Laban Movement Analysis Using a Bayesian Model and Perspective Projections

Human movement is essentially the process of moving one or more body parts to a specific location along a certain trajectory. A person observing the movement might be able to recognize it through the spatial pathway alone. Kendon (Kendon, 2004) holds the view that willingly or not, humans, when in co-presence, continuously inform one another about their intentions, interests, feelings and ideas by means of visible bodily action. Analysis of face-toface interaction has shown that bodily action can play a crucial role in the process of interaction and communication. Kendon states that expressive actions like greeting, threat and submission often play a central role in social interaction. In order to access the expressive content of movements theoretically, a notational system is needed. Rudolf Laban, (1879-1958) was a notable central European dance artist and theorist, whose work laid the foundations for Laban Movement Analysis (LMA). Used as a tool by dancers, athletes, physical and occupational therapists, it is one of the most widely used systems of human movement analysis. Robotics has already acknowledged the evidence that human movements could be an important cue for Human-Robot Interaction. Sato et al. (Sato et al., 1996), while defining the requirements for 'human symbiosis robotics' state that those robots should be able to use non-verbal media to communicate with humans and exchange information. As input modalities on a higher abstraction level they define channels on language, gesture and unconscious behavior. This skill could enable the robot to actively perceive human behavior, whether conscious and unconscious. Human intention could be understood, simply by observation, allowing the system to achieve a certain level of friendliness, hospitality and reliance. Fong, Nourbakhsh and Dautenhahn (Fong et al., 2003) state in their survey on 'socially interactive robots' that the design of sociable robots needs input from research concerning social learning and imitation, gesture and natural language communication, emotion and recognition of interaction patterns. Otero et al. suggest (Otero et al., 2006) that the interpretation of a person’s motion within its environment can enhance Human-Robot Interaction in several ways. They point out, that a recognized action can help the robot to plan its future tasks and goals, that the information flow during interaction can be extended and additional cues, like speech recognition, can be supported. Otero et al. state that body motion and context provide in many situations enough information to derive the person’s current activity.