Improving Object Localization Resolution using a Wearable Ultrasonic Sensor Array

Technologies for navigation, obstacle detection, and object location have greatly benefited from the growth in wearable sensors and connected devices. Ultrasonic sensors are a common sensing modality used for the purpose of identifying the presence of an object and the distance at which they are located. However, to improve real-time obstacle and object detection systems that make use of ultrasonic sensors, approaches to increase the directionality and distance resolution are needed. This paper presents a new 3-ultrasonic sensor array design and experimentally analyzes several location models to define an optimum computational approach for the detection of small objects using this new 3-ultrasonic sensor array. Model optimization was performed considering computational complexity and object detection resolution. Results show that the optimum model achieves an accuracy of 99.2% when identifying objects at seven different location zones, five within 18 inches from the sensors, one within 18 inches and 4 feet from the sensors, and another one outside of a range of 4 feet. The use of this ultrasonic array and object detection model could help improve systems that provide real-time feedback, for example to visually impaired individuals, by providing information that accurately narrows down the position of a detected object.