This paper presents an interval-observer-based actuator fault detection and isolation (FDI) method for underactuated autonomous underwater vehicles (AUVs) under bounded hydrodynamic uncertainty and time-varying ocean currents. A locally frozen linear time-invariant (LTI) representation enables deterministic set-membership analysis, and the robust adaptive propagated interval observer (RAPIO) propagates admissible center–radius state bounds within a Lyapunov framework. Adaptivity is introduced through a reinforcement learning (RL)-augmented uncertainty-bound modulation mechanism, where an offline-trained agent scales a nonnegative channel-wise slack term without modifying the scheduled observer-gain rule or the nominal center predictor. Under the stated observer and disturbance-envelope conditions, positivity, stability, and diagnostic-channel inclusion hold for any bounded learning signal. Actuator loss-of-effectiveness (LoE) faults are represented through the actuator-effectiveness channel and detected through interval-consistency violations, enabling axis-wise isolation of surge, yaw-rate, and pitch-rate actuator faults. The same schedule-blind decision layer is additionally evaluated with structurally distinct additive-bias and stuck/jam actuator models. All stuck/jam events are detected, and bias-magnitude sweeps identify channel-wise 100%-detection boundaries with zero false alarms. A structured 72-case scenario sweep shows reliable detection, strong false-alarm rejection, and acceptable detection delays compared with benchmark observers.

Unmanned marine vehicles (UMVs) are one of the most crucial components of the underwater communication and surveillance system in the marine and oceanographic environment. This research introduces a novel approach to stabilize the connection in an unstable underwater environment by inheriting features of communication velocities. The innovative proposed technique effectively divides the entire communication underwater by distributing the burden and ensuring trouble-free and efficient communication and surveillance. In existing approaches, troubleshooting communication issues often results in delays or slow performance. However, the proposed method focuses on establishing an efficient communication link to increase the throughput of the deployed underwater network. To achieve this, we have implemented two different communication channels, acoustic and wireless optical, that provide redundant pathways for communication, which enhances reliability and resilience. In the event of a denial-of-service attack or failure in one communication channel, the other channel becomes the primary means of communication, ensuring uninterrupted data transmission. We have used an attack prediction machine-learning model to further improve system security. Random forest (RF), na & iuml;ve Bayes classifiers, and machine learning methods like support vector machines (SVM), multilayer perceptrons (MLP), and k-nearest neighbors (KNN) are used to forecast the likelihood of an attack. To stop unwanted access or malicious activity, the associated communication channel is shut when a model shows a high probability of an attack. The efficiency of our recommended control strategy is confirmed by a simulation study of the proposed networked UMVs system. The suggested system delivers improved robustness and secure communication for real-time situations in the underwater environment by using machine learning-based attack prediction, redundant communication channels, and varied velocities to stabilize the connection. SVM achieves high accuracy at 98%. MLP is not far behind, with a 96% accuracy rate. The higher accuracy of SVM and MLP in predicting DoS assaults in UMVs is highlighted by the lesser accuracies of 66.3.

Underwater Acoustic Sensor Networks (UASNs) play a critical role in underwater exploration, yet face challenges like high energy consumption, and uneven load distribution in multi-hop routing. While existing clustering protocols like Anchor Nodes assisted Cluster-based Routing Protocol (ANCRP) have improved energy-efficiency through uniform cluster formation, they still suffer from unequal energy depletion among cluster heads (CHs), particularly in shallower layers. This paper introduces key innovations that advance cluster-based routing in UASNs. We propose an Energy-Adaptive Non-uniform Clustering Protocol (EANCP) with three novel aspects: 1) depth-optimized cluster sizing where deeper CHs manage larger clusters with lower communication demands while shallower CHs handle smaller clusters to prevent early energy depletion; 2) an adaptive beaconing mechanism where transmission rate varies with depth to conserve energy in stable deep regions; and 3) a multi-criteria CH selection strategy considering residual energy, congestion, and distance to sink. The protocol was designed and simulated in Python, leveraging its scientific computing ecosystem (NumPy, SciPy, and Matplotlib) for flexible modeling of underwater acoustic channels and energy consumption. Unlike ANCRP’s uniform approach, EANCP’s depth-aware design addresses the uneven energy consumption in UASNs. Simulations demonstrate significant improvements: a 14% lower energy consumption, 11.5% longer network lifetime, $5-8\% $ higher packet delivery ratio (PDR) and 20% lower delays compared to ANCRP. The protocol particularly excels in large-scale deployment by preventing routing bottlenecks through congestion-aware forwarding. These advancements make EANCP a significant improvement in underwater routing protocols, offering a more robust solution to balancing energy efficiency with reliable communication in harsh underwater environments.

