Follow and avoid a walking human being using real-time 3D depth flow generation

This paper proposes a walking human being avoidance and following under complex environment using 3D Depth Flow Generation method. 3D Depth Flow measures 3D motion vector of every pixels between two time sequential images. At first, real-time 3D Depth Flow Generation method by enhancing recursive correlation algorithm and online spacio-temporal consistency checking method are denoted. Then ego-motion estimation method, mobile obstacle or human avoiding method, and mobile human being following method are denoted. Furthermore, as an application, walking human avoidance and following task using mobile robot by combining with real-time 3D Depth Flow generation system is denoted.