Meeting Abstract

114.2  Monday, Jan. 7  Continuous metrics for classification of bipedal gaits and predictions of gait transition fine structure in turkeys WILSHIN, SD*; DALEY, MA; Royal Veterinary College; Royal Veterinary College swilshin@rvc.ac.uk

An accurate method for differentiating walking and running bipedal gaits is presented and applied to experimental data from turkeys. It is known that bipedal walking and running gaits can be distinguished based on the phase difference of the kinetic and potential energy of the center of mass. It has also been suggested that the energy stored in the legs may also differentiate these gaits. It is, however, conventional to use a discrete estimate of these phase differences (such as the relative timing of peaks in these energies). Such estimates are prone to error, especially during non-steady locomotion, and are of limited utility when examining gait transitions which typically occur over short time scales (one or two strides). It is the short time scale of transitions that makes them interesting, as it is likely that energy cost is of diminished importance.  Other factors such as stability and robustness may play a greater role in the form of transitions than in ordinary locomotion. To investigate these issues, we need a continuous-in-time classification of gait. We present and discuss a continuous-in-time classification of the gait of turkeys (n=5) on a treadmill. Gait classification was constructed by applying continuous-time phase extraction techniques to kinematic data. We show that the resulting gait classifier has a high performance, average 93% correct from a testing set with at least 16 strides per bird, and can correctly classify partial strides. We will discuss the potential application of this classifier to investigating locomotor dynamics and transition fine structure. Funding: HFSP:RGY0062/2010