Carleton University - School of Computer Science Honours Project
Winter 2021
Adding Ladder Climbing to Phase-Functioned Neural Networks
Steven Thomas
SCS Honours Project Image
ABSTRACT
The goal of this project was to modify the research done in “Phase-Functioned Neural Networks for Character Control” so that the system would be able to climb ladders and ledges. The original research used a neural network to control the animations of a character in 3d space. This project seemed simple enough in its theory and through most of its implementation, but encountered issues when the time came to create the AI system. The manner required to train the model to move vertically correctly was not figured out during this project. Due to issues with implementation, much of the original research needed to be reused to compensate for the issues creating a new AI for the task. The result of the project showed that while it may be possible to adjust the network to climb ladders, it is not an easy task to perform.