Object Shape Classification Using 4×4 Ultrasonic Sensor Array

This research paper presents a comprehensive study on an ultrasonic sensor-based system designed for real-time object shape recognition using machine learning (ML) algorithms grounded in echolocation principles. Echolocation, a sound-based sensory system employed by animals such as bats and dolphins, enables precise identification of object shape, location and size through the interpretation of emitted and received echoes. This paper details the design and implementation of the sensor array system and evaluates the performance of various deep learning and machine learning algorithms for this application. The machine learning algorithm XGBoost classified object shapes with an accuracy of 99.1%.