“…Motion control in such environment is still a challenging problem since the working environment models are not known a priori and sensing is limited and uncertain. There are many methods proposed to solve this problem, such as 3D information retrieval including multi-view geometric techniques using single movable camera [12,13], simultaneous localization and Mapping [14], visual servoing with uncalibrated cameras [15,16], However due to the limitation of visual sensor or strict requirements for prior knowledge of robot dynamics, they cannot meet the requirements for high precise and flexible robot teaching applications [6,17],…”