Li Wang

With the growing applications of maritime unmanned systems in environmental monitoring, ocean patrol, and emergency response, achieving efficient multi-platform cooperation in complex and dynamic marine environments remains a critical challenge. Unmanned Aerial Vehicles (UAVs) provide flexible and high-coverage sensing capabilities but are constrained by limited energy capacity, whereas Unmanned Surface Vehicles (USVs) offer long endurance and can serve as mobile platforms and energy supply nodes. Existing studies mostly focus on single-factor optimization, lacking a systematic analysis of the coupled relationships among energy, information (communication and positioning), and task decision making. To address this problem, this paper proposes an Energy–Information–Decision Coupling Optimization Method for Cooperative Maritime Unmanned Systems. A unified coupling model is established to integrate task completion, energy consumption, communication delay, and replenishment scheduling into a multi-objective optimization framework. A bi-level optimization algorithm is designed: the upper layer optimizes USV trajectories and energy supply strategies, while the lower layer optimizes UAV path planning and task allocation. A closed-loop adaptive mechanism is incorporated to achieve optimal cooperation under dynamic tasks and energy constraints. Extensive simulations combined with real-world experimental data are conducted to evaluate the method in terms of mission efficiency, energy balance, communication latency, and system robustness, with ablation studies quantifying the contribution of the coupling module. Results demonstrate that the proposed method significantly outperforms non-coupled or single-factor optimization strategies across multiple performance metrics: it achieves a task completion rate exceeding 93%, reduces total energy consumption by approximately 6% and replenishes waiting latency by over 28% compared with the decoupled baseline method. This effectively enhances the cooperative efficiency and robustness of maritime unmanned systems, and provides theoretical and methodological guidance for large-scale, complex ocean missions.

Uneven terrains and complex obstacles in off-road environments present significant challenges to the stability and safety of vehicle path planning. This paper presents a hierarchical DRL-DWA path planning framework for unmanned ground vehicles (UGVs). At the global level, an energy-aware D* Lite algorithm generates cost-efficient waypoints considering both distance and energy consumption. At the local level, a deep reinforcement learning-enhanced DWA controller adaptively adjusts the weighting factors of evaluation functions in real time to ensure dynamic feasibility on rough terrain. The parameter selection is formulated as a Markov decision process (MDP), where a novel reward function based on elevation maps, vehicle pose, goal, and obstacle information guides the optimization for off-road navigation. Furthermore, the single UGV framework is extended to a multi-UGV system, where formation control is achieved through the leader–follower strategy. To evaluate the performance of our algorithm, we conduct experiments in 3D simulation environments featuring various terrains and obstacles. The results indicate that the proposed approach outperforms existing path planning techniques, showing a higher success rate and a lower average elevation gradient in uneven terrains.

The Transformer-based target detection model, DETR, has powerful feature extraction and recognition capabilities, but its high computational and storage requirements limit its deployment on resource-constrained devices. To solve this problem, we first replace the ResNet-50 backbone network in DETR with Swin-T, which realizes the unification of the backbone network with the Transformer encoder and decoder under the same Transformer processing paradigm. On this basis, we propose a quantized inference scheme based entirely on integers, which effectively serves as a data compression method for reducing memory occupation and computational complexity. Unlike previous approaches that only quantize the linear layer of DETR, we further apply integer approximation to all non-linear operational layers (e.g., Sigmoid, Softmax, LayerNorm, GELU), thus realizing the execution of the entire inference process in the integer domain. Experimental results show that our method reduces the computation and storage to 6.3% and 25% of the original model, respectively, while the average accuracy decreases by only 1.1%, which validates the effectiveness of the method as an efficient and hardware-friendly solution for target detection.

In an intelligent driving environment, monitoring the physiological state of drivers is crucial for ensuring driving safety. This paper proposes a method for monitoring and analyzing driver physiological characteristics by combining electronic vehicle identification (EVI) with multimodal biometric recognition. The method aims to efficiently monitor the driver’s heart rate, breathing frequency, emotional state, and fatigue level, providing real-time feedback to intelligent driving systems to enhance driving safety. First, considering the precision, adaptability, and real-time capabilities of current physiological signal monitoring devices, an intelligent cushion integrating MEMSs (Micro-Electro-Mechanical Systems) and optical sensors is designed. This cushion collects heart rate and breathing frequency data in real time without disrupting the driver, while an electrodermal activity monitoring system captures electromyography data. The sensor layout is optimized to accommodate various driving postures, ensuring accurate data collection. The EVI system assigns a unique identifier to each vehicle, linking it to the physiological data of different drivers. By combining the driver physiological data with the vehicle’s operational environment data, a comprehensive multi-source data fusion system is established for a driving state evaluation. Secondly, a deep learning model is employed to analyze physiological signals, specifically combining Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks. The CNN extracts spatial features from the input signals, while the LSTM processes time-series data to capture the temporal characteristics. This combined model effectively identifies and analyzes the driver’s physiological state, enabling timely anomaly detection. The method was validated through real-vehicle tests involving multiple drivers, where extensive physiological and driving behavior data were collected. Experimental results show that the proposed method significantly enhances the accuracy and real-time performance of physiological state monitoring. These findings highlight the effectiveness of combining EVI with multimodal biometric recognition, offering a reliable means for assessing driver states in intelligent driving systems. Furthermore, the results emphasize the importance of personalizing adjustments based on individual driver differences for more effective monitoring.

