Lecture Notes in Computer Science 1 Gesture Analysis: Invariant Laws in Movement Sylvie Gibet, Jean-François Kamp, Franck Poirier Valoria, Université de Bretagne Sud, Campus de Tohannic, rue Yves Mainguy, F-56000 Vannes, France {Sylvie.Gibet, Jean-Francois.Kamp, Franck.Poirier}@univ- ubs.fr Abstract. This paper presents gesture analysis under the scope of motor control theory. Following the motor program view, some studies have revealed a num- ber of invariant features that characterize movement trajectories in human hand- arm gestures. These features express general spatio-temporal laws underlying coordination and motor control processes. Some typical invariants are described and illustrated for planar pointing and tracing gestures. We finally discuss how these invariant laws can be used for motion edition and generation. 1 Introduction With the massive development of Human-Computer Interaction (HCI), new systems try to take advantage of the expressive power of gestures. At first, gesture interaction has been reduced to simple command interfaces. More recently, techniques such as capturing body movements, recognizing and interpreting human actions, and animat- ing virtual humans, have given rise to a number of virtual reality applications with more natural interfaces. In such applications, the produced gestures should be per- ceived as natural and should follow biomechanical or psychomotor laws characteriz- ing human movement. This paper presents gesture analysis results under the scope of motor control the- ory. The analysis leads to the identification of significant parameters that characterize some classes of gesture (simple pointing, point-to-point and rhythmic gestures), ex- tracted from real data in the Cartesian space. The main issues are regarded as a way to understand the mechanisms underlying the control and coordination of movement, both in terms of general principles involving central processes of task planning and in terms of peripheral processes emerging from the bio-mechanical system. We examine the principles that can be expressed as invariant features in movement trajectories, the most well-known ones being the Fitt’s law and the Two-Third Power law. The over- view is restricted to planar pointing and tracing movements, with patterns which are not previously learnt. We don’t intend to demonstrate that these laws prefigure organ- izational principles explicitly used by the Central Nervous System (CNS) to produce movement, but we give some elements that can be used for motion edition or genera- tion in Human Machine interaction or computer animation.