The strengths and weaknesses of inverted pendulum models of human walking.

An investigation into the kinematic and kinetic predictions of two "inverted pendulum" (IP) models of gait was undertaken. The first model consisted of a single leg, with anthropometrically correct mass and moment of inertia, and a point mass at the hip representing the rest of the body. A second model incorporating the physiological extension of a head-arms-trunk (HAT) segment, held upright by an actuated hip moment, was developed for comparison. Simulations were performed, using both models, and quantitatively compared with empirical gait data. There was little difference between the two models' predictions of kinematics and ground reaction force (GRF). The models agreed well with empirical data through mid-stance (20-40% of the gait cycle) suggesting that IP models adequately simulate this phase (mean error less than one standard deviation). IP models are not cyclic, however, and cannot adequately simulate double support and step-to-step transition. This is because the forces under both legs augment each other during double support to increase the vertical GRF. The incorporation of an actuated hip joint was the most novel change and added a new dimension to the classic IP model. The hip moment curve produced was similar to those measured during experimental walking trials. As a result, it was interpreted that the primary role of the hip musculature in stance is to keep the HAT upright. Careful consideration of the differences between the models throws light on what the different terms within the GRF equation truly represent.