Аңдатпа
This work presents a low-cost acoustic sensing framework for underwater localization and perception using single-beam sonars (SBS) and a forward-looking sonar (FLS). A Kalman filter (KF) based sensor fusion approach is developed to integrate SBS measurements with inertial sensing, incorporating pre-filtering and confidence-based gating to mitigate acoustic outliers and reflective disturbances. Multiple motion configurations, including manual underwater robot guiding followed by thruster-propelled controlled motion are employed to evaluate robustness under controlled and non-ideal conditions. Optical flow and visual tracking are used as independent references for localization validation and perception. Experimental results demonstrate significant drift reduction and improved localization compared with inertial sensing. When compared to the ground truth, SBS-assisted fusion results in an average localization error of 17% per 100 𝑚 displacement. The framework further supports parent–child localization using FLS tracking, enabling relative pose estimation in multi vehicleconfigurations, with a localization error of 5% per 100 𝑚 displacement. Improved localization coupled with image perception was explored for planar scene reconstruction using mosaicking. The proposed system therefore highlights the feasibility of using low-cost sonar driven localization for underwater inspection and robotic navigation for small scale underwater robots.