This study presents a PRISMA-based systematic review of 343 publications focused on tracking how AI-based methods have evolved within robot manipulator control from 2015 to 2025. The review examines how AI has been incorporated into the control pipeline by organizing prior work according to functional roles, including perception and estimation, planning, learning-based control, interaction and safety, and learning and adaptation. In addition to this functional taxonomy, the study analyzes publication growth, application domains, robot types, evaluation settings, and methodological patterns to characterize the evolution of the field over the past decade. The results show that research activity has been concentrated primarily in learning control, while other functional roles have received comparatively less attention. The data also reveals an uneven distribution across application areas and robot platforms in the literature, with a strong reliance on simulation and limited evidence of integrated real-world deployment. These patterns indicate that, despite rapid growth and methodological diversity, the field remains imbalanced in both research focus and validation maturity. Rather than summarizing individual studies in isolation, this review provides a high-level perspective on where effort has been concentrated, where major gaps persist, and which directions are most critical for advancing AI-based robot manipulator control toward reliable and scalable real-world use.
Nguyen et al. (Mon,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: