Identifying and modeling motion primitives for the hydromedusae Sarsia tubulosa and Aequorea victoria

The movements of organisms can be thought of as aggregations of motion primitives: motion segments containing one or more significant actions. Here, we present a means to identify and characterize motion primitives from recorded movement data. We address these problems by assuming that the motion sequences can be characterized as a series of dynamical-system-based pattern generators. By adopting a nonparametric, Bayesian formalism for learning and simplifying these pattern generators, we arrive at a purely data-driven model to automatically identify breakpoints in the movement sequences. We apply this model to swimming sequences from two hydromedusa. The first hydromedusa is the prolate Sarsia tubulosa, for which we obtain five motion primitives that correspond to bell cavity pressurization, jet formation, jetting, cavity fluid refill, and coasting. The second hydromedusa is the oblate Aequorea victoria, for which we obtain five motion primitives that correspond to bell compression, vortex separation, cavity fluid refill, vortex formation, and coasting. Our experimental results indicate that the breakpoints between primitives are correlated with transitions in the bell geometry, vortex formation and shedding, and changes in derived dynamical quantities. These dynamics quantities include terms like pressure, power, drag, and thrust. Such findings suggest that dynamics information is inherently present in the observed motions.