U-SLAM: Ultrasonic Simultaneous Localization and Mapping for Underwater Pipeline Defect Inspection

This paper presents an Ultrasonic Simultaneous Localization and Mapping (U-SLAM) system for underwater pipeline inspection. Addressing the limitations of existing SLAM technologies in water environments, including visual degradation, laser absorption, and bulky sonar equipment, we developed the compact U-SLAM system based on a designed ultrasonic radar formed by high-frequency ultrasonic probe arrays. The system features a miniaturized design (diameter = 13 cm, weight = 465 g) that achieves 1.8° high-precision 360° scanning via a symmetric four-probe configuration, with a dedicated SLAM algorithm. The algorithm combines multi-sensor fusion-based coarse localization, linear-stage-controlled fine-motion positioning, and point-cloud loop-closure constraints, enabling millimeter-level positioning accuracy and highly real-time point cloud modeling. We derived and simulated the model of the ultrasonic transducer in static and flowing water, designed the ultrasonic radar and SLAM algorithm, and analyzed the SLAM error. Finally, we developed a radar prototype and performed experiments. Results in simulated pipeline environments (50cm diameter) showed the system’s capability to reconstruct non-small-area structures (e.g., flanges, sediment deposits), achieving a mean ultrasonic ranging error of 1.14mm. Further round-trip scanning in a realistic pipeline with an integrated robot confirmed that the U-SLAM system reliably performs localization and 3D mapping, producing structurally consistent point-cloud models. While challenges remain in modeling small-area fine features like cracks, U-SLAM’s real-time performance, miniaturization, and environmental adaptability provide an innovative technical solution for underwater inspection robots.