Optimizing time-delay estimation for underwater acoustic positioning in tank inspection robots to mitigate reflection interference

Abstract

Robot inspection of large storage tanks has become the primary method for addressing hidden risks in the petrochemical industry. Accurate positioning of Tank Inspection Robots (TIR) is crucial in this process. However, the unique characteristics of the liquid medium and tank walls introduce challenges such as noise, reverberations, and multipath effects, which frequently complicate the acoustic positioning of robots. These issues are further complicated by numerous interface reflections. To address the problems caused by reflections, this paper focuses on time-delay estimation for underwater acoustic positioning. It analyzes the impact of reflections on various acoustic positioning methods and simplifies the time-delay estimation process within the Time Difference of Arrival (TDOA) localization algorithm in reverberant environments. To this end, the paper introduces the AMCOCEP algorithm, which integrates Generalized Cross-Correlation, the cepstrum, and the Average Magnitude Difference Function (AMDF). The effectiveness of the proposed positioning model and method is validated by positioning a robot within a storage tank and comparing positioning accuracy before and after algorithm optimization. According to the experimental results, the AMCOCEP algorithm reduces the average error by 1.07 cm and improves the average accuracy by 30.7 %.