Human trajectory prediction with machine learning: An integrated systematic review and bibliometric analysis (2015-2025)

This paper presents an in-depth bibliometric analysis of the evolving landscape of human trajectory prediction through the lens of machine learning (ML), leveraging both bibliographic coupling and keyword co-occurrence methodologies. Drawing from a curated corpus of high-impact publications indexed in Scopus from 2015 to 2025, the study unveils the intellectual and conceptual structure of the field. The bibliographic coupling analysis identifies five dominant thematic clusters: (1) sensor-driven human activity and mobility monitoring, (2) deep learning approaches for pedestrian and crowd trajectory forecasting, (3) autonomous vehicles and intelligent driving behaviors, (4) multi-agent coordination and vehicle-to-vehicle dynamics, and (5) vision-based trajectory estimation and behavioral understanding. In parallel, the co-occurrence analysis reveals four emergent conceptual clusters centered around: (1) autonomous systems and environmental perception, (2) human-centered motion modeling, (3) predictive artificial intelligence (AI) and spatiotemporal learning, and (4) cognitive modeling and adaptive learning architectures. The findings underscore a paradigm shift toward hybrid deep learning frameworks, socially-aware multi-agent models, and explainable trajectory forecasting systems. Furthermore, the study identifies key future directions including graph-based motion prediction, privacy-preserving analytics, and real-time edge-AI deployments. This bibliometric roadmap serves as a foundational reference for scholars, system architects, and policymakers engaged in advancing intelligent mobility systems and human-machine interaction in complex environments. Beyond intelligent mobility systems, the findings offer direct implications for educational environments, particularly in smart campuses and learning analytics. ML-based human trajectory prediction can support student flow optimization, campus safety monitoring, and adaptive learning space design. This study therefore provides a conceptual and practical foundation for integrating predictive mobility analytics into educational and institutional decision-making.