Search Results (764)

Continuous monitoring of vehicle engine vibration is a key enabler of real-time diagnostics, yet conventional accelerometers require an external power supply and fit poorly into the distributed sensor networks envisioned for next-generation vehicles. Triboelectric nanogenerators offer an attractive self-powered alternative, but their direct application to the vibration of a running passenger vehicle engine, and the explicit link between sensor design parameters and individual engine operating states, remains largely unexplored. Here, we address this gap by co-tuning the air gap and the substrate rigidity of a contact-separation triboelectric vibration sensor to the vibration spectrum of an automotive engine. A systematic 3 × 3 design sweep across three gap distances and three substrate types identifies a single configuration that simultaneously resolves the low-frequency idle band and the higher-frequency acceleration band of a four-cylinder gasoline engine. A frequency-amplitude response map confirms that the real engine operating points fall within the sensitive region of the optimized device, and an on-vehicle test demonstrates clean discrimination of all seven operating states, from ready to shut-down, without any external power. The results establish design guidelines for source-matched triboelectric vibration sensors and outline a practical path toward self-powered, wireless-ready engine health monitoring in future vehicles. Full article

Post-quantum cryptographic implementations in Internet-of-Things (IoT) devices are significantly threatened by physical side-channel attacks, where practical attack risks are increased by physical accessibility and resource limitations. In particular, recent work has shown that belief propagation-based attacks can recover secret keys from lattice-based digital signatures using only a single side-channel trace of the Number Theoretic Transform (NTT). This work introduces the Quantum-Randomized Number Theoretic Transform (QR-NTT), an implementation-level defense mechanism that integrates quantum-derived entropy directly into the execution flow of lattice-based signature algorithms. Rather than treating randomness as a static input, QR-NTT uses quantum entropy to introduce controlled variability in execution ordering, arithmetic factor usage, and memory access behavior while preserving mathematical correctness and constant-time execution. The proposed framework is designed for embedded platforms and remains compatible with existing post-quantum cryptographic standards and IoT communication protocols. A complete implementation on an ARM Cortex-M4 platform, coupled with commercial quantum random number generator (QRNG) hardware, demonstrates that QR-NTT significantly degrades the effectiveness of template matching and belief propagation attacks. Experimental evaluation shows a reduction in single-trace attack success rates from over 90% to below 3% and an increase of approximately two orders of magnitude in the number of traces required for successful key recovery. These security gains are achieved with moderate overheads of 18.3% in execution time and 1.8 KB of additional memory while remaining well within practical IoT constraints. The results indicate that quantum-derived entropy can be leveraged as a practical implementation-level defense against physical attacks, complementing algorithmic post-quantum security. QR-NTT demonstrates a viable path toward strengthening the real-world resilience of post-quantum IoT systems without sacrificing deployability. Full article

In deep geothermal engineering, concrete slabs are prone to thermal cracking. The aggregates and pores are the core influencing factors for this failure behavior. However, existing research methods are unable to accurately capture the microscopic evolution process of thermal cracking and cannot clarify the intrinsic mechanism of how the characteristics of aggregates and pores affect the initiation and propagation of cracks. This limitation restricts the in-depth understanding of the laws of concrete thermal cracking. To address this deficiency, this study employs the discrete element method (DEM) and combines the particle flow program PFC2D to construct a microscopic model of concrete disks. By setting reasonable temperature parameters and thermal load boundaries, a numerical simulation system matching the actual deep geothermal high-temperature environment is established. Three sets of quantitative variables were designed: aggregate particle size (0.003, 0.004, 0.005, 0.006), aggregate volume fraction (0.35, 0.40, 0.45, 0.50), and porosity (0.11, 0.12, 0.13, 0.14). Through controlled variable simulations, the influence laws of each variable on the formation, propagation path, and time evolution of concrete thermal cracks were explored. The quantitative research results show that an increase in aggregate particle size significantly accelerates the generation and propagation of cracks. When the particle size is 0.006, the number of cracks is the highest and the propagation rate is the fastest. The aggregate volume fraction is negatively correlated with the final number of cracks, and 0.50 is the optimal fraction, at which the number of cracks is the smallest. A decrease in the fraction will lead to intensified stress concentration in the cement paste and a sudden increase in the number of cracks. An increase in porosity significantly disrupts the material continuity. When the porosity is 0.14, the bifurcation and connection of cracks are the most significant, while a low porosity of 0.11 can effectively inhibit the overall development process of thermal cracks. In addition, compared with traditional experimental methods and continuous medium numerical simulation techniques, the discrete element method has unique advantages in revealing the internal mechanism of concrete thermal cracking at the microscopic level. It can achieve real-time tracking of the evolution of discrete micro-cracks and the internal stress distribution characteristics. This study enriches the microscopic theoretical system of concrete thermal cracking and provides reliable quantitative references and technical support for the design of thermal crack resistance of concrete in deep geothermal engineering and the optimization of material composition. Full article