This paper presents a novel nonlinear Stewart control method, which combines real-time dynamics estimation and an intelligent nonlinear control approach. The proposed solution combines (i) a Fractal-inspired Radial Basis Function (FRBF) network to estimate unknown system dynamics and (ii) a Deep Reinforcement Learning based Fractal-inspired Sliding Mode Controller (DRL-FSMC) to track the trajectory in uncertain conditions. The FRBF network estimates unknown disturbances with 43% parameter efficiency compared to standard RBF networks. The DRL-FSMC offers more robustness, less chattering and automatically tunes control gains online using a Soft Actor-Critic (SAC) algorithm, which enables robust trajectory tracking under parametric uncertainty and external disturbances. The reliability and accuracy of the proposed controller is proved by extensive simulations in different motion scenarios, such as step-inputs, sinusoidal paths, and uncertainty scenarios. The proposed controller exhibits better convergence rate, better tracking, and greater resistance to disturbances and modeling errors compared to the traditional PID and classical SMC approaches. The proposed approach, according to these results, has a considerable potential of applications that require agility and stability, including flight simulation and robotic-assisted surgery.

This research presents an innovative stacked autoencoder-based receiver design to address the challenges of efficient signal recovery for low probability of detection (LPD) constrained covert underwater acoustic communication (CUAC), thereby promoting secure and eco-friendly communications in underwater environments. Communicating through underwater channels presents significant challenges due to the long propagation of acoustic waves, signal attenuation, and ambient noise. These challenges are further intensified in LPD-constrained CUAC systems due to severely degraded SNR of the received signal at the desired range. The proposed system employs a data-driven approach for receiver design to effectively recover the signal in a low SNR environment by learning intrinsic channel features and replaces traditional physical-layer operations of de-spreading, demodulation, and decoding with a unified, joint feature learning framework, creating an efficient processing pipeline and enhancing the BER performance of the receiver. The performance of the stacked autoencoder-based receiver is validated through a lake experiment conducted at a range of 1.22 km, along with simulations. The experiment demonstrates that the proposed receiver achieves reliable performance with SNR levels as low as -15 dB, yielding a significant 3 dB improvement at a BER of 10-2 compared to conventional receivers.

This paper presents a novel multi-threaded autoencoder-based multi-hydrophone receiver designed for covert underwater acoustic communications under lowprobability-of-detection constraints. In the proposed architecture, each thread processes the signal from a different hydrophone in parallel through a dilated convolutional autoencoder, enhanced with a sequence fusion head to capture long-range temporal dependencies and spatial correlations across the array of received signals. This design jointly optimizes signal reconstruction and cross-hydrophone coherence, significantly improving computational efficiency and enabling reliable signal recovery in low-SNR environments. The data-driven receiver is evaluated against both traditional multi-filter-bank receivers and deep-learning baselines. Comprehensive testing on watermarked benchmark channels and an experiment in Qingdao Lake demonstrates substantial performance gains over all comparison models. The system achieves a BER of $\mathbf{1 0}^{\mathbf{- 2}}$ at an SNR as low as -17 dB on an experimental channel, and a multi-threaded parallel processing reduces inference latency by 40% compared to sequential hydrophone processing, confirming its practical viability.

Autonomous underwater vehicles (AUVs) require reliable, low-impact communication to ensure minimal environmental disturbance and facilitate coordinated operations across diverse marine regions. These applications necessitate adherence to stringent low-probability-of-detection (LPD) standards to maintain operational stealth and communication reliability. However, conventional acoustic communication technologies underperform in complex underwater settings. This paper presents a two-stage deep learning-based receiver for AUVs that integrates a dilated convolution in a convolutional autoencoder (CAE) framework for noise-resilient latent feature extraction, coupled with an LSTM classifier that performs spatiotemporal symbol detection using the resulting latent space. The CAE compresses channel-distorted signals into a structured latent space, while LSTM exploits temporal correlations to enhance detection, the training pipeline combines unsupervised encoder pretraining with supervised classifier fine-tuning, to test the performance of the proposed model simulated, benchmark NOF1 and the field-measured lake channels are utilized and the system is validated against traditional receivers as well as TCN and CNN based deep learning models. Experimental results on the benchmark dataset demonstrate a 5 dB SNR improvement over conventional systems, while maintaining bit error rates below 10-2 at-16 dB SNR. Field trials in Qiandao Lake confirm real-time operation, achieving a 4 dB performance gain over comparable deep learning approaches. This work demonstrates that CAE-based neural receivers enable low-impact AUV communications by reducing acoustic emissions per node while preserving coordination, supporting scalable, secure, and environmentally responsible ocean monitoring and exploration.

