A point-mass model of gibbon locomotion

Abstract

In brachiation, an animal uses alternating bimanual support to move beneath an overhead support. Past brachiation models have been based on the oscillations of a simple pendulum over half of a full cycle of oscillation. These models have been unsatisfying because the natural behavior of gibbons and siamangs appears to be far less restricted than so predicted. Cursorial mammals use an inverted pendulum-like energy exchange in walking, but switch to a spring-based energy exchange in running as velocity increases. Brachiating apes do not possess the anatomical springs characteristic of the limbs of terrestrial runners and do not appear to be using a spring-based gait. How do these animals move so easily within the branches of the forest canopy? Are there fundamental mechanical factors responsible for the transition from a continuous-contact gait where at least one hand is on a hand hold at a time, to a ricochetal gait where the animal vaults between hand holds? We present a simple model of ricochetal locomotion based on a combination of parabolic free flight and simple circular pendulum motion of a single point mass on a massless arm. In this simple brachiation model, energy losses due to inelastic collisions of the animal with the support are avoided, either because the collisions occur at zero velocity (continuous-contact brachiation) or by a smooth matching of the circular and parabolic trajectories at the point of contact (ricochetal brachiation). This model predicts that brachiation is possible over a large range of speeds, handhold spacings and gait frequencies with (theoretically) no mechanical energy cost. We then add the further assumption that a brachiator minimizes either its total energy or, equivalently, its peak arm tension, or a peak tension-related measure of muscle contraction metabolic cost. However, near the optimum the model is still rather unrestrictive. We present some comparisons with gibbon brachiation showing that the simple dynamic model presented has predictive value. However, natural gibbon motion is even smoother than the smoothest motions predicted by this primitive model.