Soft grippers utilize compliant materials to achieve adaptable grasping, yet they often face challenges in accommodating objects with widely varying dimensions due to their fixed kinematic structures. This paper presents the design, fabrication, modeling, and control of a novel soft gripper featuring a rigid-flexible coupled variable range adapter. By integrating a motorized crank-slider mechanism with soft pneumatic fingers, the gripper achieves a dynamic volumetric workspace capable of manipulating objects ranging from compact to large geometries. Theoretical modeling and experimental characterization reveal that the adapter serves a dual purpose: it not only expands the effective workspace but also functions as a mechanical force amplifier, capable of exponentially boosting the contact force through kinematic reconfiguration. Furthermore, an intuitive Human-in-the-Loop (HITL) teleoperation strategy is established using wearable flex sensors. This control framework maps human gestures to robotic actuation, leveraging human visual feedback as a high-level perception loop to validate the open-loop response of the soft actuators. Experimental results demonstrate that this integrated system significantly improves adaptability and payload stability for diverse object geometries compared to fixed-base counterparts.

Achieving autonomous navigation requires Maritime Autonomous Surface Ships (MASS) to overcome difficulties in recognizing intricate multi-ship encounter situations and developing appropriate collision avoidance strategies. To address the above issues, this study first integrates the Velocity Obstacle (VO) with Closest Point of Approach/Time to Closest Point of Approach (CPA/TCPA) into the collision risk identification of ships. This enables the own ship (OS) to determine the risk posed by target ships (TSs) in the scenario. Secondly, the paper further discusses complex multi-ship encounter scenarios, based on 1972 International Regulations for Preventing Collision at Sea (COLREGs), it classifies responsibilities to determine the set of give-way ship and stand-on ship obligations. Finally, the derived responsibility set is incorporated into Dynamic Window Approach (DWA), allowing the enhanced Multi-Vessels Velocity Obstacle and Improved Dynamic Window Approach (MVO-IDWA) algorithm to automatically select optimal decisions in complex multi-ship encounter scenarios and ensure the ship reaches its destination safely. Centering on the challenge of avoiding collision decision in multi-ship complex encounter scenarios, this paper proposes the integration of ship responsibility sets with the MVO-IDWA algorithm. Analysis of the results establishes that the proposed algorithm can consider the inter-ship responsibilities, risk levels, and movement trends of TSs in multi-ship encounters, thereby realizing autonomous ship collision avoidance.

This survey provides a comprehensive synthesis of methods, datasets, metrics, and deployment strategies from the evolution of convolutional neural network (CNN)-based detectors to emerging transformer and hybrid architectures. It unifies fragmented literature into a structured taxonomy while integrating results from 2014–2025 studies. The paper reviews benchmark datasets, discusses evaluation protocols and reproducibility standards, and proposes a deployment playbook considering latency, energy, and hardware constraints. Beyond technical performance, it addresses responsible AI practices and ethical challenges in marine observation. By highlighting open problems in multimodal fusion, self-supervised learning, and on-device adaptation, this work aims to guide future research and practical deployment of underwater vision systems. A comprehensive survey of underwater object detection covering classic CNN-based detectors, modern transformer and hybrid models, training and evaluation practices under challenging aquatic conditions, the dataset landscape, deployment constraints (latency/VRAM/energy), and open problems for real-world marine applications.