8 November 2024 in IEEE Journal of Oceanic Engineering
In extreme environments, such as polar oceans, where potential hazards like sea ice are prevalent, deploying autonomous surface vessel (ASV) can enhance operational efficiency and safeguard personnel. As these extreme environments necessitate higher performance standards, particularly in terms of path-following accuracy and control stability, we introduce in this research an ASV path-following control method predicated on an enhanced proportional–integral–derivative (PID) parameters tuning algorithm aimed at reducing path-following errors and bolstering control stability. First, the adaptive line-of-sight (ALOS) guidance algorithm is devised to determine the desired ASV heading by designing the forward-looking range adjustment strategy. Second, the improved sparrow search algorithm (ISSA) is proposed for PID parameters tuning. Since the lack of stability of the standard Sparrow Search Algorithm (SSA), the producer update strategy is modified, and the Brown–Levy mutation strategy is designed to improve the global search ability of the algorithm. Finally, the virtual ASV simulation platform is built, and the real-time PID controller is constructed by designing the PID real-time tuning strategy. The parameters of the Nomoto ship motion model are fitted in the simulation platform according to different marine environments, and the PID controller parameters are updated in real-time by ISSA to improve the path following accuracy. Experimental results of the marine environment simulation test and the real-world experiment show that the ALOS guidance algorithm can effectively generate the current desired rudder angle. The PID controller based on ISSA has the best performance in computer simulation. The average overshoot is 2.79%, and the average convergence time is 20.1 s. In the real-world experiment, the average path following error of ISSA Real-Time is reduced by 51.0% compared with that of SSA and 27.2% compared with that of ISSA. The improved control method can better satisfy the control requirements of the ASV, enhance control stability, and achieve more precise path following.
28 February 2025 in IEEE Sensors Journal
Ocean sensor networks (OSNs) can sense long time series data, which plays a crucial role in tasks that rely on massive sensory data such as ocean ecosystem monitoring and maritime regulation. Unfortunately, ocean nodes are prone to failures, and harsh ocean channels, and natural environments can interfere with the data collection and transmission process, which can lead to large-scale loss of ocean sensory data. Based upon the comprehensive and in-depth analysis of the ocean’s time-varying dynamic environment, we proposed a Multi-head ProbSparse Self-attention Mechanism based Model (MPSM) to predict and reconstruct missing ocean data. Specifically, MPSM can learn key features of ocean observation data on both sides of missing gaps with a dual encoder architecture. Subsequently, the multi-head probsparse self-attention mechanism can efficiently capture inter-sequence dependencies in parallel with lower computational complexity. Moreover, its residual connection will enhance model’s expressive ability and alleviate gradient vanishing. Simulation experiment results demonstrate that metrics of MPSM are lower than the selected benchmark algorithms on different ocean datasets, which means MPSM has a stronger robust and generalizability. MPSM can provide high-precision, low-latency missing data restoration capabilities for marine monitoring scenarios. It enhances the continuity and reliability of ocean observation systems.
1 October 2026 in IEEE Internet of Things Journal
29 July 2026 in SSRN Electronic Journal
Accurate vessel trajectory prediction is essential for navigational safety, maritime traffic management, and port operations. Existing methods mainly model Automatic Identification System (AIS) data alone, while multimodal approaches often rely on fixed fusion strategies that cannot adapt to changing marine environments and remain vulnerable to error accumulation over long prediction horizons. This study proposes SCANet, a scene-conditioned attention network for long-horizon multimodal vessel trajectory prediction. A dual-branch encoder extracts AIS motion and environmental representations, while a soft scene prior guides cross-modal multi-head attention and adaptive gated fusion. A dual-context decoder then retrieves relevant historical motion states and environmental cues during autoregressive prediction. Experiments use AIS, coastal weather-station, and ERA5 data from Los Angeles in summer and New York in winter. The two datasets contain 2,565 and 771 vessels and 10.44 and 6.27 million trajectory samples, respectively. SCANet achieves the lowest prediction error among the evaluated models across horizons of 1–6 h and exhibits more stable error growth as the horizon increases. Ablation studies confirm the complementary contributions of scene-conditioned fusion and dual-context decoding. Cross-region experiments further demonstrate that SCANet can efficiently adapt to geographically and environmentally distinct maritime regions through limited fine-tuning.
17 June 2026 in SSRN Electronic Journal
With the rapid growth of cruise business, modern cruises are becoming more complex and intelligent, raising safety demands for indoor personnel localization. However, the metal-rich structure and variable environment cause multipath effects and noise, leading to low localization accuracy. To address this, this paper proposes an indoor localization system to enhance localization accuracy. First, the impact of environmental factors on the received signal strength indicator (RSSI) is analyzed to determine the input for network model. Then, the distributed cluster optimization learning (DCOL) model is proposed based on a distributed network architecture for distance estimation, which can balance the communication and calculation of the anchor nodes and improve the distance estimation accuracy. Finally, statistical pruning principle is introduced to design the minimum error localization (MEL) algorithm. The number of anchor nodes to be retained is determined according to the statistical results to improve the localization accuracy. Through the actual deployment on the experimental ship "Wusong" of Shanghai Maritime University, the experimental results show that with the node density is 0.11/m2, the localization error is less than 0.87m in 80% cases. Compared with the baseline method, the localization following effect is improved by 55%.