ABSTRACT. Self-driving technology is rapidly transforming the transportation industry; however, existing autonomous navigation approaches often address safety, explainability, or regulatory compliance in isolation, limiting real-world trust and deployability. This paper proposes a Responsible Artificial Intelligence (RAI) framework that unifies safety assurance, explainability, and policy compliance within a single decision-level architecture, rather than treating these aspects as post-hoc or loosely coupled add-ons as seen in prior work. Safety is ensured through risk-aware decision modeling and real-time uncertainty monitoring, while navigation transparency is achieved via a combination of intrinsic interpretability and post-hoc explanation mechanisms. In contrast to existing Responsible AI approaches that lack operational policy enforcement, the proposed framework integrates rule-based ethical constraints and regulatory ontologies directly into trajectory planning to ensure continuous legal compliance. Experimental evaluation using simulation and benchmark driving datasets demonstrates that the proposed approach significantly improves safety awareness, decision transparency, and regulatory adherence without compromising navigation efficiency. These results position Responsible AI as a practical enabler of trustworthy, deployable autonomous navigation systems.

ABSTRACT. This article investigates the potential avenues of value creation via the integration of AI technologies in project management practices within the construction industry which is a critical area contextually in Industry 4.0. The research zones in on key questions such as the numerous advantages and challenges of AI adoption in construction, particularly the ability of AI to enhance decision-making, improve operational efficiency, enable predictive analytics for optimised project outcomes. To address these questions, a mixed-methods approach involving a Systematic Literature Review (SLR) using PRISMA framework and a thematic qualitative analysis was employed. This approach was integral in thoroughly exploring existing AI applications in construction, highlighting the crucial absence of a framework for AI-driven project management. This study endeavoured to address that research gap by developing and testing a new AI-driven project management framework with an iterative prompt library at the heart of it. This prompt library was designed specifically for AI output refinement to improve accuracy of responses and lessened hallucinations, eventually leading to the AI model being able to handle more complex tasks. The findings unveil that AI has immense potential to significantly enhance construction project management by optimising decision-making and streamlining operations. Despite this, the research highlights multiple limitations, such as ethical concerns, data quality problems and the necessity of robust training programs to improve AI literacy in construction professionals. To conclude, although AI presents tantalising opportunities for project management optimisation in construction, its successful consideration requires careful consideration of technical, ethical and educational factors. These findings harbour significant implications for future research practices, highlighting how crucial they are in the ongoing development of AI frameworks and tools bespoke to the needs of the construction industry.

ABSTRACT. Examining pipelines with restricted dimensions presents ongoing difficulties across various sectors, especially where accessibility, size limitations, and debris buildup obstruct traditional inspection techniques. This research delineates the design, modelling, and functionality of a novel self-towed modular robotic system developed for the autonomous inspection and cleaning of small-diameter pipes (76–100 mm). The robot features a multi-link articulated chassis and Mecanum wheels, facilitating omnidirectional movement in curved or sloping settings. A dynamic model utilising the Euler-Lagrange formulation facilitates two control strategies: the traditional Linear Quadratic Regulator (LQR) and an advanced Fractional-Order Linear Quadratic Regulator (FO-LQR). Simulation and hardware validation demonstrate that FO-LQR markedly enhances tracking precision and damping under nonlinear situations. The system, designed for industrial pipeline maintenance, features a compact, self-contained design, adaptability to confined spaces, and a robust control architecture, making it highly applicable to future space robotics—especially for inspecting space station conduits, life support tubing, or planetary habitat infrastructure, where autonomous operation in restricted environments is essential

ABSTRACT. Underwater Wireless Sensor Networks (UWSNs) face significant challenges, including high attenuation, limited bandwidth, and environmental interference. This research explores the integration of quantum sensing technologies to address the open issues and limitations. We examine quantum-enhanced detection methods, entanglement-based communication protocols, and quantum magnetometry for underwater applications. Our analysis reveals that quantum sensing can improve signal detection sensitivity by 10-20 dB, extend communication ranges by 30-40\%, and enhance positioning accuracy to sub-meter levels. We also provide a comprehensive comparison with existing surveys, identify critical research gaps, and propose a road map for practical implementation of quantum-enhanced UWSNs.

