UWB-Based Watermelon Sweetness Assessment With Smartphones for Everyday Shoppers

When making purchasing decisions based on fruit sweetness, everyday shoppers cannot rely on destructive or costly assessment methods. However, relative comparison among displayed items is often sufficient. To address this need, this paper presents a portable, smartphone-based assistance system that leverages the ultra-wideband (UWB) capabilities of modern smartphones. The system includes a compact UWB transponder that returns channel impulse response (CIR) data generated by the smartphone’s UWB ranging packets. A lightweight neural network on the smartphone processes this data and displays the result to the user. Our prototype, costing approximately $30 excluding the smartphone, achieves 88% accuracy in two-class classification of watermelon sweetness, and correctly selects either the sweetest or second-sweetest watermelon in 87% of cases from a group of five samples within the 10–13∘Bx range. Contrary to some prior assumptions, our results show that UWB signals from smartphones can effectively sense sweetness through the thick rind of watermelons.