Human-robot Friendly Interaction Using Facial And Body Motion For A Humanoid Robot

For a long time, humans get used to interacting with others in terms of voice, facial expressions, body movement, and body contact. A humanoid robot should have human-habitual, natural, and human-like interaction form in order to be widely used in human environment. Up to now, most of the existing humanoid robots are difficult to interact with humans in a human-like way. On the one hand, the interactive mode is not so complicated to meet the demands of social communication; on the other hand, it is difficult to ensure the safety of humans when interacting between a human and a humanoid robot. Therefore, this dissertation focuses on this issue and develops a friendly interactive system with whole-body movement on a humanoid robot. It is supported by the projects from National High-Tech Research and Development Program of China and National Natural Science Foundation of China. The main research issues and results are summarized as follows.First, based on the mechanism and procedure of human facial expressions, this dissertation adopts motion capture system to acquire and analyze the motion data of human face, and obtains the active and passive relationships of human facial motion points. Then, based on the active and passive relationships of human facial motion points, an active drive points(ADPs) model is proposed for facial expression with less active degree of freedom, this method overcomes the challenge of excessive DOFs caused by complex head model.Second, in order to overcome the unnatural interactive issue of the current humanoid robots, this dissertation presents an interactive method for humanoid robots based on kinematics and dynamic mapping of human interactive data, according to the captured motion data of human body, take the number of DOFs, motion range of every joint, the ground contact constraint into consideration, the human interactive data are then mapped to obtain the interactive data of a humanoid robot.Third, a new flexible controller that considers the interaction between visual and force information for a humanoid robot is designed; the proposed method can meet the demands of accurate safety interaction. Also, in order to overcome the difficulty of unbalance issue during the whole-body motion of a humanoid robot, this dissertation presents a coordinated whole-body motion control method based on visual information to enhance the safety of the interaction and guarantee the balance of a humanoid robot.Finally, the effectiveness of the proposed methods was validated through experiments on a humanoid robot. The recognition experiments of facial expression verified the proposed active drive points method; the body language interaction experiment verified the effectiveness of the proposed body language interactive system; the physical interaction between human and robot experiment verified the effectiveness of proposed physical contact interactive method.