Getting into the Swing of Things

Abstract

Over the course of the past 30 years, many scholars have explored the physical mechanisms that allow humans to go up on swing sets. Parametric and driven oscillation have been shown to produce the driving factors of angular momentum on solid-rod swing analysis. Each paper assigns a single method of swinging to analyze what they claim is representative enough of how humans swing. Furthermore, very few of these analyses acknowledge that humans use their arms to deliver torque on the rope, periodically bending the rope at points of the swing. In this report we explore how different swing methods on a solid-rope swing effect performance. We also explore how the analysis changes from the solid-rod swing when we instead assume a rope that can bend where it is held by the hand. We use data from humans on swing sets to find a more representative method of swinging. These comparisons are done using simulations representative of our theoretical model and using a humanoid robot, the Adorabot, that is used to execute the swing styles we are exploring. We find that swinging algorithms that move the system’s center of mass in the same way as an ideal parametric oscillator have better performance characteristics. We relate maximum angle achieved by a swinger using a human-swinging algorithm to the maximum lean-angle of a swinger’s torso. We show that assuming a rope to be a solid-rod causes worse swinging performance, and that assuming a solid-rod in simulation for the Adorabot results in an error of 38.4% for predicting the maximum angle achieved by the rope. While the solid-rod assumption can be made for describing the general behavior of the swing, it is a large oversimplification and would be inadequate for attempting to accurately model a human on a swing-set.