Robotic manipulation of deformable objects: a comprehensive review

Purpose Deformable objects are pervasive in manufacturing, service and medical domains, yet nonlinear, high-dimensional behaviour makes reliable robotic manipulation difficult. This paper aims to clarify progress in modelling, sensing, planning and control of deformable-object manipulation (DOM), addressing critical knowledge gap and guiding future research. Design/methodology/approach This paper provides a comprehensive review of recent research in robotic manipulation of deformable objects (DOM), with a particular focus on developments from the past five years. Studies are organized along a perception–planning–control pipeline, and their assumptions, metrics and experimental settings are compared to expose convergences and open problems. Findings Analysis shows that multi-modal perception fused with learning-augmented physics models is improving state estimation; hierarchical planners increasingly exploit environmental constraints and strain limits to prevent damage; and hybrid model-based/model-free controllers are extending autonomy to surgical industrial and collaborative scenarios, although safety-guaranteed learning and unified benchmarks remain scarce. Originality/value Unlike earlier domain-specific surveys, the authors explore advancements in non-prehensile dynamic manipulation, surgical applications and emerging areas such as human-robot collaborative tasks. By addressing these areas, the authors aim to provide a holistic view of the current state of DOM, identify key challenges and highlight future research opportunities that integrate human–robot interaction to advance the field.