The increasing demand for clandestine communication in underwater acoustic environment reflects the remarkable growth of research in underwater acoustic communication and networking. Mariners are driven to transmit information covertly in the ocean keeping it hidden from unfriendly users and intruders. This research introduces a novel technique of covert underwater acoustic communication that mimics false killer whale whistles. The secret information is embedded using cepstrum transform to imitate Pseudorca crassidens whistles. This covert communication can be achieved even in the presence of eavesdroppers, who are unable to recognize the communication signal due to unique watermarking characteristics. The proposed model uses machine learning to assess imperceptibility and demonstrates exceptional robustness and improved capacity. To validate the model for secure communication and networks, underwater experiments were conducted, resulting in superior bit error rate and high watermark capacity with a perfect low probability of recognition constraint covert communication.

This paper presents a novel deep reinforcement learning based extended fractal radial basis function (DRL-EFRBF) network for accurate state-of-charge (SOC) estimation in lithium iron phosphate (LiFePO4) batteries. Unlike conventional methods such as open circuit voltage (OCV) and unscented Kalman filter (UKF), the proposed DRL-EFRBF framework eliminates the need for equilibrium conditions and predefined battery models. The proposed DRL-EFRBF approach combines adaptive policy learning with fractal-based nonlinear modeling to capture dynamic battery behaviors without relying on predefined battery models or requiring equilibrium states. The system was implemented using a Texas Instruments BQ76930EVM development kit and interfaced via an STM32 microcontroller. Experimental evaluations under constant, pulsed, and dynamic load profiles demonstrate that DRL-EFRBF achieves SOC estimation errors below 0.5%, outperforming OCV and UKF by over 90% and 70%, respectively. The proposed method exhibits rapid convergence, adapts to fluctuating conditions, and integrates effortlessly with IoT systems. These findings underscore its appropriateness for real-time battery management systems in electric vehicles and intelligent energy applications.

A continuously variable transmission can improve the energy efficiency of actuators with rotary output by providing an optimum transmission ratio. A continuously variable transmission based on circumferentially arranged disks (CAD CVT) is a new type of CVT that is highly beneficial for applications requiring large torques, like heavy road transport. However, its major drawback is that its efficiency drops in the low torque region. To overcome this problem, the current paper proposes an improved mechanical design in which the force on traction disks is changed according to the instantaneous torque requirement, thus resulting in improved efficiency in low torque regions. Furthermore, a hydraulic-actuation-based control system has been designed to ensure the optimum control of the improved mechanical design. The improved mechanical design of the CAD CVT is named CAD CVT-II, which is highly beneficial for variable torque applications such as road transport and wind turbines.

This research presents a review of low probability of detection (LPD) constrained underwater acoustic communication (UWAC). The recent surge in seabed resource exploration and overseas military operations has reignited interest in LPD CUAC. The demand for secure communication in underwater acoustic channels has increased as it is environmentally friendly and enables communication in the presence of potential adversaries, making it attractive for various sectors, including the oil and gas industry, military, and marine applications. Despite these advantages, a challenge lies in communicating without being detected. This article scrutinizes the pertinent research to provide insights into the mechanisms of this technology and attract communication system designers to this emerging research area. It starts with an overview of covert underwater acoustic communication and a detailed understanding of LPD analysis; then a range of applied and experimental work is explored, covering waveform designs, modulation, signal processing, and the role of neural networks in low signal-to-noise ratio communications. It also presents a combination of techniques for future research into stabilizing covert communication in underwater acoustic channels.

