The recent advancements in underwater robotics and localization techniques have significantly enhanced the capabilities of autonomous underwater vehicles (AUVs). Innovations in data-driven frameworks for Doppler velocity logs (DVL) calibration have led to substantial improvements in accuracy and calibration time, enabling AUVs to operate with higher precision and efficiency. Additionally, the integration of advanced computer vision, sonar, and acoustics hardware in AUV design has resulted in more robust and reliable systems, capable of complex underwater tasks. Neural network-based approaches, such as the Electric Field Inversion-Localization Network (EFILN) and Physics-informed neural networks (PINNs), have shown remarkable performance in high-precision underwater positioning and 3D localization, respectively. These developments not only enhance the operational capabilities of AUVs but also open new avenues for research and application in marine robotics.

Noteworthy papers include:

• DCNet: Introduces a data-driven framework for rapid DVL calibration, significantly improving accuracy and time efficiency.
• EFILN: Proposes a neural network for high-precision underwater positioning, demonstrating robustness and strong small-sample learning capabilities.