PUB - Publikationen an der Universität Bielefeld

The focus of this thesis is the nonverbal communication via robotic and human whole-body motions in HRI. Humans have the natural ability to nonverbally communicate with each other via whole-body motion and are very able to interpret and predict motions of others, e.g., in order to coordinate themselves with others in narrow environments. This thesis investigates the communicative potential of whole-body motions not only on the robotic side of conveying motion but also on the human side in HRI. To observe human interpretation of robot motion and to observe human communicative whole-body motion in interaction, two studies have been conducted. A result of the studies is that abrupt motions and motions which are bounded with pauses are clearer to interpret. A human interprets robotic motions as reaction to her / his own presence if the robotic motion is abrupt, e.g., humans attribute intentional behaviour more to a driving robot, which stops or to a robot which has not been moving and then moves compared to a robot which has been moving constantly. Another important result of the studies is the creation of a coding scheme based on a detailed description of human whole-body motions and their associated characteristics. Small steps are conveyed and bounded with pauses. Characteristics of motion (trajectory, distance, position, orientation) are analysed as to whether they form any patterns that represent specific communicative whole-body motions. This thesis discusses that a mere representations of two moving objects is not enough for a robot to interpret communicative whole-body motions. A relative and qualitative framework of representing motions and its advancements are used and discussed. This thesis identifies the need for a system to represent possible interpretations or meanings according to the observed communicative motions for robots. This PhD thesis is a fundamental and interdisciplinary work and the basis for different future projects which have been already initiated. This thesis provides pieces of advice of how a robot should cope with the communicative power of whole-body motions in HRI. A robot should: a) consider human communicative whole-body motion and ask for its meaning. b) be aware of the own motion. c) express goal-directed actions.