Ship path planning represents a fundamental challenge in intelligent navigation, requiring careful balance between route optimality, safety in complex marine environments. To address the limitations of conventional A* algorithms, this paper proposes an improved multi-factor and multi-scale A* algorithm. The methodology begins with processing ENC data, where canny edge detection combined with adaptive thresholding constructs obstacle maps. A novel dual-layer multi-scale grid framework is established: They are used to rapid global path searching, and precise collision avoidance. The algorithm innovatively integrates a multi-factor function that simultaneously considers obstacle distribution, environment effects, navigation rules, and ship dynamic constraints, with adaptive weight adjustment optimizing the search process. Path refinement employs smoothing algorithms to significantly reduce waypoint numbers. Simulation experiments conducted in Dalian port demonstrate the algorithm’s superior performance: maintaining safe clearance even in obstacle-dense areas and using the shorter length. Experimental results confirm that generated paths better satisfy practical navigation requirements.

Safe navigation of maritime autonomous surface ships (MASS) relies on two capabilities: path planning and collision avoidance. This review surveys classical algorithms and modern AI techniques for embedding the International Regulations for Preventing Collisions at Sea (COLREGs) into autonomous navigation. We organise prior work into three families—classical search/optimisation, real-time reactive methods, and learning-based approaches—and discuss their strengths and limitations with respect to rules compliance, computational cost, and onboard constraints. Building on these insights, we outline a large-language-model framework, Navigation-GPT, which couples reasoning-and-acting (ReAct) prompting with low-rank adaptation (LoRA). We further propose a three-phase deployment roadmap for MASS: core model integration, domain fine-tuning, and integrated operations. The paper concludes with open challenges and research directions toward reliable, explainable, and fully compliant MASS navigation.

Maritime safety faces growing challenges due to an expanding global fleet, tighter schedules, and increasingly complex stakeholder interactions. This study integrates multiple data sources to determine a more accurate representation of major marine accident causative factors in the United Kingdom. Logistic regression and data modelling are applied to Automatic Identification System data (2011–2017) and reported accidents from the Marine Accident Investigation Branch (2013–2019). Results show that larger vessels, daytime transits, service ships, winter conditions, and confined high-density areas such as ports impact accident likelihood. Interviews validate the data and emphasize the influence of port geometry and channel complexity. Among major UK ports, London, Plymouth and Milford Haven exhibit the highest accident-to-traffic densities. While maritime regulations and safety management systems in ports and vessels are seen as adequate by industry professionals, human factors require the greatest attention to improve maritime safety.

The paper explores the accuracy of WiFi-Round Trip Timing (RTT) positioning in indoor environments. Filtering techniques are applied to WiFi-RTT positioning in indoor environments, enhanced by Residual Signal Strength Indicator (RSSI)-based outlier detection. A Genetic and Grid filter are compared with a Particle filter and single-epoch least-squares across a range of test scenarios. In static scenarios, 67% of trials had sub-metre accuracy and 90.5% had a root mean square error (RMSE) below 2 m. In Non-Line-of-Sight (NLOS) conditions, 38% of trials had sub-metre accuracy, whereas for environments with full Line-of-Sight (LOS) conditions, 95.2% of trials had sub-metre accuracy. In scenarios with motion, 22.2% of trials had sub-metre accuracy. RSSI-based outlier detection in NLOS conditions, provided an average improvement of 41.3% over no outlier detection across all algorithms in the static and 14% in the dynamic tests. The Genetic filter achieved a mean improvement of 49.2% in the static and 47% in the dynamic tests compared with least squares.

Adapting Barker’s ((2019). The Journal of Navigation, 72(3), 539–554) taxonomy of wayfinding behaviours – originally developed for man-made environments, paper and screen – we examined which behaviours are also found in the outdoors. In the analysis of the collected data from a questionnaire (n=401), we find that participants employ every category in Barker’s framework of social, semantic and spatial behaviours. Our respondents report the use of digital maps on a mobile phone as the most common behaviour, with following directional signs as the second most used. Furthermore, social wayfinding behaviours figure prominently and the participants express preferences for various information sources. We demonstrate similarities of behaviours across the different types of environments and we confirm the applicability of Barker’s taxonomy of wayfinding behaviours also in nature. Our study generates knowledge that potentially can make navigation simpler and more efficient through wayfinding design, and lead to heightened feeling of safety in the outdoors. Wayfinding behaviour studies, like this one, can serve as a bridge between human psychology and practical design.