A vision-based closed-loop framework for adaptive ergonomic human-robot collaboration

Abstract

Introduction Although collaborative robots can safely share workspaces with human operators, most collaborative workstations remain static and cannot adapt to changes in operator posture during task execution, potentially compromising ergonomics and interaction quality. This study presents a vision-guided closed-loop adaptive control framework that enables a collaborative robot to continuously adjust workpiece position according to the operator’s posture. Methods Using real-time vision-based human pose estimation, the system automatically modifies workpiece height and reach distance to support ergonomic working postures. The framework was evaluated with 30 participants performing screw-driving tasks under fixed and adaptive workstation conditions. Results Compared with the fixed workstation, the adaptive workstation significantly reduced Rapid Upper Limb Assessment scores, improved physical comfort, reduced perceived postural load, and increased trust and acceptance without compromising perceived safety. In addition, it reduced the influence of participant height on ergonomic performance, demonstrating improved anthropometric adaptability. Discussion These findings demonstrate that real-time posture perception can serve as an effective feedback mechanism for adaptive robot control, providing a practical framework for human-centered collaborative workstations that accommodate operator variability while improving ergonomic outcomes and user experience.