International Journal on Robotics, Automation and Sciences

Robotics and Automation, Computer Science, and Artificial Intelligence

2025-12-04T17:18:23+08:00

This thematic issue highlights the convergence of robotics and automation, computer science, and artificial intelligence as the foundation of modern intelligent systems. Advances in robotic design, control systems, and autonomous operation continue to expand the capabilities of automated technologies across diverse sectors. Concurrently, developments in computer science—including algorithmic innovation, distributed systems, and secure computing—provide the essential infrastructure that supports complex and scalable digital solutions. Artificial intelligence strengthens this ecosystem through machine learning, deep learning, and data-driven decision-making, enabling enhanced perception, prediction, and adaptability. Together, these fields drive transformative progress in industrial automation, smart systems, human–robot collaboration, and intelligent applications. This issue brings together research contributions that address theoretical advancements, practical implementations, and interdisciplinary perspectives, offering insight into the evolving landscape of intelligent and autonomous technologies.

Sentiment Analysis of Indonesian Nickel Downstreaming on X Using Naïve Bayes and K-Nearest Neighbors

2025-09-16T11:12:10+08:00

Nickel downstreaming has become one of Indonesia’s most prominent industrial policies, positioned as a pathway to economic growth and global relevance in the electric vehicle supply chain. Despite its ambitions, the policy has triggered intense debate on social media, where concerns about ecological damage and foreign dominance intersect with narratives of national pride. This study employs sentiment analysis to examine public perceptions of the policy through 337 tweets collected from X (formerly Twitter). Two machine learning algorithms, Naïve Bayes and K-Nearest Neighbors, were applied to classify sentiment into positive, negative, and neutral categories, followed by evaluation using confusion matrices, accuracy, precision, recall, and F1-score. The results show that negative sentiment dominates across both models, with Naïve Bayes achieving higher accuracy and recall, while KNN displayed strengths in precision and F1-score. Wordcloud analysis further revealed that positive sentiment is associated with industrial progress and national identity, negative sentiment emphasizes environmental risks and foreign control, and neutral sentiment reflects factual reporting of events. These findings confirm that nickel downstreaming remains a contested policy, viewed as an economic opportunity by some and as a source of social and ecological concern by many others. This study demonstrates the value of integrating sentiment analysis with policy research, as social media provides real-time insights into how citizens perceive government initiatives. The evidence highlights the importance of addressing environmental sustainability and equitable resource management to build trust and legitimacy. Sentiment analysis therefore serves not only as a tool for understanding public opinion but also as a guide for shaping more inclusive governance.

Manuscript received: 5 Aug 2025 | Revised: 8 Oct 2025 | Accepted: 15 Oct 2025 | Published: 30 Nov 2025

Hybrid Phishing Detection Model: Integrating BERT with TF-IDF for Enhanced Email Security

2025-08-19T14:12:10+08:00

Phishing emails remain a major cybersecurity problem because they cleverly exploit our natural trust by impersonating real messages. While standard NLP methods like TF-IDF and FastText are efficient, they often miss the subtle, contextual tricks found in today's sophisticated phishing attempts. On the other hand, advanced deep learning models like BERT are fantastic at understanding context, but they require a lot of computational power. In this paper, we suggest a hybrid solution. We merge the lightweight, statistical strengths of TF-IDF with the deep contextual power of BERT's embeddings to create a more robust phishing detection system. To test this, we ran experiments on datasets of 1,000, 5,000, and 10,000 emails, putting five different models head-to-head. Our results were clear: the hybrid models consistently beat the single-method ones. Interestingly, the TF-IDF + BERT combo was the most accurate on the smaller dataset (1,000 samples). However, for larger datasets (5,000 and 10,000 samples), TF-IDF + FastText offered the best balance of accuracy and speed. While the BERT hybrid was slightly more accurate, its slower processing time is a real hurdle for scaling up. We believe our proposed framework offers a practical and effective tool for real-world cybersecurity teams.

