MPC-SCBF: Model Predictive Control with Social Zone-Driven Control Barrier Function for Socially-Aware Following Robots

Embodied intelligence has improved human-robot interaction in human society. However, human-following behaviors of socially-aware robots pose challenges to human comfort. In this study, a Model Predictive Control with Social zone-driven Control Barrier Function (MPC-SCBF) is proposed to achieve the robot avoiding intrusions into the intimate zone of human targets and other pedestrians during obstacle avoidance, thereby enabling safe and comfortable navigation. This method simultaneously optimizes three objectives: maintaining the comfort of both human targets and other pedestrians, generating natural robot motion, and achieving socially compliant obstacle avoidance. Experimental results demonstrate that the proposed method effectively achieves human following and obstacle avoidance in dynamic environments.