ABSTRACT. Machine learning has become a transformative force in VLSI design and semiconductor manufacturing. This study provides a critical review of state-of-the-art ML approaches from across the digital IC design flow (from HLS, via RTL PPA prediction, to logic synthesis, placement, routing, verification, DFT, and test), and major semiconductor manufacturing domains, such as lithography, etch modeling, metrology, defect classification, yield prediction, and equipment health monitoring. This study provides comparative insights, cross-domain analysis, and taxonomy-driven organization for ML methods such as supervised learning, GNNs, RL, AutoML, generative AI, PINNs, and LLM-based approaches. There is the bottlenecks-domain shift, data scarcity, compute cost, and explainability identified -and put forward a future roadmap that includes hybrid physics-ML modeling, industrial-scale datasets, explainable models, and ML-driven digital twins. This systematic synthesis shows that ML is transforming EDA workflows and making next-generation semiconductor manufacturing intelligence achievable

ABSTRACT. Electric vehicles (EVs) are becoming increasingly popular due to their environmental benefits, energy efficiency, and lower long-term operational costs. However, ensuring their long-term reliability, safety, and optimal performance depends heavily on implementing effective diagnostic and monitoring systems. This project introduces an innovative Internet of Things (IoT)- based solution designed specifically for real-time monitoring and diagnostics of critical EV parameters, including but not limited to battery voltage, current, charge level, motor temperature, and overall system health. The system incorporates a set of on-board sensors integrated with a microcontroller (such as the ESP32), which collects data continuously during vehicle operation. The collected data is then securely transmitted to a cloud platform using encrypted telemetry channels and authenticated connections. This ensures the protection of sensitive data such as battery health metrics, temperature fluctuations, and even GPS- based location tracking, from unauthorized access or tampering. Real-time alerts generated from the system allow for early fault detection, facilitating timely preventive maintenance and enhancing vehicle safety. Data visualization dashboards enable intuitive analysis for both drivers and technicians. The developed monitoring solution is cost-effective, lightweight, and scalable—making it suitable for commercial as well as research applications. It contributes to smarter, safer, and more efficient electric vehicle management.

ABSTRACT. This paper compares contour-based image processing techniques with the YOLOv5 model's deep learning approach for automated ceramic tile inspection. It also introduces a prototype for grading tiles based on their defects. In industries where manual inspections are still common, maintaining quality and efficiency in tile production is crucial. Manual inspection practices are often inefficient, error-prone, and time-consuming. As a result, tile production companies face challenges such as raw material wastage and lost profits from defective products. To address these issues, this research proposes an automatic tile inspection system and provides a comparative analysis of the accuracy rates of each model, including their respective strengths and weaknesses. The goal of this research is to enhance quality control by detecting defects, which in turn will reduce costs and human errors. The effectiveness of the proposed system will be measured by determining whether each tile is rejected as faulty.

ABSTRACT. The rising urban population causing the demand for electrical infrastructures. As the population expands the infrastructure of the cities is also required to expand and relatively the streets and street lighting systems also escalate. This produces an extra burden on the power supply companies and the provision of power supply to cater for areas expansion cannot meet the power demand. To solve these crises of shortage of power supply, the use of Traffic Wind Turbines (TWTs) provides a realistic alternative for sustainable street lighting. It provides the viability of using wind energy generated by passing automobiles to reduce the load on traditional power networks, which helps with energy conservation and environmental sustainability. The system uses an axial flux alternator to produce power to charge the battery of 12 volts and the battery is then connected to an inverter which provides 220 volts AC power to the street light to illuminate the street. Since LED lights are the most economical lights and have high luminosity as well as consume much less power than halogen, or tube lights therefore, these lights are preferred over others which also fulfill the purpose of alternate power production. This research article describes the design, execution, and analysis of a vertical axis wind turbine (VAWT) system used to power LED streetlights.

ABSTRACT. In industrial quality control, automated surface inspection is still a major challenge, especially for small, low-contrast, or patterned defects. This paper offers a comparative analysis of contemporary deep-learning detectors, traditional image processing techniques, and a new Hybrid Contour–CNN framework that combines convolutional neural network classification with contour-driven defect localization. Through further acquisition and synthetic augmentation, the ceramic tile defect dataset was increased from 1,750 to 2,550 images, encompassing a variety of surface textures, lighting conditions, and defect types. 5-fold stratified cross-validation was used to benchmark the suggested hybrid model against YOLOv5s, YOLOv8n, SSD-MobileNetV2, a lightweight CNN, and a contour-based pipeline. According to the results, when compared to standalone detectors (YOLOv5s F1 = 70.6% ± 2.8, YOLOv8n F1 = 74.7% ± 1.9), the hybrid model achieves the highest F1 score (88.3% ± 1.5) and improves the detection of subtle defects.

