FULL, R.J.*; KUBOW, T.; GARCIA, M.; SCHWIND, W.; KODISTCHEK, D.; Univ. of California, Berkeley; Univ. of California, Berkeley; Univ. of California, Berkeley; Univ. of Michigan, Ann Arbor; Univ. of Michigan, Ann Arbor: Can a simple neural oscillator generate rapid running in cockroaches?

Animals are complex, high dimensional, dynamical systems. Yet, their musculo-skeletal systems can be modeled by a low degree of freedom single mass atop a spring and their control system by simple neural oscillators. We tested the hypothesis that self-stabilized, high-speed running might be excited by a single neural clock. We performed principle component analysis (PCA) on a 42 degree of freedom kinematic data set captured from the cockroach Blaberus discoidalis. A high degree of rhythmicity and stereotypy does not guarantee simple control signals. Multiple control signals could be required to move one joint followed by the next and timing may differ among legs. PCA on joint angle data from straight ahead running revealed that three to six principle components could account for nearly all of the systematic variation of the limbs with a single component representing over 80%. PCA reveals linear correlations between joint angles within and among all legs at all points in time, bounding the effective number of joint degrees of freedom actually used in performing the motion. PCs generated from a reduced population of data were able to reconstruct data of different strides and other individuals. Furthermore, PC vectors generated from different subpopulations of the data spanned very similar subspaces within the ambient space of possible joint motions. Combined, these results suggest that cockroaches running rapidly at steady-state operate within the same low dimensional subspace of the much higher possible available degrees of joint freedom. A single neural clock-like signal can stably excite a tuned mechanical system as it does in a physical model such as the hexapedal robot RHex.