Most neural architectures model time as a one-dimensional real-valued variable, constraining temporal reasoning to sequential propagation along a single axis. We introduce Complex-Time Neural Networks (CTNN), a new class of architectures in which temporal coordinates are elements of the complex plane T = t + iτ ∈ ℂ, where Re(T) preserves chronological ordering and Im(T) encodes an orthogonal experiential dimension. Within this geometry, Im(T) < 0 defines a memory domain enabling retrospective retrieval, Im(T) = 0 corresponds to present-moment computation, and Im(T) > 0 defines an imagination domain for prospective projection. We prove the Expressive Separation Theorem (Theorem 1), establishing that, within the temporally coupled function class G_TCP and under explicit Assumptions A1–A4 (in particular the bounded projection Assumption A3), CTNN accesses temporally coupled functions at O(1) cost with respect to temporal distance Δ₁, Δ₂, while real-time architectures incur Ω(Δ₁ + Δ₂) sequential steps. For layered compositions, this yields an exponential composition gap within G_TCP under A1–A4. These advantages hold under the stated assumptions and may not directly generalize to broader function classes or large-scale settings where A3 cannot be maintained. Therefore, Theorem 1 provides a formal separation result for G_TCP, while CTNN more broadly defines a geometric framework for temporal computation. As the first concrete instantiation of this framework, we develop Complex-Time Convolutional Neural Networks (CTCNN). CTCNN achieves state-of-the-art performance on Something-Something V2 (70.2 ± 0.4%, +1.1% over VideoMAE v2, p < 0.01), strong performance on Kinetics-400 (78.4 ± 0.3%), and substantial gains on Long Range Arena Path-X (87.3% vs. 79.6%, +7.7%), using 3.4× fewer parameters than VideoMAE v2. Learnable angular parameters α and β provide computationally interpretable parameters related to memory-access span and prospection breadth, with values varying systematically across task families. Full article

Deep reinforcement learning (DRL) for autonomous mobile robot navigation faces several inherent limitations. The stochastic nature of actions generated by DRL policies can undermine performance consistency, while inefficient exploration frequently delays the learning process or prevents the discovery of optimal solutions. This research aims to enhance the robustness of path planning by addressing these challenges. To achieve this goal, we propose a hybrid approach that integrates the flexible decision-making capabilities of deep reinforcement learning with the stability of traditional path planning. The proposed model adopts the Twin Delayed Deep Deterministic Policy Gradient (TD3) network as its base. Notably, we pre-process LiDAR point cloud data to extract only essential features for the state representation, thereby preventing performance degradation from high-dimensional inputs and improving computational efficiency. Our model optimizes the learning process through two core strategies. First, it prioritizes experience data generated during training based on negative rewards, guiding the model to learn more frequently from critical failures rather than redundant successes. Second, it dynamically compares the action proposed by the TD3 network with a goal-oriented action from a classical path-planning algorithm in real time. By selecting the action with the higher estimated value, the model guides the policy toward a stable and effective trajectory from the earliest stages of training. To validate the efficacy of our approach, we conducted simulation-based experiments comparing the performance of the proposed model with existing reinforcement learning networks. To ensure statistical significance and mitigate the impact of random initialization, all reported results are averaged over 10 independent runs with different random seeds. The results quantitatively demonstrate that our model achieves significantly higher and more stable reward values, confirming a robust improvement in the path-planning process. Full article

