A complexity scoring framework and its effect on the human-likeness of robotic arm movements

Abstract In the context of Industry 5.0, the ability of robotic arms to perform human-like movements is crucial for enhancing Human-robot collaboration. These movements allow robots to operate more intuitively in shared workspaces, while improving safety, precision, and efficiency. This human-robot synergy not only boosts productivity but also enables personalized manufacturing, where robots assist workers in creating customized products with greater flexibility and minimal errors. Previous studies indicate that the greater the complexity of a movement, the more challenging it can be to generate smooth, human-like, and collision-free trajectories. Further, while much research has addressed how to evaluate the human-likeness of robotic movements, the impact of task complexity on trajectory generation has received limited attention. To bridge this gap, the present work introduces a novel scoring framework that quantifies the complexity of arm movements in industrial environments. The score is derived from seven individual metrics, weighted according to their relevance, and designed to reflect spatial and kinematic properties of motion. Using data from two representative tasks, we analyzed the relationship between the proposed complexity score and established human-likeness metrics. Moreover, the human-likeness and complexity of the movements of the robot is compared against human movements in an industrial setting. The results indicate that movements with greater displacement and more frequent reorientations tend to exhibit higher complexity. Also, the movements performed by the robot closely resemble those of the human, providing evidence of the human-likeness of the robotic motions.