Problem C: Dog Cannot Catch a Treat - Using Differential Equations and Artificial Intelligence...
0Presented by:
Nathan Tam, Penn State Berks Division of Science, Reading PA USA
Mark Longenberger, Penn State Berks Division of Science, Reading PA USA
Lillie Mohn, Penn State Berks Division of Science, Reading PA USA
https://qubeshub.org/community/groups/simiode/expo
Abstract: Problem C: Dog Cannot Catch, from SCUDEM VIII 2023 was the problem our team investigated. Here, we attempt to answer the question of why Fritz, a dog from the popular YouTube video "Fritz Learns to Catch," cannot catch food that is thrown at him. This presentation will discuss the background work and techniques that ultimately led to our final presentation. Background discussion will include the various approaches our team considered such as models taking root in statistics and multivariable calculus. Then, we will discuss our chosen model, which was divided into two distinct components: the kinematics of the flight of the food object and the decision-making tendencies of Fritz. The model for the flight of the food object was bounded by differential equations considering air resistance. To model the dog's decision-making, our team utilized the NEAT (NeuroEvolution of Augmenting Topologies) method to train an AI model. Finally, the presentation will discuss our results and future improvements our team would take the provide a more accurate model of Fritz.