A dual-resolution acoustic-sensing robot for autonomous in-pipe inspection

This paper presents the development of an acoustic-sensing robotic system designed for autonomous in-pipe inspection. Existing methods, such as camera-based visual sensing and laser-based distance scanning systems, offer high-resolution and rapid spatial data but are often unsuitable for long-term deployment in adverse environments. Recent developments show that acoustic sensing methods can offer power- and data-efficient inspection for long-term service. This paper explores autonomous in-pipe inspection using only acoustic measures for remote pipe monitoring. We propose a dual-resolution acoustic sensing strategy that employs low-frequency acoustic waves for long-range coarse sensing and navigation, paired with high-frequency acoustic waves for short-range, high-resolution imaging. A robotic system and corresponding data characterisation and classification method was developed based on this strategy, enabling autonomous inspection. Experimental findings show that the robot can effectively localise multiple features within the pipe setup, achieving an average localisation variation of 8 cm over a total length of 1800 cm. Additionally, the system effectively classifies features, specifically aligned pipe structures, tilted pipe structures, blockages and empty pipes, with an average accuracy of 76% during autonomous inspections.