19.1 Wednesday, Jan. 4 How is dog gait affected by natural rough terrain? WILSHIN, Simon; HAYNES, G. Clark; REEVE, Michelle; REVZEN, Shai; SPENCE, Andrew J.*; Royal Veterinary College; University of Pennsylvania; Royal Veterinary College; University of Pennsylvania; Royal Veterinary College firstname.lastname@example.org
In nature legged animals depend on locomotion over uneven terrain for survival and reproduction. One way in which animals may optimize their locomotor behaviour for this task is by adjusting the relative timing of their leg recirculation, or gait. Therefore, we asked how the relative leg timing of quadrupeds changes during locomotion over natural, uneven terrain, and compared this to our idealised notions of the walk, trot and gallop. Five male dogs of shoulder height 522.0 ± 62.6 mm (mean ± s.d.) and body mass 20.0 ± 2.5 kg (mean ± s.d.) were trialled at nominal walk, trot, and gallop speeds over flat and uneven terrain. Mean perturbation size on uneven terrain was 54.8 ± 44.6 mm versus 4.2 ± 3.1 mm on flat. Dogs were fitted with a wirelessly synchronized suite of five sensors, comprised of Global Position System and inertial measurement units. One device was attached to the proximal-most segment of each leg, and a fifth on the midline of the back at the front legs. Raw sensor data were used to compute animal speed, position, and a continuous estimate of leg phases. The centroids of relative leg phase (averaged across time within each stride), describing the gait used by the dog on each terrain at each nominal gait speed, were significantly different on the rough terrain (linear mixed-model; n=5 dogs, p<0.05). At walking speeds on the rough terrain, dog gait moves towards the trot. Averages and distances between gaits in relative leg phase space do not account for the dynamical and geometric structure of these phase variables, however. Theoretical developments required to handle these data will be discussed. To explain why we observe these changes in dog gait, we propose experiments in a physical model, the robot XRL.