Manuscript received: 3 Jul 2025 | Revised: 25 Aug 2025 | Accepted: 7 Sep 2025 | Published: 30 Nov 2025

Forecasting High-Risk Traffic Zones Using Machine Learning for Enhanced Road Safety

2025-09-16T11:19:24+08:00

Road traffic accidents continue to pose serious global public health and economic challenges. In Malaysia alone, traffic-related incidents caused an estimated RM25 billion in losses in 2023. This study presents a two-part machine learning framework: Part A focuses on predicting accident severity, while Part B uses these predictions to forecast high-risk traffic zones through spatial and temporal analysis. Accident data from 2023 was selected from the UK Road Safety dataset to reflect current traffic patterns, infrastructure, and enforcement efforts. Five classifiers, Logistic Regression, Decision Tree, Random Forest, XGBoost, and K-Nearest Neighbors, were trained and evaluated. A stacking ensemble combining the top three models was constructed to enhance predictive accuracy. The models were assessed using accuracy, precision, recall, and F1-score, with results showing that the ensemble method outperformed individual classifiers. The findings demonstrate the potential of ensemble learning in identifying high-risk zones and supporting proactive road safety planning.

Manuscript received: 3 Aug 2025 | Revised: 21 Sep 2025 | Accepted: 28 Sep 2025 | Published: 30 Nov 2025

A Reproducible Benchmark of AdamW-Augmented Lightweight Models for Trash Classification

2025-07-20T15:01:33+08:00

Global waste generation is projected to reach 3.40 billion tonnes by 2050, creating urgent demands for automated waste classification systems that can overcome the limitations of manual sorting methods. Current deep-learning research on waste classification lacks standardised evaluation protocols, preventing meaningful architectural comparisons and hindering the progress of reproducible research. This paper establishes a reproducible benchmark framework for lightweight neural network models designed explicitly for trash classification research applications. Lightweight models are designed for optmised architecture and computation cost while maintain accuracy. Four representative lightweight models, including MobileNet V3 Large, Vision Transformer (ViT) Small, EfficientFormer, and ShuffleNet V2, were systematically evaluated on the TrashNet dataset using identical training protocols. All models employed AdamW optimisation with a learning rate of 1 × 10^-4, weight decay of 1 × 10^-4, and CosineAnnealingLR scheduling through 5-fold stratified cross-validation on RTX 2080 Ti hardware. Experimental results demonstrate that ViT Small achieved the highest classification accuracy at 0.815 but required 21.67M parameters, while MobileNet V3 Large delivered superior computational efficiency with 0.768 accuracy and 0.72ms inference time using only 4.21M parameters. Statistical analysis revealed significant performance differences across models (p = 0.0002), with hardware-aware architectural optimisations proving more critical than raw parameter reduction for computational performance on data centre GPU hardware. The standardised evaluation framework and open-source implementation provide rigorous baselines for advancing automated waste classification research.

Manuscript received: 12 Jun 2025 | Revised: 7 Aug 2025 | Accepted: 11 Aug 2025 | Published: 30 Nov 2025

Cyclone Nature Prediction with the help of a Customized SVM Model

2025-08-11T16:10:19+08:00

Efficiently predicting the nature of tropical cyclones through machine learning techniques has always posed a challenge in the quest to save human lives. While existing research has proposed various methods to accurately predict cyclone behavior and reduce its impact on humanity, this paper introduces a unique customized Support Vector Machine (SVM) model. Unlike existing models, this machine learning-based custom model enhances evaluation metrics, offering significant improvements in binary classification forecasting. The paper also presents a schematic diagram outlining an architectural design for cyclone nature detection utilizing satellite images. The proposed customized SVM model achieves impressive classification metrics, with accuracy at 95%, precision at 94.78%, recall at 94.5%, and an F1-score of 94.9%. In contrast, other models such as Random Forest (RF), SVM, decision tree (DT), and Logistic Regression (LR) fall short, failing to reach an accuracy exceeding 92%. Furthermore, future work may involve the development of hybrid models.