ABSTRACT. Small and Medium Enterprises (SMEs) serve as key drivers of Malaysia’s economic growth, job creation, and community well-being. However, their resilience continues to be challenged by financial constraints, stakeholder expectations, and rapidly evolving digital demands. This paper proposes a tech-enabled conceptual framework that integrates financial sustainability, stakeholder collaboration, and community development to strengthen Malaysian SMEs in the digital era. The framework emphasizes the importance of active engagement with government bodies, financial institutions, industry partners, and community organizations, highlighting how digital tools can enhance transparency, resource efficiency, and strategic decision-making. By aligning stakeholder objectives with digital and financial sustainability goals, SMEs can adopt innovative practices such as digital financial management, online customer engagement, and technology-driven business models. The framework also positions community development as a core dimension of SME resilience, linking digital adoption and financial stability with broader socio-economic contributions. This study provides a theoretical foundation for future empirical research and offers actionable insights for policymakers and practitioners seeking to enhance SME performance, adaptability, and long-term viability in an increasingly digital landscape.

ABSTRACT. Small and Medium Enterprises (SMEs) serve as key drivers of Malaysia’s economic growth, job creation, and community well-being. However, their resilience continues to be challenged by financial constraints, stakeholder expectations, and rapidly evolving digital demands. This paper proposes a tech-enabled conceptual framework that integrates financial sustainability, stakeholder collaboration, and community development to strengthen Malaysian SMEs in the digital era. The framework emphasizes the importance of active engagement with government bodies, financial institutions, industry partners, and community organizations, highlighting how digital tools can enhance transparency, resource efficiency, and strategic decision-making. By aligning stakeholder objectives with digital and financial sustainability goals, SMEs can adopt innovative practices such as digital financial management, online customer engagement, and technology-driven business models. The framework also positions community development as a core dimension of SME resilience, linking digital adoption and financial stability with broader socio-economic contributions. This study provides a theoretical foundation for future empirical research and offers actionable insights for policymakers and practitioners seeking to enhance SME performance, adaptability, and long-term viability in an increasingly digital landscape.

ABSTRACT. Concept drift is the phenomenon of changes in the statistical characteristics of a data stream which effect the relationship between the inputs and the target variable leading to ineffectiveness of an ML model, and in extreme cases makes it obsolete. There are multiple types of concept drift and widely recognized methods are used to detect them but is widely accepted that there is no single method for all types of drift. Ensemble methods have been proposed but their base learners themselves are prone to drift. In this work statistical drift detection base learners are applied to three synthetically created datasets for each type of concept drift and then the results are compared. Furthermore, the drift detection methods are also applied to three real-world datasets. Afterwards, three stacking ensembles are created using different combinations of the drift detection methods and they are also applied to the datasets. Three Meta-learners are used and compared namely Logistic regression, XG Boost and Random Forest. The results show that all three outperform stand-alone drift detection algorithms on the simulated datasets and in case of real-world datasets, they are able to make correct predictions even when the stand alone base learners fail to do so. Furthermore, statistical base learners make sure that the methodology for drift detection isn’t circular i.e. prone to drift themselves as is the case with ML based drift detection algorithms.

ABSTRACT. Accurate tumor segmentation in medical imaging is crucial for effective diagnosis and treatment planning in cancer care. This research compares the performance of the classical Active Contour Model (ACM) with that of the Medical Segment Anything Model (MedSAM) for tumor segmentation across datasets, including ultrasound, mammograms, and digital breast tomosynthesis (DBT). The findings revealed that MedSAM significantly outperforms ACM in both ultrasound and mammogram applications, owing to its transformer-based architecture, which adeptly handles noise and the diverse appearances of tumors. However, both methods demonstrated comparable performance in DBT. The findings advocate integrating MedSAM into clinical workflows for 2D breast imaging, while highlighting specific challenges in 3D modalities that necessitate further specialized algorithmic development.

