producing a small, low-cost and portable pollutant detector based on miniaturized radar technology.
Our long-term plan is to allow environmental researchers to make cheaper, faster and thus more numerous measurements in the field and therefore create a richer data set. When miniaturised radar technologies are deployed into smartphones and other devices, AquaCat could be available to a much larger section of the population. This will enable the collection of crowd-sourced pollution data over a large geographic area and over a long time. By having a higher number of samples over a longer period, policymakers will gain a better understanding of the distribution of pollutants and their sources, allowing them to better target regulatory interventions and infrastructure improvements.
 Object Recognition with the Project Soli in St Andrews: https://sachi.cs.st-andrews.ac.uk/2016/05/projectsoliatstandrews/
 Hui Y, Flamich G., Schrempf P, Harris-Birtill D., Quigley A. RadarCat: Radar Categorization for Input & Interaction, submitted to ACM UIST 2016, Tokyo Japan (under review)
 Alves, Renato IS, et al. “Metal concentrations in surface water and sediments from Pardo River, Brazil: human health risks.” Environmental research 133 (2014): 149155.