Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
DY: Fachverband Dynamik und Statistische Physik
DY 20: Focus Session: Modelling of Non-linear Dynamics in Biological Movement (joint session BP/ DY)
DY 20.3: Vortrag
Mittwoch, 2. April 2014, 14:45–15:00, ZEU 250
Humans run like pogosticks - with ankles — •Horst-Moritz Maus1, Shai Revzen2, John Guckenheimer3, Christian Ludwig1, Johann Reger4, and Andre Seyfarth1 — 1TU Darmstadt, Deutschland — 2University of Michigan, Ann Arbor, USA — 3Cornell University, Ithaca, USA — 4TU Ilmenau, Deutschland
Running is an essential mode of human locomotion. Its large number of biomechanical and neural degrees of freedom are often modeled as a simple Spring Loaded Inverted Pendulum (SLIP). The SLIP body bounces as if on a pogo stick, pivoting on its spring leg and then jumping through a ballistic aerial phase. Updating SLIP model parameters to fit each step can result in trajectories that follow an observed path much more closely. These parameter updates represent a control input modulating the uncontrolled SLIP dynamics to obtain human-like movement and stability. Here we systematically construct a minimalistic *ankle-SLIP* model from measurements of humans running on a treadmill. Using Data Driven Floquet Analysis we identify candidate predictors for the parameter changes. Selecting a predictor related to ankle state allows us to predict running motion stride to stride and mimic rates of recovery from perturbation. We reveal inherent limitations in predictions made by other SLIP variants. Our methods produce a systematic means to search for prediction enhancing, yet low dimensional models of rhythmic processes in the physical sciences. More directly the ''ankle-SLIP'' models may impact gait assessment in sports and in clinical contexts and suggest control strategies for humanoid robots and prosthetic limbs.