From Machinery to Biology: A Review on Mapless Autonomous Underwater Navigation

摘要

Autonomous navigation in unknown; map-free environments is a core requirement for advanced robotics. While significant breakthroughs have been achieved in terrestrial scenarios, extending this capability to the unstructured, dynamic, and harsh underwater domain remains an enormous challenge. This review comprehensively analyzes the mainstream technologies underpinning mapless autonomous underwater navigation, with a primary focus on conventional Autonomous Underwater Vehicles (AUVs). It systematically examines key technical pillars of AUV navigation, including Dead Reckoning and Simultaneous Localization and Mapping (SLAM). Furthermore, inspired by the emerging concept of fourth-generation submersibles—which leverage living organisms rather than conventional machinery—this review expands its scope to include live fish as potential controlled platforms for underwater navigation. It first dissects the sophisticated sensory systems and hierarchical navigational strategies that enable aquatic animals to thrive in complex underwater habitats. Subsequently, it categorizes and evaluates state-of-the-art methods for controlling live fish via Brain-Computer Interfaces (BCIs), proposing a three-stage control hierarchy: Direct Motor Control, Semi-Autonomous Control with Task-Level Commands, and Autonomous Control by Biological Intelligence. Finally, the review summarizes current limitations in both conventional AUV technologies and bio-hybrid systems and outlines future directions, such as integrating external sensors with fish, developing onboard AI for adaptive control, and constructing bio-hybrid swarms. This work bridges the gap between robotic engineering and biological inspiration, providing a holistic reference for advancing mapless autonomous underwater navigation.