Classifying humpback whale calls plays a vital role in understanding the ecological behavior and social communication patterns of marine mammals in diverse underwater environments. This study explores the impact of regional acoustic variations in humpback whale calls on deep learning-based classification performance. Recordings were collected from four distinct regions: Tortola (British Virgin Islands), St. David's Island, Argus Island, and Castle Harbour (Bermuda). These vocal differences, often referred to as dialects, reflect regional adaptations in communication patterns and are essential to this classification task. A novel Multi-scale Deep Feature Aggregation (MSDFA) model was developed and evaluated alongside CNN, Residual-Net, and Dense-Net architectures. Preprocessing involved bandpass filtering and spectral subtraction-based denoising, followed by MFCC-based feature extraction. The MSDFA model achieved a classification accuracy of 95 %. To ensure model robustness and reliability, K-fold cross-validation, comprehensive error analysis, and advanced data augmentation techniques are employed. The results demonstrate how regional vocal characteristics influence classification accuracy and provide insights into the geographic diversity of humpback whale communication.

The performance of the UAVs while executing various mission profiles greatly depends on the selection of planning algorithms. Reinforcement learning (RL) algorithms can effectively be utilized for robot path planning. Due to random action selection in case of action ties, the traditional Q-learning algorithm and its other variants face the issues of slow convergence and suboptimal path planning in high-dimensional navigational environments. To solve these problems, we propose an improved deep Q-network (DQN), incorporating an efficient tie-breaking mechanism, prioritized experience replay (PER), and L2-regularization. The adopted tie-breaking mechanism improves the action selection and ultimately helps in generating an optimal trajectory for the UAV in a 3D cluttered environment. To improve the convergence speed of the traditional Q-algorithm, prioritized experience replay is used, which learns from experiences with high temporal difference (TD) error and avoids uniform sampling of stored transitions during training. This also allows the prioritization of high-reward experiences (e.g., reaching a goal), which helps the agent to rediscover these valuable states and improve learning. Moreover, L2-regularization is adopted that encourages smaller weights for more stable and smoother Q-values to reduce the erratic action selections and promote smoother UAV flight paths. Finally, the performance of the proposed method is presented and thoroughly compared against the traditional DQN, demonstrating its superior effectiveness.

Currently, AI and AR are the most advanced technologies inspiring innovations in numerous areas. While AR provides novel, engaging user experiences, it also creates tedious and unstructured development issues like content production and marker detection. Recently, AI has been investigated to provide solutions to those AR limitations. The results from those studies inspired this work to present DeepReality, a Unity 3D plug-in that combines Deep Learning (DL) models with AR using Barracuda inference engine and AR Foundation. The ease with which DeepReality provides DL-AR integration allows developers to deploy DL-based iOS and Android mobile apps to extract visual elements and overlay AR content onto physical objects. DeepReality enhances feature-based environmental monitoring through incongruous items and semantic processing of objects. This study also presents a smart gym device that uses AIto adjust posture in real time to ensure optimal form during workouts like the chest press and to increase effectiveness. The system analyzes the violation of postures and gives audio feedback using speech recognition by recognizing the body key points through the Mediapipe framework. In addition, while creating custom training programs, diet recommendations, and motivational music playlists, user goals will be integrated into the design, meanwhile, 3D virtual trainers demonstrate exercises. This AI-powered concept merges augmented reality with real-time feedback to create long-term exercise habits and overall wellness. Evaluation of DeepReality's execution time and memory consumption prove it to be user-friendly and menial device compatible. Extensive metrics and analysis, operational validation, and modularity for DL integration are provided by this open-source toolkit available from the Unity asset store. By making the AI AR integration available for all applications on AR, developers can leverage the transformative power of AIto enhance their AR applications. Thereby, the future of AR-AI synergy is forged with DeepReality.

This study introduces a technique for determining surface orientations by projecting a monochrome, spatial pixel-encoded pattern and calculating the surface normals from single-shot measurement. Our method differs from traditional methods, such as shape from shading and shape from texture, in that it does not require relating the local surface orientations of adjacent points. We propose a multi-resolution system incorporating symbols varying in sizes from 8 × 8, 10 × 10, 12 × 12, 14 × 14, and 16 × 16 pixels. Compared to previous methods, we have achieved a denser reconstruction and obtained a 5.2 mm resolution using an 8 × 8 pattern at a depth of 110 cm. Unlike previous methods, which used local point orientations of grid intersection and multiple colors, we have used the monochrome pattern and deterministic centroid positions to compute the unit vector or direction vector between the neighboring symbols. The light plane intersections are used to calculate the tangent vectors on the surface. Surface normals are determined by the cross-product of two tangent vectors on the surface. A real experiment was conducted to measure simple plane surfaces, circular surfaces, and complex sculptures. The results show that the process of calculating surface normals is fast and reliable, and we have computed 1654 surface normals in 29.4 milliseconds for complex surfaces such as sculptures.