Manuscript received: 3 May 2025 | Revised: 30 Jun 2025 | Accepted: 13 Jul 2025 | Published: 30 Nov 2025

Deep Learning-based Obstacle Detection for Human Interaction Robots: A Review

2025-07-03T11:53:51+08:00

Obstacle detection is the foundation of autonomous robotics, enabling robots to perceive and understand the world around them to move safely. Deep learning has emerged as one of the driving forces in today’s research, with various algorithms employed for learning and making effective decisions based on vast and complex datasets. In recent years, numerous deep learning methods have been developed and studied to detect obstacles. This paper provides an end-to-end overview of over 40 state-of-the-art deep learning models (from 50 papers) for obstacle detection in human-interacting robots, with a focus on deployment viability, real-time running, and energy efficiency. We also delve into the architecture of deep learning, highlight key challenges in real-world deployment, offer a comparative analysis of basic and advanced deep learning approaches, and examine the trade-offs between accuracy, speed, and power consumption, providing insights into practical considerations. This review categorizes obstacle detection techniques into two groups: Core CNN-based methods and Advanced Deep Learning Methods. Comparisons were made between these two groups, concentrating on computational requirements, deployment feasibility, and hardware configuration. Several key findings emerged. It was determined that models with high accuracy were computationally expensive and unsuitable for embedded deployment. While some models experience accuracy-speed trade-offs, others are limited by hardware constraints and power limitations. Finally, this review concludes with a structured discussion of real-world deployment considerations, prioritizing model efficiency, scalability, and potential future research directions in deep learning-based obstacle detection.

Manuscript received: 30 Jun 2025 | Revised: 28 Jul 2025 | Accepted: 11 Aug 2025 | Published: 30 Nov 2025

Simulation based Analysis of Encoder Resolution on Differential Drive AMR Odometry

2025-07-29T16:49:29+08:00

This research explores the impact of encoder resolution on the odometry accuracy and navigational performance of a differential-drive Autonomous Mobile Robot (AMR), using the Automated Trash Mobile Robot (ALTO) as a test platform. Encoder pulse-per-revolution (PPR) values ranging from 40 to 4096 were simulated in Gazebo. A custom encoder and odometry simulation algorithm were developed and integrated into the ROS1-based navigation stack. Controlled experiments—including straight-line, rotational, and dynamic path tests—were conducted in virtual environments to compare positional accuracy using /odom, /amcl_pose, /global_pose, and /world_pose. Results showed that higher PPR values improved odometry precision, particularly in orientation estimation, but had limited influence on global pose accuracy under AMCL-based sensor fusion. While lower resolutions caused noticeable drift, AMCL maintained robust localization. The findings offer practical guidance for optimizing encoder selection, balancing cost and performance in industrial AMR deployments.

Manuscript received: 30 Jun 2025 | Revised: 8 Aug 2025 | Accepted: 16 Aug 2025 | Published: 30 Nov 2025

The Application of Hybrid Renewable Energy Systems

2025-07-09T11:31:13+08:00

Hybrid Renewable Energy Systems (HRES) integrate solar, wind, and other renewable energy to deliver more sustainable, dependable, and affordable energy for rural, urban, and industrial areas. Based on 20 articles/journal from 2020–2025 that were taken from Google Scholar, IEEE Xplore, and Scopus, this paper evaluates HRES applications, technologies, barriers, and future development. Storage will increase to 204.47 GW, when solar and wind power dominate with capacities increased by 937% and 118% throughout 2014 to 2020. Optimization tools like HOMER Pro and Particle Swarm Optimization (PSO) can reach up to 1.10% error in energy predictions. HRES can reduce costs and emissions by 86% (solar) and 61% (wind) by prioritizing renewable energies usage. Regulatory loopholes, intermittency, and high initial costs are some of the challenges in the application of HRES. MATLAB visualizations show capacity trends and cost reductions, which supports economic viability. Examples that demonstrate sustainability and highlight reliability include mining activities in Iran and microgrids in Makkovik, Canada. This paper identifies HRES based on the literature, AI, IoT, and policy incentives. Future advancements must go beyond technical constrains and standardize regulations to scale HRES for global energy transformations, smart cities, mining industries, and resilient communities.