Conventional heading-keeping autopilot logic exhibits well-known performance limitations under complex route geometry and environmental disturbances. Motivated by this limitation, this paper proposes a line-adhering intelligent navigation framework for disturbance-aware path-following of autonomous ships. The core idea is based on numerical simulation scenarios representing curved inland/coastal routes under wind- and current-disturbance conditions. The addressed gap lies in the limited integration of route-geometry adherence, human-like maneuvering logic, and disturbance-aware controller reconfiguration within conventional heading-centered ship path-following frameworks. Therefore, a rough-set classifier identifies disturbance modes and reconfigures PID, LQR, and MPC controllers in real time. Moreover, a vessel-dynamics constrained Bézier refinement method generates high-resolution reference paths aligned with navigational curvature limits. Mathematical models including the Nomoto and MMG formulations are incorporated to ensure controllability and dynamic feasibility. Results show that the proposed framework improves path-following precision, robustness, and comfort under the considered simulation conditions. Full article

To mitigate local congestion and address the adaptability limitations of traditional static routing under dynamic traffic, this paper proposes an end-to-end routing method based on a Graph Attention Network (GAT), termed Congestion-Aware Graph Attention Routing (CA-GAR). To alleviate the issue of local optima in traditional heuristic iterative optimization, we design a dynamic link cost optimization algorithm with multi-start parallel exploration. This algorithm employs a ”penalty–reselection–reward” closed-loop feedback mechanism, performing global searches from multiple random initial states to generate a high-quality, empirically near-optimal cost matrix as supervised labels. Building on this, CA-GAR leverages a multi-head attention mechanism to adaptively aggregate high-order topological features of nodes and edges, and incorporates a staged hierarchical hyperparameter optimization strategy to map real-time network states to link costs. Simulation results demonstrate that CA-GAR outperforms traditional static routing under light, medium, and heavy loads. Under high-load burst conditions, the method exhibits effective congestion avoidance capability, reducing end-to-end delay by approximately 50% and lowering the packet loss rate to as low as 2%. Compared with QLRA, CA-GAR shows promising performance in multi-path traffic splitting and possesses robust fast rerouting capabilities during node failures, thereby achieving intelligent traffic distribution and global load balancing. Full article

Industrial surface inspection is fundamental to advanced manufacturing, yet reliable robotic image acquisition in complex geometries remains challenging due to severe occlusions and the inherent trade-off between resolution and coverage. Inspired by human visual inspection behaviors and perception–action coordination mechanisms, this paper proposes a hierarchical trajectory optimization framework for robotic image acquisition based on measured point clouds. Specifically, a multi-constraint preprocessing model is developed to emulate human-like active perception strategies, enabling occlusion-aware viewpoint generation over complex concave and convex surfaces with adaptive camera orientation. Building upon this, a multi-objective trajectory optimization method is introduced to coordinate global coverage and local motion efficiency, jointly optimizing viewpoint sequencing, path length, and motion smoothness hierarchically. To further enhance flexibility in constrained environments, a Pose Reachability Augmented Generative Adversarial Network (PRAGAN) is proposed to learn feasible and adaptable imaging postures under kinematic constraints. Experimental results on an industrial robotic platform equipped with 2D and 3D vision systems demonstrate 100% coverage of key surface areas, a 47.0% reduction in path length, and a 37.5% decrease in solution time compared with the baseline in the physical experiments, while ensuring collision-free operation. Both simulation and real-world experiments validate that the proposed framework effectively captures human-inspired perception and motion coordination, providing a practical and scalable solution for complex industrial surface inspection. Full article