This article addresses the challenges encountered in underwater acoustic communication (UWAC) and presents a novel approach for chirp spread spectrum (CSS) communication. CSS is recognized for its ability to adjust to multipath and Doppler dispersion in underwater conditions, despite it usually demands a large bandwidth time product to achieve optimal performance. To address this constraint and improve data rate, the paper proposes a neural network-based receiver for spectral efficient M-ary CSS communication. M-ary communication is accomplished by transmitting chirps with different start and stop frequencies. At the receiver, a multilayer perceptron (MLP) artificial neural network and a one-dimensional convolutional neural network (1D CNN) are used for supervised classification. The neural network is trained offline using a comprehensive dataset developed by the BELLHOP ray tracing algorithm, which simulates various underwater acoustic channels. The application of VTRM pre-processing equalization aims to enhance performance. The simulation results illustrate the superior performance of the proposed receiver when compared to a conventional receiver based on a matched filter. The 16-ary chirp spread spectrum 1D CNN and MLP receivers show a gain of 6 and 4 dB, respectively, in a simulated channel after undergoing VTRM pre-processing. Furthermore, the utilization of a 16-ary 1D CNN receiver results in a noticeable 6 dB enhancement in two recorded channels. However, the MLP receiver outperforms the traditional receiver in terms of bit error rate. The article emphasizes the possibility of higher data rates and enhanced performance in underwater communication systems by employing the proposed M-ary CSS neural network-based method.

Deploying and effectively utilizing wireless sensor networks (WSNs) in underwater habitats remains a challenging task. In underwater wireless sensors networks (UWSNs), the availability of a continuous energy source for communicating with nodes is either very costly or is prohibited due to the marine life law enforcement agencies. So, in order to address this issue, we present a Q-learning-based approach to designing an energy-efficient medium access control (MAC) protocol for UWSNs through collision avoidance. The main goal is to prolong the network’s lifespan by optimizing the communication methods, specifically focusing on improving the energy efficiency of the MAC protocols. Factors affecting the energy consumption in communication are adjustments to the interference ranges, i.e., changing frequencies repeatedly to obtain optimal communication; data packet retransmissions in case of a false acknowledgment; and data packet collision occurrences in the channel. Our chosen protocol stands out by enabling sensor (Rx) nodes to avoid collisions without needing extra communication or prior interference knowledge. According to the results obtained through simulations, our protocol may increase the network’s performance in terms of network throughput by up to 23% when compared to benchmark protocols depending on the typical traffic load. It simultaneously decreases end-to-end latency, increases the packet delivery ratio (PDR), boosts channel usage, and lessens packet collisions by over 38%. All these gains result in minimizing the network’s energy consumption, with a proportional gain.

In underwater environments, the accurate estimation of state features for passive object is a critical aspect of various applications, including underwater robotics, surveillance, and environmental monitoring. This study presents an innovative neuro computing approach for instantaneous state features reckoning of passive marine object following dynamic Markov chains. This paper introduces the potential of intelligent Bayesian regularization backpropagation neuro computing (IBRBNC) for the precise estimation of state features of underwater passive object. The proposed paradigm combines the power of artificial neural network with Bayesian regularization technique to address the challenges associated with noisy and limited underwater sensor data. The IBRBNC paradigm leverages deep neural networks with a focus on backpropagation to model complex relationships in the underwater environment. Furthermore, Bayesian regularization is introduced to incorporate prior knowledge and mitigate overfitting, enhancing the model’s robustness and generalization capabilities. This dual approach results in a highly adaptive and intelligent system capable of accurately estimating the state features of passive object in real-time. To evaluate the efficacy of this intelligent computing approach, a controlled supervised maneuvering trajectory for underwater passive object is constructed. Real-time estimations of location, velocity, and turn rate for dynamic target are scrutinized across five distinct scenarios by varying the Gaussian observed noise’s standard deviation, aiming to minimize mean square errors (MSEs) between real and estimated values. The effectiveness of the proposed IBRBNC paradigm is demonstrated through extensive simulations and experimental trials. Results showcase its superiority over traditional nonlinear filtering methods like interacting multiple model extended Kalman filter (IMMEKF) and interacting multiple model unscented Kalman filter (IMMUKF), especially in the presence of noise, incomplete measurements and sparse data.