Algorithms and graphic interface design to control and teach a humanoid robot through human imitation

Abstract

Given that human-like robots are finding their place in different areas of our world, research is being carried out in order to improve human-robot interaction. Humanoid robots not only require a human appearance but also require human-like movements. Along these lines we present this project which tries to give a solution to the question of human to robot arm mapping with the objective to control in real time and to teach a robot by imitation. For the technical implementation we have worked with a SR3000 ToF camera to sense the human movements which allows us to perform a markerless arm tracking based on proximity information. The robot used is a Robotis Bioloid; still, the software has been designed to be adapted with few modifications to other robots having similar arm structures.