A critical component of bus network adjustment is the accurate prediction of potential risks, such as the likelihood of complaints from passengers. Traditional simulation methods, however, face limitations in identifying passengers and understanding how their travel patterns may change. To address this issue, a pre-evaluation method has been developed, leveraging the spatial distribution of bus networks and the spatio-temporal behavior of passengers. The method includes stage of travel demand analysis, accessible path set calculation, passenger assignment, and evaluation of key indicators. First, we explore the actual passengers’ origin and destination (OD) stop from bus card (or passenger Code) payment data and biometric recognition data, with the OD as one of the main input parameters. Second, a digital bus network model is constructed to represent the logical and spatial relationships between routes and stops. Upon inputting bus line adjustment parameters, these relationships allow for the precise and automatic identification of the affected areas, as well as the calculation of accessible paths of each OD pair. Third, the factors influencing passengers’ path selection are analyzed, and a predictive model is built to estimate post-adjustment path choices. A genetic algorithm is employed to optimize the model’s weights. Finally, various metrics, such as changes in travel routes and ride times, are analyzed by integrating passenger profiles. The proposed method was tested on the case of the Guangzhou 543 route adjustment. Results show that the accuracy of the number of predicted trips after adjustment is 89.6%, and the predicted flow of each associated bus line is also consistent with the actual situation. The main reason for the error is that the path selection has a certain level of irrationality, which stems from the fact that the proportion of passengers who choose the minimum cost path for direct travel is about 65%, while the proportion of one-transfer passengers is only about 50%. Overall, the proposed algorithm can quantitatively analyze the impact of rigid travel groups, occasional travel groups, elderly groups, and other groups that are prone to making complaints in response to bus line adjustment.

Owing to the attractive properties of decentralization, unforgeability, transparency, and traceability, blockchain is increasingly being used in various scenarios such as supply chain and public services, where massive Spatial-Temporal-Keywords (STK) transactions need to be packaged to the blockchain systems. However, due to the multi-dimensionality of STK transactions and transaction randomness among different blocks, existing solutions fail to enable queries in a verifiable and efficient way for blockchains storing multidimensional transactions. To this end, this paper takes the first step to innovatively propose an authenticated and efficient query approach in hybrid blockchain systems consisting of on-chain and off-chain parts. We first design a new data structure named MRK-Tree in the block body, which organizes STK transactions for efficient nodes pruning of both kNN and range queries. Then we propose an improved block header, which improves the efficient pruning of blocks on the basis of ensuring the authentication of query results. Also, we design a cross-block searching algorithm named Efficient Block Pruning (EBP) and intra-block searching algorithms named Authenticated kNN/Range Query (AKQ/ARQ) to accelerate authenticated queries for multiple MRK-Trees in the hybrid blockchain systems. Authentication mechanisms are proposed to ensure the soundness and completeness of query results. Rigorous security analysis validates the practicability of the proposed approach. We build a blockchain prototype to comprehensively evaluate the performance of proposed query schemes. Extensive evaluation results with real datasets reveal that our approach can ensure authenticated queries, meanwhile improving the time efficiency by up to 36.45x and space efficiency by up to 4 orders of magnitude compared with the well-known benchmark query schemes.

Background: Protein–protein interactions (PPIs) are of great importance in cellular systems of organisms, since they are the basis of cellular structure and function and many essential cellular processes are related to that. Most proteins perform their functions by interacting with other proteins, so predicting PPIs accurately is crucial for understanding cell physiology. Results: Recently, graph convolutional networks (GCNs) have been proposed to capture the graph structure information and generate representations for nodes in the graph. In our paper, we use GCNs to learn the position information of proteins in the PPIs networks graph, which can reflect the properties of proteins to some extent. Combining amino acid sequence information and position information makes a stronger representation for protein, which improves the accuracy of PPIs prediction. Conclusion: In previous research methods, most of them only used protein amino acid sequence as input information to make predictions, without considering the structural information of PPIs networks graph. We first time combine amino acid sequence information and position information to make representations for proteins. The experimental results indicate that our method has strong competitiveness compared with several sequence-based methods.

Multiagent reinforcement learning (MARL) is commonly considered to suffer from non-stationary environments and exponentially increasing policy space. It would be even more challenging when rewards are sparse and delayed over long trajectories. In this paper, we study hierarchical deep MARL in cooperative multiagent problems with sparse and delayed reward. With temporal abstraction, we decompose the problem into a hierarchy of different time scales and investigate how agents can learn high-level coordination based on the independent skills learned at the low level. Three hierarchical deep MARL architectures are proposed to learn hierarchical policies under different MARL paradigms. Besides, we propose a new experience replay mechanism to alleviate the issue of the sparse transitions at the high level of abstraction and the non-stationarity of multiagent learning. We empirically demonstrate the effectiveness of our approaches in two domains with extremely sparse feedback: (1) a variety of Multiagent Trash Collection tasks, and (2) a challenging online mobile game, i.e., Fever Basketball Defense.

Because of the continuous progress of vehicle hardware, the condition where the vehicle cannot load a complex algorithm no longer exists. At the same time, with the progress of vehicle hardware, the number of texts shows exponential growth in actual operation. In order to optimize the efficiency of mass data transmission in actual operation, this paper presented the text information (including position information) of the maximum entropy principle of a neural network probability prediction model combined with the optimized Huffman encoding algorithm, optimization from the exchange of data to data compression, transmission, and decompression of the whole process. The test results show that the text type vehicle information based on compressed algorithm to optimize the algorithm of data compression and transmission can effectively realize data compression. It can also achieve a higher compression rate and data transmission integrity, and after decompression it can basically guarantee no distortion. The method proposed in this paper is of great significance for improving the transmission efficiency of vehicle text information, improving the interpretability and integrity of text information, realizing vehicle monitoring, and grasping real‐time traffic conditions.