What I Have in My Cup? A Liquid Identification Mechanism Based on Electrical Connectivity

摘要

Designing and Developing of Sensors to assess and evaluate the presents of different physical phenomena in the environments has heavily introduced in the past decade. Different methods and techniques have been studied to develop a direct micro-sensors. With the introduction of microcontrollers and machine learning, sensors data may be used to predict other indirect phenomena in the environment. In this work, the electrical conductivity is utilized to predict the type of a solution or liquid in a cup. A simple sensor electric circuit based on ATmega328 micrcontroller has been designed to detect the electrical conductivity values in different liquids. Subsequently, these values have been utilized to train an artificial neural network model (ANN) to predict the liquids commercial names. The designed circuit and the ANN model have been tested with different 15 liquids diluted with water. eleven liquids are drinkable, the other 4 are chemical cleaning liquids. Our simple ANN model with this tiny dataset and two features only can detect the liquid type with an accuracy of more than 80% and a MSE value of less than 1%.