ABSTRACT. This paper explores the critical role of Magnetic Induction (MI) in establishing robust communication and power transfer for Internet of Underwater Things (IoUT) networks. MI technology presents significant advantages over traditional acoustic methods, including a wider frequency band, shorter propagation delay, and enhanced conductivity. Crucially, it also enables underwater wireless power transfer (WPT). As a use case, we proposed a hybrid model that uses both underwater communications using MI coils and surface communication via an IoT network for an underwater monitoring system. In this work, a floating buoy design is proposed that uses solar energy to provide power to the batteries and wireless power using MI coils to power up IoT controller that is responsible for data acquisition underwater. The results show the data fetched from sensors and distance covered using the coils in a relayed network. This demonstrates that it is possible to conduct future research into other underwater applications by implementing MI as a means of underwater communication.

ABSTRACT. The future of an intelligent and autonomous vehicle cyber-physical systems (AV-CPS) highly depends on secure communications and AI-driven decision systems. These autonomous vehicles (AVs) are data-driven and involve various connected sensing devices, which, if exploited, can expose the AV to advanced AI-based cyber threats such as Distributed Denial of Service (DDoS) attacks. This research highlights the role of Generative AI (GenAI) by introducing a novel Diffusion-GAN architecture that combines diffusion-based latent conditioning with residual fusion for enhanced DDoS detection in AV environments. The proposed model achieves an accuracy of 96.07\% on the \emph{UL-ECE-5G-AV-DDoS2025} dataset. Beyond detection, the system incorporates an LLM-based post-detection analysis using GPT-3.5 Turbo with zero-shot prompting to interpret anomalies and generate actionable insights. This combination of Diffusion-GAN detection and LLM-driven analysis provides an adaptive real-time security mechanism for 5G-enabled autonomous vehicles.

ABSTRACT. The intelligent Autonomous Vehicles (AVs) rely on 5G connectivity to exchange positional data, perception outputs, and control commands in real time. This communication enables coordinated motion and safety-critical response. However, it also increases the environments exposure to Distributed Denial-of-Service (DDoS) attacks, that disrupt message flow and impair operational safety. To investigate this risk, a simulated driving and DDoS attack scenario is developed in the CARLA simulator under a 5G-enabled AV setting. This setup yields the \emph{UL-ECE-5G-AV-DDoS2025} dataset, which contains regular driving traffic and DDoS attack traffic. The dataset includes Global Positioning System (GPS) coordinates, speed, acceleration, throttle, steering, braking values, and network-layer metrics. Hence, this \emph{UL-ECE-5G-AV-DDoS2025} dataset supports reproducible benchmarking for Generative-AI (Gen-AI), machine learning, deep learning, and traffic-flow analysis in 5G-connected transport environments.

ABSTRACT. Increasing infrastructure resilience and sustainability requirements have led to the creation of innovative Structural Health Monitoring (SHM) systems that can identify damage, anticipate deterioration, and facilitate prompt maintenance decisions. The presented study is an artificial intelligence (AI)-based smart SHM framework based on wireless sensor networks, deep learning algorithms, and digital twin technology to monitor critical infrastructure in real-time. To obtain multi-modal data in different operating and environmental conditions, a hybrid sensing system made up of accelerators, strain-guidelines, acoustic emission sensors, and fiber-optic temperature sensors was installed. Adaptive wavelet filtering coupled with the statistical and modal analysis were used to preprocess the collected signals and feature engineer them, and the principle component analysis was used to reduce them. Several AI models (Support Vector Machine (SVM), Random Forest (RF), Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) in the damage classification were tested, and LSTM model had more accuracy (97.5) and receiver operating characteristic area under the curve (ROC-AUC) of 0.99. The predictive maintenance was accomplished by the use of the LSTM-based forecasting model that could predict a remaining useful life (R 2 = 0.94) and offer a set of nine-days early-warning. Tests of robustness showed the strength of the framework at temperature, humidity, noise, and biased sensor behavior with more than 92 percent accuracy. The combination of AI, edge computing, and visualization of the digital twin allowed the uninterrupted, almost real-time identification of anomalies, scheduling of maintenance based on its condition, and enhancement of situational awareness. The suggested AI-powered SHM can be considered an important step toward autonomous, intelligent, and sustainable infrastructure management.