Manuscript received: 16 Jun 2025 | Revised: 29 Jul 2025 | Accepted: 10 Aug 2025 | Published: 30 Nov 2025

Development of an Autonomous Vacuum Cleaner Robot: CASARICA

2025-07-03T13:17:09+08:00

This article introduces CASARICA, an autonomous vacuum robot whose aim is to counteract the long process of manual household cleaning, particularly for Indonesia's urban regions. With the use of an Arduino Mega 2560 microcontroller, CASARICA pairs ultrasonic, infrared, and dust sensors to independently move around and effectively sweep away dirt. Powered by a 5000mAh power bank, it works for 8 hours, employing a roller brush and DC motors to suck in dust in an onboard waste bin. Even when in the process of testing, it has been known to survive environments both with smooth ceramics or wooden floor, navigate through obstacles, and find high-dust areas with little user effort. Tests showed strong performance in battery life, mobility, and vacuuming which makes CASARICA to be a low-cost option for easier and automated cleaning.

Manuscript received: 15 Jun 2025 | Revised: 29 Jul 2025 | Accepted: 10 Aug 2025 | Published: 30 Nov 2025

Laser Induced Thermal Effects and Accuracy in Polycarbonate Cutting

2025-06-14T17:06:13+08:00

This research investigates into the CO2 laser cutting process applied to polycarbonate materials of varying thicknesses. Investigating laser power, cutting speed, standoff distance, and cutting diameter, the study focuses on responses such as heat-affected zones and kerf diameters. Through advanced microscopy and coordinate measuring tools, interaction effects were assessed using variance analysis. Measurements were made using Meiji Techno MT7000 Metallurgical Microscope and the CRYSTA-Apex S 900 CNC Coordinate Measuring Machine. Interactions effects of data were calculated by analysis of variance. Notably, higher cutting speeds coupled with lower laser power yielded optimal heat-affected zones. Standoff distance emerged as a critical factor influencing material cut-through capacity. The results show that optimum levels of heat affected zone were possible by applying higher cutting speeds and lower laser power. Standoff distance had the most impact on the ability of the material to be cut through.

Manuscript received: 14 Jun 2025 | Revised: 19 Aug 2025 | Accepted: 30 Aug 2025 | Published: 30 Nov 2025

A Narrative Review of Data Mesh Architecture Principles and Implementation Outcomes

2025-07-17T12:19:04+08:00

Centralised data architectures often create operational bottlenecks that limit organisational agility. Data Mesh offers a distributed alternative through domain ownership and federated governance. This narrative review synthesises 52 sources published between 2001 and 2024, examining the evolution from traditional data architectures to Data Mesh implementations across financial services, healthcare, e-commerce, and technology sectors. The review traces the progression from centralised data warehouses through distributed computing frameworks to Data Mesh's emergence, identifying four foundational principles domain-oriented decentralisation, data as a product, self-serve infrastructure, and federated governance. Analysis of recent implementation studies reveals mixed outcomes. Successful adoptions demonstrate improved domain autonomy and reduced central bottlenecks. However, multiple case reports significant coordination complexity and extended implementation timelines, with transformations requiring substantial investments in platform engineering. Consistent challenges emerge, including skill gaps in domain teams transitioning to data ownership, policy conflicts in federated governance structures, infrastructure investments that exceed traditional architectures, and cultural resistance to distributed accountability. Implementation success correlates with existing DevOps maturity, sustained executive sponsorship, phased adoption approaches, and robust metadata management capabilities. The review identifies critical research gaps in standardised success metrics, quantitative failure analysis, privacy-preserving techniques for federated environments, and long-term sustainability assessment. Based on the analysed cases, Data Mesh appears most suitable for large enterprises with diverse data domains and established platform engineering capabilities. Smaller organisations may find centralised approaches more appropriate given the complexity and resource requirements of distributed architectures. This synthesis provides practitioners with evidence-based insights while highlighting priorities for future research.