Existing studies of street-related emotional perception mainly rely on static scene evaluations, which cannot capture the cumulative effects of environmental exposure during continuous walking. To address this limitation, this study proposes a method for predicting pedestrian physiological responses in sequential audiovisual street environments. Four real-world walking routes were selected, with outbound and return directions treated as independent paths, yielding eight paths and 32 valid samples. EEG, ECG, sound pressure level, first-person video, and GPS data were synchronously collected to construct a 1 s multimodal time-series dataset. Pearson correlation, Kendall correlation, and mutual information analyses were used to examine linear, monotonic, and nonlinear relationships between environmental variables and physiological indicators, and the resulting weights were incorporated into a Long Short-Term Memory (LSTM) model for multi-step prediction. Visual elements and noise exposure were the main factors influencing physiological responses. Among the models, the mutual-information-weighted LSTM performed best, achieving an R² of 0.77 for heart rate variability (RMSSD), whereas prediction of the EEG ratio (β/α and θ/β) remained limited. An additional independent street sample outside the training set was then used to generate a dual-dimensional EEG-ECG physiological response map, demonstrating the model’s potential for identifying emotional risk segments and supporting street-level micro-renewal. Full article

To address trajectory-tracking degradation of unmanned surface vehicles (USVs) in constrained waters caused by model uncertainty, strong environmental disturbances, and actuator limitations, this paper proposes a robust disturbance-observer-based optimization model predictive control method. First, a nonlinear tracking error model is established for a 3-DOF USV by incorporating environmental loads, parametric perturbations, and unmodeled dynamics into the kinematic and dynamic equations. Based on this model, a prediction model suitable for model predictive control is derived through linearization and discretization. Then, to estimate complex unknown disturbances online, a robust disturbance observer integrating a radial basis function neural network (RBFNN) with an adaptive sliding-mode mechanism is developed, enabling real-time approximation and compensation of lumped disturbances in the surge and yaw channels. Furthermore, to overcome actuator saturation caused by the direct superposition of feedforward compensation and feedback control in conventional composite strategies, a dynamic constraint reconstruction mechanism is introduced. By feeding the observer-generated compensation signal back into the MPC optimizer, the feasible control region is updated online so that the total control input satisfies both magnitude and rate constraints of the propulsion system. Theoretical analysis based on Lyapunov theory proves the uniform ultimate boundedness of the observation errors and neural-network weight estimation errors, while input-to-state stability theory is employed to establish closed-loop stability. Comparative simulations under sinusoidal trajectories, time-varying curvature paths, and large-maneuver turning conditions demonstrate that the proposed method significantly improves tracking accuracy, disturbance rejection capability, and control feasibility under severe disturbances and parameter mismatch. Full article

The detection technology for metal surface defects plays a crucial role in improving metal product quality and production efficiency in various manufacturing and 3-D printing factories. Metal defect detection faces scale variation and irregular shapes, which limit the adaptability of general object detection models in industrial scenarios. Deep learning-based methods are widely used for metal surface defect detection due to their strong adaptability and high automation. Yet, their existing studies pay limited attention to adaptability, evaluation, and recommendations across different detection methods for metal surface defects. This work mainly discusses YOLO, R-CNN, and transformers, as well as FPN, and analyzes their applications in metal surface defect detection according to their respective characteristics, to provide guidance for future research. YOLO has advantages in real-time industrial online detection, while R-CNN and transformer models show potential advantages in handling complex defect cases. Additionally, this work summarizes commonly used datasets and evaluation metrics for metal surface defect detection and analyzes the benchmark performance of different types of detection methods. It also discusses future research directions, including the current status and improvement paths of different models in terms of accuracy, real-time performance, and adaptability. Future models should focus on balancing accuracy and real-time performance, exploring new hybrid architectures, and improving adaptability to different metal surface defects to support further development in this field. Full article

As industries deploy more connected devices in factories, warehouses, and smart facilities, the need for artificial intelligence (AI) systems that can operate securely in distributed, data-intensive environments is growing. Traditional centralized learning and online education platforms struggle when students and systems have to process real-time streams (sensors, video, text) with strict latency and privacy requirements. To address this challenge, a blockchain-secured, edge-enabled multimodal federated learning framework tailored for Industrial IoT (IIoT) environments is proposed. The model integrates four key layers: (i) a blockchain layer that provides credentialing, transparency, and token-based incentives; (ii) a multimodal community layer that supports group formation, peer consensus, and cross-modal learning across text, images, audio, and sensor data; (iii) an edge computing layer that enables low-latency task offloading and secure training within Intel SGX enclaves; and (iv) a data layer that applies pre-processing, differential privacy, and synthetic augmentation to safeguard sensitive information. Experiments on industrial multimodal datasets demonstrate 42% faster model aggregation, 78.9% multimodal accuracy, and 1.9% accuracy loss under ε = 1.0 differential privacy. This shows a scalable and practical path for decentralized AI training in next-generation IIoT systems, confirming the possibility of technical support for educational processes. However, the conducted research requires a validation of pedagogical effectiveness. Full article