ABSTRACT. Effective patient management requires quick healthcare services, but this is often interfered with by delayed diagnosis of patients, poor appointment scheduling, and insufficient real-time support in decision making. This paper presents a machine based Clinical Decision Support System (CDSS) that addresses these challenges through intelligent symptom analysis methods and risk stratification. The system provides an easy to use web interface in which patients feed in symptoms and health data to identify and treat respiratory problems like influenza, seasonal sore throats, and lung cancer at an early stage. Three machine learning models were compared on a publicly available lung cancer dataset, which were Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Decision Tree; the latter provided the optimal disease-classification results. The trained model will analyze patient-reported symptoms to estimate the probability of disease preventing unnecessary hospital visits and ensure that high-risk patients receive care on time. The real-time alert system informs the nursing staff about severe cases to allow them to prioritize and intervene quickly. In addition, the CDSS has an intelligent appointment-scheduling feature, which provides data driven guidance, which is available based on automated risk identification, reducing care delays. Implemented as a web-based solution, the system is shown to be practically useful in clinical practice, and our findings indicate that this interconnected solution helps to bridge the gap between patient symptoms and the proper medical response to improve the use of healthcare resources and patient outcomes.

ABSTRACT. Artificial Intelligence (AI) integrated with Hospital Information Systems (HIS) provided a new way of delivering better health care with great hope toward improving patient outcomes, achieving efficiency, and reducing costs. The paper is an analysis of the implementation of AI and HIS in 108 hospitals from around the globe toward that end measurable enhancement in patient care and system performance. Appropriate data were meticulously collected from healthcare settings in North America, European countries, Asian countries, and developing regions across large academic medical centers, community hospitals, and specialty care facilities. Results yielded a substantially positive direction in attaining patient safety whereby mortality rates reduced between 7% and 22% among different implementations. Diagnostic accuracy increased between 15% and 30%, while operational efficiency improved between 20% and 40%. The economic analysis shows very good ROI prospects. In five years, AI-powered radiology systems would produce average returns of 451%. This gives a national cost saving possibility between $200-360 billion per year. We found seven major fields for the application of AI: predictive analytics; clinical decision support systems; automated diagnosis and imaging analysis; natural language processing; machine learning for treatment recommendations; patient monitoring systems and virtual health assistants. Success of implementation depends on such factors as strong leadership, effective stakeholder involvement, solid data infrastructure, and smart integration with the current clinical workflows. This study provides practical guidelines to healthcare organizations that are about to embrace the AI and HIS integration and serves as a foundation for more advanced future endeavors in AI supported patient care.

ABSTRACT. Fractured bones can require complicated restorative procedures due to the need for specialized biomaterials that can temporarily provide the necessary mechanical support while the body continues to heal biologically. Prefabricated scaffolds serve as essential supportive frameworks for engineers constructing bone tissues as they simulate the human extracellular matrix and aid the body in the construction of new tissue. This study focuses on the impact of scaffold structure on its mechanical properties by examining scaffold porosity using finite element analysis (FEA), in which a comparative analysis of porous as well as solid titanium alloy scaffolds was conducted. In ANSYS, potential scaffold design models were created and all models underwent identical loading scenarios. To examine how the presence of pores affects stress concentration, the deformation of the scaffold in terms of shape and structure, along with the Von Mises stress and strain, were the variables that were mainly observed when compared to the scaffold without pores it showed the highest stress value. This occurred because stress become concentrated near the edges containing pore at the point where the strut meet. Still the scaffold with pores showed the load distribution was uniform and these findings highlight the balance that must be achieved between the strength of the scaffold in bearing loads and its ability to interact effectively with living tissues.

