An Adaptive Spatial Positioning Algorithm for Ultrasonic-Only Parking Slot Detection System

Abstract

In the field of automated parking system, achieving precise detection of parking slot edge only through ultrasonic sensors remains the fundamental challenge for implementing an ultrasonic-only parking system. In this work, an adaptive spatial positioning (ASP) algorithm is proposed utilizing a single ultrasonic sensor based on the principle of multiple reflections of ultrasonic waves at the same obstacle edge, enabling precise edge detection. The edge detection results from ipsilateral ultrasonic sensors are subsequently fused through a decision tree algorithm, followed by noise reduction via clustering methods to eliminate discrete outliers, ultimately achieving accurate parking space detection. The proposed ASP algorithm demonstrates superior computational efficiency, exhibiting a memory footprint of 420 bytes and an execution time of $44~\mu $ s. Ultimately, the ultrasonic-only slot detection system is tested in real-world scenarios, demonstrating an 87.5% obstacle (vehicle) edge correction rate, 86.4% bilateral slot between vehicles (100% for slot between vehicle and pillar) detection rate, and 100% unilateral slot detection rate, with an average positioning error of <25 cm. It demonstrates its applicability in an ultrasonic-only parking system.