A model-experiment comparison of system dynamics for human walking and running.

The human musculo-skeletal system comprises high complexity which makes it difficult to identify underlying basic principles of bipedal locomotion. To tackle this challenge, a common approach is to strip away complexity and formulate a reductive model. With utter simplicity a bipedal spring-mass model gives good predictions of the human gait dynamics, however, it has not been fully investigated whether center of mass motion over time of walking and running is comparable between the model and the human body over a wide range of speed. To test the model's ability in this respect, we compare sagittal center of mass trajectories of model and human data for speeds ranging from 0.5 m/s to 4 m/s. For simulations, system parameters and initial conditions are extracted from experimental observations of 28 subjects. The leg parameters stiffness and length are extracted from functional fitting to the subjects' leg force-length curves. With small variations of the touch-down angle of the leg and the vertical position of the center of mass at apex, we find successful spring-mass simulations for moderate walking and medium running speeds. Predictions of the sagittal center of mass trajectories and ground reaction forces are good, but their amplitudes are overestimated, while contact time is underestimated. At faster walking speeds and slower running speeds we do not find successful model locomotion with the extent of allowed parameter variation. We conclude that the existing limitations may be improved by adding complexity to the model.