Predictive Modeling of Healthy and Amputee Walking using a Simple Planar Model

It is generally accepted that the design of the foot can influence gait for both robots and amputees. However, the nonlinear nature of bipedal systems combined with the intermittent contacts required for walking makes rigorous investigation of the influence of foot design difficult. Further, it is not known how to predict even normal human walking across a range of speeds. This work extends a robot modeling and control technique to create highly accurate, yet computationally inexpensive, predictive models of human walking. The models were validated using new robotic hardware experiments and existing data from human subject studies. The results have implications both for the design and control of robots and for models of amputee gait. The models are grounded on a hybrid zero dynamics (HZD)-based control technique originally developed for planar bipedal robots with point feet and instantaneous transfer of support. To account for the function of the human foot, curved feet were incorporated into the HZD framework, which required rederiving the model of the transfer of support and the stability criteria. The effects of foot design on gait were systematically investigated in both simulation and hardware. Due to interaction effects between foot radius and center of curvature location, there are two distinct design strategies that yield energetically efficient gaits. In addition, models of human walking must also include the effects of toe-off. This can be achieved either with an impulsive force at the hip during the impact phase or with ankle joints. Using a six-link planar model with ankle joints and curved feet, normal human joint kinematics and energy expenditures can be predicted very accurately at walking speeds ranging from very slow to very fast using a torque-squared-based objective function. By removing one of the ankle joints from the symmetric human model, asymmetric amputee gait can be investigated. It is shown that small reductions in amputated/contralateral leg symmetry lead to large improvements in energetic efficiency.