Human-robot collaboration empowered by hidden semi-Markov model for operator behaviour prediction in a smart assembly system

Human-robot collaboration (HRC) in assembly scenarios, particularly a human center HRC system, demands the robot to react smartly. How the robot should respond in time to collaboration requests with anticipation of human behavior instead of waiting for visible cues is challenging. This study proposes a hidden semi-Markov model (HSMM) for a human center HRC assembly system that enables the robot to accurately predict the human assembly rate. The HSMM model links different assembly rates (visible states) with distinct operator perceptions (invisible states). Thus, the robot responds by adjusting to the predicted human behavior using the proposed HSMM. The proposed model was verified through a simulation of the collaborative scenario on a graphics processing unit assembly line. The results indicate that the proposed HSMM-based HRC plan significantly reduces the assembly cycle time and operator waiting time. A successful partnership between humans and robots in a smart assembly system using the proposed recognition intelligence and precise predictions was achieved in this study.