Constraining frisbee tracking methods through Bayesian analysis of flying disc models

Abstract

Since the invention of the original Frisbee in the 1940s, disc sports such as Ultimate Frisbee and disc golf have exploded in popularity. Yet despite the expansion of disc sports, the body of literature surrounding Frisbee flight dynamics remains limited. Understanding flying disc physics – which is both easy to observe and nontrivial to model – may improve the performance of disc sport athletes. This Honors Thesis builds on the work of Frisbee Flight Simulation and Throw Biomechanics, published as a master’s thesis by Sarah Hummel in 2003. After exploring the Frisbee model described by Hummel, the work presented here uses a Bayesian approach to compare this model against simulated flight data. By generating a set of synthetic data and using an MCMC algorithm to recover the parameters used to produce the data set, we illustrate a possible approach for assessing accuracy of Frisbee models. We propose that this type of analysis can be used to constrain properties of cameras and develop enhanced tracking methods for flying discs. Since Frisbee data can be easily captured using basic smartphones, the work herein offers an accessible way to test modern parameter estimation and data-driven modeling methods.