The integration of Cyber-Physical Systems (CPS), Digital Twins (DT), and robotics with Advanced Data Science Platforms (ADSP) is rapidly transforming industrial and research landscapes by enabling real-time data processing, intelligent decision-making, and enhanced automation. Over the past decades, with the growing demand for adaptable, expandable, and secure platforms, numerous groundbreaking studies have emerged in the field of ADSP for CPS, DT, and robotics. However, most existing research addresses isolated aspects of AI, CPS, DTs, or robotics, lacking a holistic view of how ADSP expands and optimizes these technologies. To address this gap, this survey provides a comprehensive and structured review of the existing methodologies, tools, and techniques at each step of the ADSP workflow, focusing on their applications in CPS, DT, and robotics. Firstly, this survey analyzes the data ingestion phase, focusing on raw data collection architecture, advanced data pre-processing techniques, and data lake integration, while identifying integration and scalability challenges. Secondly, the development methodologies and data analysis techniques within experimental workflows are explored by highlighting widely used tools and real-world case studies. Thirdly, a detailed overview of optimization techniques and deployment strategies is presented, including cloud, edge, and hybrid models, supported by practical deployment examples. Finally, the continuous learning mechanisms are investigated for adaptive system updates, challenges in real-time adaptation, and expanding model performance. Additionally, this survey focuses on evaluation metrics, benchmark studies, and performance comparisons to assess platform efficiency across various domains. This study also explores emerging challenges such as data quality & availability, platform expandability, model transparency, and security, offering insights into future research directions. To ensure the rigor and breadth, this survey is conducted on 319 high-quality studies, state-of-the-art methodologies, and platforms that are selected using the PRISMA methodology from over 600 reviewed publications. By presenting a structured overview of ADSPs, this work aims to serve as a valuable resource for researchers, engineers, and industry professionals aiming to utilize data science for innovation in CPS, DT, and robotics.

Depression among university students is a critical global problem that affects academic performance, well-being, and dropout rates. Although machine learning models offer strong predictive power, their opacity limits their adoption in educational and clinical contexts. Most existing approaches emphasize either global or individual explanations, leaving a gap in multi-level interpretability that can inform institutional policy and personalized support. This paper introduces a novel multi-level explainable AI (XAI) framework that unifies SHAP for global importance ranking, H-LIME for hierarchical subgroup explanations, and counterfactual reasoning for actionable individual recourse. By establishing a generalizable theory of hierarchical interpretability that bridges global, group, and local levels of explanation, the framework advances both methodological rigor and practical relevance. Using a dataset of 27,901 students with demographic, academic, lifestyle and psychosocial factors, Random Forest achieved a strong predictive performance (accuracy = 0.839, ROC–AUC = 0.916). The results consistently identified suicidal ideation, academic pressure, and financial stress as dominant factors, while subgroup and individual analyses revealed contextual and personalized nuances that global explanations alone obscure. Individual-level counterfactuals further suggested feasible workload or lifestyle adjustments, offering practical recourse for at-risk students. Limitations include dependence on self-reported survey data and the fact that the sample is from a single country, which may limit the generalizability of the findings. This unified approach offers a reusable methodological framework for achieving multi-level interpretability in sensitive AI applications. It also provides evidence-based guidance for the early identification of at-risk individuals, customized interventions, and the development of equity-based institutional policies.

The rapid advancement of ICT, robotics, data science, and cyber-physical systems (CPS) has intensified the need for agile, future-ready human resource development (HRD) strategies. This scoping review explores current HRD practices, challenges, and opportunities across these technological domains. Guided by the PRISMA-ScR framework, 791 records were initially retrieved from four databases, with 43 peer-reviewed publications ultimately included following a rigorous screening process. Thematic analysis revealed a strong emphasis on digital and AI literacy, interdisciplinary training, and the adoption of immersive technologies such as virtual reality and AI-driven competency tracking systems. Persistent gaps, including misalignment between academic training and industry needs, a lack of scalable HRD models, and underdeveloped policy frameworks. Statistical comparisons revealed no significant differences across fields, indicating the cross-cutting relevance of the identified strategies. The findings underscore the need for structured, empirically grounded, and industry-integrated HRD approaches to prepare the workforce for the demands of the Fourth Industrial Revolution.

This study introduces a Preference-Based Reinforcement Learning (PbRL) approach tailored for autonomous vehicle (AV) applications within a simulated environment. Traditional RL methods often struggle with the complexities of reward function engineering, failing to perform behaviors of human desire. The proposed framework integrates human preferences directly into the training loop, our framework offers a novel methodology for enhancing the decision-making processes of autonomous system. Our results demonstrate that PbRL can refine the strategies of AVs to align more closely with human-like decision-making, highlighting the potential for increased adaptability and safety in autonomous technologies.

Field robots deployed in subsea, planetary, disaster-response, subterranean, and industrial environments frequently operate beyond nominal design assumptions, where environmental stressors degrade hardware, sensing, communication, and the assumptions underlying autonomous operation. Existing surveys typically organize harsh-environment robotics by application domain, limiting comparison of the cross-domain mechanisms through which exposure propagates into system-level failures. This survey addresses that gap through an environment-centered, evidence-coded synthesis of 348 records spanning field deployments, robotic experiments, component-level investigations, standards, and domain-specific surveys identified primarily in database searches (2015–May 2026), with supplementary sources. We classify environmental stressors into five categories, extreme physical, liquid/moisture-related, electromagnetic/radiation, chemical, and biological, with explicit evidence-basis labels distinguishing field observations, robotic tests, component evaluations, and proxy evidence, rather than assuming universal thresholds. Building on this taxonomy, we propose a qualitative synthesis lens, not a predictive or fitted model, connecting environmental exposure through hardware degradation or direct changes in operating conditions to autonomy-level assumption violations and mission-level reliability loss, applied to seven domain-based case syntheses spanning deep-sea, planetary, subterranean, nuclear, disaster-response, construction/mining, and defense robotics. We close with a three-tier benchmarking protocol, digital-twin simulation, hardware-in-the-loop testing, and controlled field validation, specifying the minimum reporting elements needed to make future robustness claims comparable across studies.