ABSTRACT. Electrohysterography (EHG) is a noninvasive technique of measuring uterine electrophysiological activity, through which the functional maturation of the myometrium in pregnancy is assessed. This paper focuses on examining gestational agerelated changes in the electrical behavior of uterine during pregnancy, through the application of high-order signal decomposition and feature selection processes. The records of EHG were obtained in 24 pregnant women (22-43 weeks of gestation) in a BIOPAC MP 36 system under standardized clinical conditions. The records of each of the three channels were first processed by a bandpass filter (0.08-4.0 Hz) to isolate uterine activity and decomposed by Variational Mode Decomposition (VMD) via Continuous Wavelet Transformer (CWT) validation. Various temporal and spectral characteristics were calculated out of every mode, such as energy, root mean square (RMS), instantaneous frequency, and entropy. The statistical comparison demonstrated that the energy of the third VMD mode of Channel 1 (M3 energy) had a strong positive correlation with the gestational age (r = 0.4528, p = 0.0263), and it showed a significant progressive increase in the energy of the uterine electrical energy at term. The KruskalWallis test (p = 0.1877) was not significant because of a small sample, but the Cohen’s d indicated a statistically significant change in the position of physiological escalation of contractile readiness (week 37-38, d = -0.15). These observations indicate that VMD-based features have the capacity to represent significant electrophysiological maturation of the uterus and indicate their possible use as quantitative biomarkers of gestational monitoring and preterm risk.

ABSTRACT. Cardiovascular diseases remain the leading global cause of mortality, necessitating enhanced medical training methodologies. This study presents the development and evaluation of a cost-effective, multi-parameter patient simulator designed to address the limited accessibility of simulation-based medical education (SBME) in resource-constrained environments. The simulator incorporates three modular components: a CPR training module with force-feedback mechanisms, an ECG simulation module generating physiologically accurate waveforms, and a heart sound reproduction system with mobile application integration. Using an Arduino-based microcontroller architecture, the system achieved 95% cost reduction compared to commercial alternatives while maintaining clinical fidelity. Evaluation with 45 healthcare professionals demonstrated significant positive outcomes across all performance metrics (p < 0.001), with mean satisfaction scores ranging from 4.1 to 4.6 on a 5-point Likert scale. The real-time compression depth feedback was provided by the CPR module; the cardiac rhythms, both normal and pathological, are also successfully generated by the ECG module, and the heart sound module effectively reproduced the different normal and diseased heart sounds across five cardiac regions. This study demonstrates that high-quality medical training can be provided to medical students by developing low-cost human simulators.

ABSTRACT. Large Language Models (LLMs) are increasingly used for automated code generation, but most benchmarks emphasise correctness alone and overlook efficiency and maintainability. This study evaluates four representative models SantaCoder, CodeGen-350M, CodeT5-Base, and PolyCoder on the HumanEval benchmark using three dimensions: correctness (pass@k), runtime efficiency (time, memory), and code quality (complexity, nesting, linting). Experiments in a standardised GPU environment show that CodeT5-Base achieves the strongest overall balance, combining near-perfect pass@10 accuracy (1.000) with 6.600 s average runtime per task and 0.042 GB peak memory. By contrast, CodeGen-350M reaches pass@10 = 0.701 at 36.100 s and 0.259 GB, SantaCoder 0.195 at 23.700 s and 0.001 GB, and PolyCoder 0.591 at 42.300 s and 3.906 GB, underscoring that accuracy gains often trade off against latency and memory; efficiency and maintainability must therefore be assessed alongside correctness.

ABSTRACT. This paper introduces a novel terminal nonsingular sliding mode control (NTSMC) strategy tailored for quadrotor unmanned aerial vehicles (UAVs), addressing the challenges of nonlinear dynamics, system uncertainties, and external disturbances that affect flight stability and performance. Drawbacks of conventional sliding mode control (SMC) methods and terminal sliding mode (TSM) approaches, while robust against uncertainties, often suffer from issues like chattering, and the TSM suffers from singularities. These drawbacks can result in reduced control precision and performance degradation in real-world applications. The proposed NTSMC framework employs a carefully designed nonsingular terminal sliding surface, which ensures finite-time convergence without the occurrence of singularities. This feature is crucial in maintaining robust control during rapid maneuvers and in the presence of dynamic disturbances, as commonly experienced by quadrotors in complex environments. Additionally, the proposed control law incorporates continuous control signals to mitigate the chattering effect, which is commonly observed in traditional SMC designs, leading to smoother control actions and enhanced system performance. The NTSMC is rigorously derived to guarantee stability and finite-time convergence based on Lyapunov theory, ensuring that the quadrotor's attitude and position tracking errors are driven to zero within finite time. The effectiveness of the proposed method is evaluated through extensive simulations under various operating conditions, including trajectory tracking, aggressive maneuvering, and disturbance rejection. The results demonstrate that the NTSMC significantly improves control accuracy, robustness, and response speed compared to conventional SMC and other existing control strategies.