Material classification via embedded RF antenna array and machine learning for intelligent mobile robots

Abstract

In this work, we present a novel design for an embedded Radio Frequency (RF) antenna array that can distinguish various materials by analyzing changes in Received Signal Strength Indicator (RSSI) values. The use of a low-cost and small-form-factor microcontroller by Espressif makes this design both cost-effective and suitable for integration into various applications, differentiating it from previous studies. To enhance the material classification performance, a combination of Kalman filter and Support Vector Machine is proposed which does not require a large amount of training data for model optimization. Results demonstrate that the proposed machine learning model is able to perform material classification within a 2 m range, with an average accuracy of over 96%. Such a system is well-suited for intelligent mobile robotic applications particularly in warehouse automation or smart manufacturing lines due to its ability for proximal remote sensing, real-time monitoring, and multimodal sensing.