Manuscript received: 9 Jun 2025 | Revised: 24 Jul 2025 | Accepted: 30 Jul 2025 | Published: 30 Nov 2025

Design and Implementation of an Arduino-Based Ultrasonic Device for Humane Dog Repellent

2025-06-30T11:35:18+08:00

Dogs can pose a nuisance and danger to people in residential areas through barking and territorial behaviour, causing discomfort and safety concerns. Dogs and humans both possess hearing capabilities, but dogs can detect ultrasonic frequencies that humans cannot perceive. This enhanced auditory sensitivity makes dogs responsive to high-frequency acoustic stimuli. In this paper, preliminary field observations of an Arduino-based ultrasonic dog deterrent device are presented to explore frequency response patterns in free-roaming dog populations. A frequency-based approach represents a potentially more environmentally safe alternative compared to traditional chemical repellents. This research presents observational data from field testing of a portable prototype that incorporates an Arduino Uno microcontroller, an ultrasonic transducer, and an amplifier to generate adjustable high-frequency sound waves. The microcontroller enables frequency control across the 38 to 42 kHz range to emit an ultrasonic sound that dogs respond to without physical harm. The device is portable, offers frequency adjustability, and is capable of field deployment. Based on the observations from forty encounters with stray dogs, the response rates increased across the frequency range. Across the frequency range, 42 kHz showed the highest observed response. These findings suggest that ultrasonic deterrent applications show promise. Further research is needed to confirm the effectiveness and optimal deployment parameters.

Manuscript received: 9 Jun 2025 | Revised: 14 Jul 2025 | Accepted: 21 Jul 2025 | Published: 30 Nov 2025

Multi-objective and Multi-disciplinary Optimization of Vertical Axis Wind Turbine Blades

2025-07-03T13:46:59+08:00

The demand for renewable energy is increasing, leading to more research on Vertical Axis Wind Turbines (VAWTs) because they can be used in cities and rural areas. This review looks at the latest methods for improving the design of VAWT blades, to summarize the advancements in the multi-objective optimization and to highlight the interdisciplinary nature of the research, encompassing aerodynamics, materials science, and structural mechanics. It examines important factors like how air flows around the blades, their strength, and the materials used. The review also identifies gaps in current research and suggests future study directions. The goal is to enhance VAWT performance for better energy capture and use in various environments, especially where wind speeds are low. This research is crucial for advancing VAWT technology and making renewable energy more accessible and efficient. Aerodynamic performance remains a key focus, with computational fluid dynamic being the dominant method used for analysis. A few of the literature review findings are AI and machine learning are valuable tools for optimization but require validation. The structural and material innovations are advancing but need to be integrated with aerodynamic studies. Sustainable materials and manufacturing techniques are underexplored in the context of multi-objective optimization.

Manuscript received: 6 Jun 2025 | Revised: 20 Jul 2025 | Accepted: 11 Aug 2025 | Published: 30 Nov 2025

Effects of Composition and Processing on the Properties of Sn-3Zn-4Bi and Sn-Ag-Cu Solder Alloys for Electronic Packaging

