User Interaction for Guided Learning Supporting Object Recognition in Service Robots

Under current technical conditions, robots often have difficulties or make errors in object recognition due to the complexity of the scene or the particularity of the object to be recognized. It is necessary for users to provide input or guided training for the robot during the recognition process. However, most of the input methods commonly used by researchers, is limited to using mouse and keyboard to mark outer and inner edges of the object on the screen. We introduce in this paper a survey of possible actions and input methods for designing human robot interactions supporting guided learning in object recognition. Specifically, we analyzed key factors and procedures in analyzing typical object recognition and proposed human robot interaction methods appropriate for each type of obstacle.