The NP-hard coordination of heterogeneous robots for time-windowed warehouse tasks remains challenging: metaheuristics are precise but slow, whereas neural methods cannot handle heterogeneous constraints, leading to infeasible allocations. This paper presents the Constraint-Aware Heterogeneous Transformer (CAHT), a lightweight encoder–decoder architecture that performs end-to-end task assignment and sequencing in a single forward pass. The central innovation is a dynamic feasibility masking mechanism that enforces capacity and energy constraints directly within the softmax computation, eliminating infeasible allocations at the architectural level. This is complemented by a spatial-bias Transformer encoder and a two-stage supervised–reinforcement learning training paradigm using ALNS-generated labels. Experiments across four problem scales (5–20 robots, 50–200 tasks) demonstrate that CAHT achieves objective values within 7–13% of the ALNS reference while being 29–91× faster (23–104 ms vs. 2–3 s). Constraint violation rates remain below 6%, with time-window satisfaction above 94%. Ablation analysis identifies dynamic masking as the dominant contribution (+213% degradation upon removal), and cross-scale generalization reveals that the optimality gap decreases from 13.0% to 10.7% as the problem scale grows. With only 0.91 M parameters, CAHT occupies a new trade-off point on the Pareto frontier, offering a practical path toward real-time autonomous warehouse coordination. Full article

For complex tasks such as search and recovery in uncharted maritime areas, the use of heterogeneous unmanned swarms (UAVs and USVs) is highly promising, yet effective cross-domain cooperative trajectory planning remains a key challenge, often leading to mission delays. This paper proposes a rapid Cooperative Cross-domain Path Planning framework (CCPP) and its associated algorithm for heterogeneous UAV–USV swarms. The framework first establishes a visual-fusion modeling pipeline, converting visual pose estimation, uncertainties, and semantic dynamic obstacles into a planning representation with robust safety margins and time-varying risk fields. A hybrid velocity-path co-optimization algorithm is then designed to simultaneously generate curvature-feasible trajectories and speed profiles under heterogeneous kinematics and explicit temporal constraints. In the end, an adaptive interpretable decision tree acts as a meta-strategy for online replanning and real-time adjustment of modes and weights. To address the critical issue of uneven arrival time distribution, this paper introduces, inspired by economic inequality analysis, a normalized Gini coefficient-based arrival time consistency index to quantify and optimize coordination timing. Comprehensive experiments validate the effectiveness of the proposed approach in enhancing cooperative efficiency and real-time adaptability. Full article

This paper presents a deep reinforcement learning framework for autonomous navigation based on multi-modal belief state representation learned from LiDAR and depth sensors. To address the challenges posed by partial observability and sensor-specific uncertainty, we propose a probabilistic representation module that models belief states as Gaussian distributions over latent environmental features. Sensor-specific encoders extract structured features from raw LiDAR and depth inputs, which are fused using a Q-value-guided weighting scheme derived from the policy critic. A motion-prediction pretraining strategy and a cross-modal coherence loss are introduced to enhance the alignment and reliability of the learned belief states. The resulting representation is integrated into a Soft Actor–Critic (SAC) framework to enable policy-driven decision-making under uncertainty. Extensive experiments in simulated environments demonstrate that the proposed method improves success rate, navigation efficiency, and generalization. Real-world experiments further validate these findings, with the proposed method outperforming a classical navigation baseline by reducing average travel time by 16% and path length by 4%. These results support the use of probabilistic multi-modal belief modeling for autonomous navigation under partial observability. Full article