2025-10-15T16:27:52+08:00

The mechanical properties of lead-free solder alloys are critical for ensuring the reliability of electronic packaging, with shear strength and hardness being particularly important as electronic devices become smaller and interconnection densities increase. Thermal fluctuations and external mechanical impacts further intensify shear stresses on solder joints, raising concerns about long-term performance. In this study, the shear stress behavior of Sn-Ag-Cu and Sn-3Zn-4Bi solder joints was examined under different reflow temperatures. Sn-Ag-Cu, a widely researched lead-free solder, demonstrated strong resistance to high stress levels, reinforcing its suitability for high-reliability applications; however, its relatively high melting temperature (~221 °C) limits its use in low-temperature reflow processes. By comparison, Sn-3Zn-4Bi solder, with a melting temperature only ~12 °C higher than eutectic SnPb solder, showed potential for low-temperature soldering, while also exhibiting higher microhardness values than Sn-Ag-Cu, suggesting improved structural robustness. Despite these advantages, concerns remain regarding its compatibility with copper substrates, where interfacial reactions may affect joint integrity. Overall, the results suggest that Sn-Ag-Cu is preferable for applications requiring high strength and thermal resistance, whereas Sn-3Zn-4Bi offers notable benefits for low-temperature processing, provided substrate interactions are properly managed.

Manuscript received: 15 Aug 2025 | Revised: 30 Sep 2025 | Accepted: 4 Oct 2025 | Published: 30 Nov 2025

Physical, Online, or Hybrid? A Study on the Preferred Mode of Learning of Multimedia University (MMU) Students

2025-07-17T12:32:12+08:00

Higher institutions in Malaysia, including Multimedia University (MMU), has adopted online and hybrid learning mode for its students during the Covid 19 pandemic and post pandemic. At present, this practice is still on going, apart from the physical learning mode norm. This sparks an interest in determining the preferred mode of learning of MMU students and is the basis for this study. A total of 363 respondents from different faculties across both Melaka and Cyberjaya Campus partake in this study. Several tests were conducted on the data collected, including Cronbach Alpha, Pearson Correlation, and multiple linear regression (MLR). Results indicated the survey instrument used was reliable across all variables. Weak relationships were found among all predictors to the three preferred learning modes. Albeit this, the MLR tests were conducted. In conclusion, upon comparing the results, it was determined that the preferred learning mode of MMU students is the online mode.

Manuscript received: 18 Jun 2025 | Revised: 30 Aug 2025 | Accepted: 5 Sep 2025 | Published: 30 Nov 2025

Intrusive and Non-Intrusive Techniques for Blood Sugar Measurement: A Practical Review

2025-08-27T14:15:12+08:00

Measurement of sugar levels in blood is the main means of diagnosing for diabetes and other complications of blood sugar levels. The established principle of the methodology for this is the extraction of blood from the subject and submission of the blood sample to chemical tests that determine the presence of substances, such as glucose, that indicate blood sugar levels. This principle is inherently intrusive; R&D into methods with this principle has the goal of improving convenience and minimizing amount of sampling needed, while maintaining reliable accuracy. There is also R&D into developing non-intrusive methods that estimate blood sugar levels without blood sampling, with the aim of producing results that can be comparable with intrusive methods. The common goal of either approach is making blood sugar measurement more convenient for as many people as possible. At this time of writing, non-intrusive methods have yet to replace the gold standard. A breakthrough in this matter can facilitate the implementation of machine learning in interpreting blood sugar levels.

Manuscript received: 7 Aug 2025 | Revised: 11 Sep 2025 | Accepted: 16 Sep 2025 | Published: 30 Nov 2025

SmartRecruit: A Fuzzy Rule-Based Expert System for Candidate Screening and Ranking

2025-05-13T16:37:11+08:00

Human Resource (HR) management plays a pivotal role in organizational success, with recruitment being one of its most critical functions. In recent years, the integration of Artificial Intelligence (AI) into HR processes has gained significant attention, particularly in automating recruitment to enhance efficiency and reduce biases. While AI-driven systems have demonstrated advanced capabilities, many lack adaptability across diverse job roles and often fail to provide transparency in decision-making. This research addresses these limitations by proposing a novel fuzzy A Fuzzy Rule-Based Expert System for Candidate Screening and Ranking (SmartRecruit). The system evaluates candidates based on key parameters such as skills, educational qualifications (e.g., CGPA), and work experience, offering an efficient, unbiased and transparent approach to hiring.

Manuscript received: 9 Apr 2025 | Revised: 12 Jun 2025 | Accepted: 30 Jun 2025 | Published: 30 Nov 2025